Task()

Task(env)

Task(env, numcon, numvar)

Task(task)


Constructor of a new optimization task.

Parameters
• env (Env) – Parent environment. (input)

• numcon (int) – An optional hint about the maximal number of constraints in the task. (input)

• numvar (int) – An optional hint about the maximal number of variables in the task. (input)

• task (Task) – A task that will be cloned. (input)

__del__()


Free the underlying native allocation.

analyzenames(whichstream,nametype)


The function analyzes the names and issues an error if a name is invalid.

Parameters
Groups

Names

analyzeproblem(whichstream)


The function analyzes the data of a task and writes out a report.

Parameters

whichstream (mosek.streamtype) – Index of the stream. (input)

Groups

analyzesolution(whichstream,whichsol)


Print information related to the quality of the solution and other solution statistics.

By default this function prints information about the largest infeasibilites in the solution, the primal (and possibly dual) objective value and the solution status.

Following parameters can be used to configure the printed statistics:

Parameters
Groups
appendacc(domidx,afeidxlist,b)


Appends an affine conic constraint to the task. The affine constraint has the form a sequence of affine expressions belongs to a domain.

The domain index is specified with domidx and should refer to a domain previously appended with one of the append...domain functions.

The length of the affine expression list afeidxlist must be equal to the dimension $$n$$ of the domain. The elements of afeidxlist are indexes to the store of affine expressions, i.e. the affine expressions appearing in the affine conic constraint are:

$F_{\mathrm{afeidxlist}[k],:}x + g_{\mathrm{afeidxlist}[k]} \quad \mathrm{for}\ k=0,\ldots,n-1.$

If an optional vector b of the same length as afeidxlist is specified then the expressions appearing in the affine constraint will instead be taken as:

$F_{\mathrm{afeidxlist}[k],:}x + g_{\mathrm{afeidxlist}[k]} - b_k \quad \mathrm{for}\ k=0,\ldots,n-1.$
Parameters
• domidx (int) – Domain index. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

appendaccs(domidxs,afeidxlist,b)


Appends numaccs affine conic constraint to the task. Each single affine conic constraint should be specified as in Task.appendacc and the input of this function should contain the concatenation of all these descriptions.

In particular, the length of afeidxlist must equal the sum of dimensions of domains indexed in domainsidxs.

Parameters
• domidxs (int[]) – Domain indices. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

appendaccseq(domidx,afeidxfirst,b)


Appends an affine conic constraint to the task, as in Task.appendacc. The function assumes the affine expressions forming the constraint are sequential. The affine constraint has the form a sequence of affine expressions belongs to a domain.

The domain index is specified with domidx and should refer to a domain previously appended with one of the append...domain functions.

The number of affine expressions should be equal to the dimension $$n$$ of the domain. The affine expressions forming the affine constraint are arranged sequentially in a contiguous block of the affine expression store starting from position afeidxfirst. That is, the affine expressions appearing in the affine conic constraint are:

$F_{\mathrm{afeidxfirst}+k,:}x + g_{\mathrm{afeidxfirst}+k} \quad \mathrm{for}\ k=0,\ldots,n-1.$

If an optional vector b of length numafeidx is specified then the expressions appearing in the affine constraint will instead be taken as

$F_{\mathrm{afeidxfirst}+k,:}x + g_{\mathrm{afeidxfirst}+k} - b_k \quad \mathrm{for}\ k=0,\ldots,n-1.$
Parameters
• domidx (int) – Domain index. (input)

• afeidxfirst (int) – Index of the first affine expression. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

appendaccsseq(domidxs,numafeidx,afeidxfirst,b)


Appends numaccs affine conic constraint to the task. It is the block variant of Task.appendaccs, that is it assumes that the affine expressions appearing in the affine conic constraints are sequential in the affine expression store, starting from position afeidxfirst.

Parameters
• domidxs (int[]) – Domain indices. (input)

• numafeidx (int) – Number of affine expressions in the affine expression list (must equal the sum of dimensions of the domains). (input)

• afeidxfirst (int) – Index of the first affine expression. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

appendafes(num)


Appends a number of empty affine expressions to the task.

Parameters

num (int) – Number of empty affine expressions which should be appended. (input)

Groups

Problem data - affine expressions

appendbarvars(dim)


Appends positive semidefinite matrix variables of dimensions given by dim to the problem.

Parameters

dim (int[]) – Dimensions of symmetric matrix variables to be added. (input)

Groups

Problem data - semidefinite

appendcone(ct,conepar,submem)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Appends a new conic constraint to the problem. Hence, add a constraint

$\hat{x} \in \K$

to the problem, where $$\K$$ is a convex cone. $$\hat{x}$$ is a subset of the variables which will be specified by the argument submem. Cone type is specified by ct.

Define

$\hat{x} = x_{\mathtt{submem}[0]},\ldots,x_{\mathtt{submem}[\mathtt{nummem}-1]}.$

Depending on the value of ct this function appends one of the constraints:

• Quadratic cone (conetype.quad, requires $$\mathtt{nummem}\geq 1$$):

$\hat{x}_0 \geq \sqrt{\sum_{i=1}^{i<\mathtt{nummem}} \hat{x}_i^2}$
• Rotated quadratic cone (conetype.rquad, requires $$\mathtt{nummem}\geq 2$$):

$2 \hat{x}_0 \hat{x}_1 \geq \sum_{i=2}^{i<\mathtt{nummem}} \hat{x}^2_i, \quad \hat{x}_{0}, \hat{x}_1 \geq 0$
• Primal exponential cone (conetype.pexp, requires $$\mathtt{nummem}=3$$):

$\hat{x}_0 \geq \hat{x}_1\exp(\hat{x}_2/\hat{x}_1), \quad \hat{x}_0,\hat{x}_1 \geq 0$
• Primal power cone (conetype.ppow, requires $$\mathtt{nummem}\geq 2$$):

$\hat{x}_0^\alpha \hat{x}_1^{1-\alpha} \geq \sqrt{\sum_{i=2}^{i<\mathtt{nummem}} \hat{x}^2_i}, \quad \hat{x}_{0}, \hat{x}_1 \geq 0$

where $$\alpha$$ is the cone parameter specified by conepar.

• Dual exponential cone (conetype.dexp, requires $$\mathtt{nummem}=3$$):

$\hat{x}_0 \geq -\hat{x}_2 e^{-1}\exp(\hat{x}_1/\hat{x}_2), \quad \hat{x}_2\leq 0,\hat{x}_0 \geq 0$
• Dual power cone (conetype.dpow, requires $$\mathtt{nummem}\geq 2$$):

$\left(\frac{\hat{x}_0}{\alpha}\right)^\alpha \left(\frac{\hat{x}_1}{1-\alpha}\right)^{1-\alpha} \geq \sqrt{\sum_{i=2}^{i<\mathtt{nummem}} \hat{x}^2_i}, \quad \hat{x}_{0}, \hat{x}_1 \geq 0$

where $$\alpha$$ is the cone parameter specified by conepar.

• Zero cone (conetype.zero):

$\hat{x}_i = 0 \ \textrm{for all}\ i$

Please note that the sets of variables appearing in different conic constraints must be disjoint.

For an explained code example see Sec. 6.3 (Conic Quadratic Optimization), Sec. 6.5 (Conic Exponential Optimization) or Sec. 6.4 (Power Cone Optimization).

Parameters
• ct (mosek.conetype) – Specifies the type of the cone. (input)

• conepar (float) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0. (input)

• submem (int[]) – Variable subscripts of the members in the cone. (input)

Groups

Problem data - cones (deprecated)

appendconeseq(ct,conepar,nummem,j)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Appends a new conic constraint to the problem, as in Task.appendcone. The function assumes the members of cone are sequential where the first member has index j and the last j+nummem-1.

Parameters
• ct (mosek.conetype) – Specifies the type of the cone. (input)

• conepar (float) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0. (input)

• nummem (int) – Number of member variables in the cone. (input)

• j (int) – Index of the first variable in the conic constraint. (input)

Groups

Problem data - cones (deprecated)

appendconesseq(ct,conepar,nummem,j)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Appends a number of conic constraints to the problem, as in Task.appendcone. The $$k$$th cone is assumed to be of dimension nummem[k]. Moreover, it is assumed that the first variable of the first cone has index $$j$$ and starting from there the sequentially following variables belong to the first cone, then to the second cone and so on.

Parameters
• ct (mosek.conetype[]) – Specifies the type of the cone. (input)

• conepar (float[]) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0. (input)

• nummem (int[]) – Numbers of member variables in the cones. (input)

• j (int) – Index of the first variable in the first cone to be appended. (input)

Groups

Problem data - cones (deprecated)

appendcons(num)


Appends a number of constraints to the model. Appended constraints will be declared free. Please note that MOSEK will automatically expand the problem dimension to accommodate the additional constraints.

Parameters

num (int) – Number of constraints which should be appended. (input)

Groups
appenddjcs(num)


Appends a number of empty disjunctive constraints to the task.

Parameters

num (int) – Number of empty disjunctive constraints which should be appended. (input)

Groups

Problem data - disjunctive constraints

appenddualexpconedomain() -> (domidx)


Appends the dual exponential cone $$\left\{ x\in \real^3 ~:~ x_0 \geq -x_2 e^{-1} e^{x_1/x_2},\ x_0> 0,\ x_2< 0 \right\}$$ to the list of domains.

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appenddualgeomeanconedomain(n) -> (domidx)


Appends the dual geometric mean cone $$\left\{ x\in \real^n ~:~ (n-1) \left(\prod_{i=0}^{n-2} x_i\right)^{1/(n-1)} \geq |x_{n-1}|,\ x_0,\ldots,x_{n-2}\geq 0 \right\}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appenddualpowerconedomain(n,alpha) -> (domidx)


Appends the dual power cone domain of dimension $$n$$, with $$n_\ell$$ variables appearing on the left-hand side, where $$n_\ell$$ is the length of $$\alpha$$, and with a homogenous sequence of exponents $$\alpha_0,\ldots,\alpha_{n_\ell-1}$$.

Formally, let $$s = \sum_i \alpha_i$$ and $$\beta_i = \alpha_i / s$$, so that $$\sum_i \beta_i=1$$. Then the dual power cone is defined as follows:

$\left\{ x\in \real^n ~:~ \prod_{i=0}^{n_\ell-1} \left(\frac{x_i}{\beta_i}\right)^{\beta_i} \geq \sqrt{\sum_{j=n_\ell}^{n-1}x_j^2},\ x_0\ldots,x_{n_\ell-1}\geq 0 \right\}$
Parameters
• n (int) – Dimension of the domain. (input)

• alpha (float[]) – The sequence proportional to exponents. Must be positive. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendprimalexpconedomain() -> (domidx)


Appends the primal exponential cone $$\left\{ x\in \real^3 ~:~ x_0 \geq x_1 e^{x_2/x_1},\ x_0,x_1> 0 \right\}$$ to the list of domains.

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendprimalgeomeanconedomain(n) -> (domidx)


Appends the primal geometric mean cone $$\left\{ x\in \real^n ~:~ \left(\prod_{i=0}^{n-2} x_i\right)^{1/(n-1)} \geq |x_{n-1}|,\ x_0\ldots,x_{n-2}\geq 0 \right\}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendprimalpowerconedomain(n,alpha) -> (domidx)


Appends the primal power cone domain of dimension $$n$$, with $$n_\ell$$ variables appearing on the left-hand side, where $$n_\ell$$ is the length of $$\alpha$$, and with a homogenous sequence of exponents $$\alpha_0,\ldots,\alpha_{n_\ell-1}$$.

Formally, let $$s = \sum_i \alpha_i$$ and $$\beta_i = \alpha_i / s$$, so that $$\sum_i \beta_i=1$$. Then the primal power cone is defined as follows:

$\left\{ x\in \real^n ~:~ \prod_{i=0}^{n_\ell-1} x_i^{\beta_i} \geq \sqrt{\sum_{j=n_\ell}^{n-1}x_j^2},\ x_0\ldots,x_{n_\ell-1}\geq 0 \right\}$
Parameters
• n (int) – Dimension of the domain. (input)

• alpha (float[]) – The sequence proportional to exponents. Must be positive. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendquadraticconedomain(n) -> (domidx)


Appends the $$n$$-dimensional quadratic cone $$\left\{x\in\real^n~:~x_0 \geq \sqrt{\sum_{i=1}^{n-1} x_i^2}\right\}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendrdomain(n) -> (domidx)


Appends the $$n$$-dimensional real space $$\{ x \in \real^n \}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendrminusdomain(n) -> (domidx)


Appends the $$n$$-dimensional negative orthant $$\{ x \in \real^n: \, x \leq 0 \}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendrplusdomain(n) -> (domidx)


Appends the $$n$$-dimensional positive orthant $$\{ x \in \real^n: \, x \geq 0 \}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendrquadraticconedomain(n) -> (domidx)


Appends the $$n$$-dimensional rotated quadratic cone $$\left\{x\in\real^n~:~2 x_0 x_1 \geq \sum_{i=2}^{n-1} x_i^2,\ x_0,x_1\geq 0\right\}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendrzerodomain(n) -> (domidx)


Appends the zero in $$n$$-dimensional real space $$\{ x \in \real^n: \, x = 0 \}$$ to the list of domains.

Parameters

n (int) – Dimmension of the domain. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendsparsesymmat(dim,subi,subj,valij) -> (idx)


MOSEK maintains a storage of symmetric data matrices that is used to build $$\barC$$ and $$\barA$$. The storage can be thought of as a vector of symmetric matrices denoted $$E$$. Hence, $$E_i$$ is a symmetric matrix of certain dimension.

This function appends a general sparse symmetric matrix on triplet form to the vector $$E$$ of symmetric matrices. The vectors subi, subj, and valij contains the row subscripts, column subscripts and values of each element in the symmetric matrix to be appended. Since the matrix that is appended is symmetric, only the lower triangular part should be specified. Moreover, duplicates are not allowed.

Observe the function reports the index (position) of the appended matrix in $$E$$. This index should be used for later references to the appended matrix.

Parameters
• dim (int) – Dimension of the symmetric matrix that is appended. (input)

• subi (int[]) – Row subscript in the triplets. (input)

• subj (int[]) – Column subscripts in the triplets. (input)

• valij (float[]) – Values of each triplet. (input)

Return

idx (int) – Unique index assigned to the inputted matrix that can be used for later reference.

Groups

Problem data - semidefinite

appendsparsesymmatlist(dims,nz,subi,subj,valij,idx)

appendsparsesymmatlist(dims,nz,subi,subj,valij) -> (idx)


MOSEK maintains a storage of symmetric data matrices that is used to build $$\barC$$ and $$\barA$$. The storage can be thought of as a vector of symmetric matrices denoted $$E$$. Hence, $$E_i$$ is a symmetric matrix of certain dimension.

This function appends general sparse symmetric matrixes on triplet form to the vector $$E$$ of symmetric matrices. The vectors subi, subj, and valij contains the row subscripts, column subscripts and values of each element in the symmetric matrix to be appended. Since the matrix that is appended is symmetric, only the lower triangular part should be specified. Moreover, duplicates are not allowed.

Observe the function reports the index (position) of the appended matrix in $$E$$. This index should be used for later references to the appended matrix.

Parameters
• dims (int[]) – Dimensions of the symmetric matrixes. (input)

• nz (int[]) – Number of nonzeros for each matrix. (input)

• subi (int[]) – Row subscript in the triplets. (input)

• subj (int[]) – Column subscripts in the triplets. (input)

• valij (float[]) – Values of each triplet. (input)

• idx (int[]) – Unique index assigned to the inputted matrix that can be used for later reference. (output)

Return

idx (int[]) – Unique index assigned to the inputted matrix that can be used for later reference.

Groups

Problem data - semidefinite

appendsvecpsdconedomain(n) -> (domidx)


Appends the domain consisting of vectors of length $$n=d(d+1)/2$$ defined as follows

$\{(x_1,\ldots,x_{d(d+1)/2})\in \real^n~:~ \mathrm{sMat}(x)\in\PSD^d\} = \{\mathrm{sVec}(X)~:~X\in\PSD^d\},$

where

$\mathrm{sVec}(X) = (X_{11},\sqrt{2}X_{21},\ldots,\sqrt{2}X_{d1},X_{22},\sqrt{2}X_{32},\ldots,X_{dd}),$

and

$\begin{split}\mathrm{sMat}(x) = \left[\begin{array}{cccc}x_1 & x_2/\sqrt{2} & \cdots & x_{d}/\sqrt{2} \\ x_2/\sqrt{2} & x_{d+1} & \cdots & x_{2d-1}/\sqrt{2} \\ \cdots & \cdots & \cdots & \cdots \\ x_{d}/\sqrt{2} & x_{2d-1}/\sqrt{2} & \cdots & x_{d(d+1)/2}\end{array}\right].\end{split}$

In other words, the domain consists of vectorizations of the lower-triangular part of a positive semidefinite matrix, with the non-diagonal elements additionally rescaled.

This domain is a self-dual cone.

Parameters

n (int) – Dimension of the domain, must be of the form $$d(d+1)/2$$. (input)

Return

domidx (int) – Index of the domain.

Groups

Problem data - domain

appendvars(num)


Appends a number of variables to the model. Appended variables will be fixed at zero. Please note that MOSEK will automatically expand the problem dimension to accommodate the additional variables.

Parameters

num (int) – Number of variables which should be appended. (input)

Groups
asyncgetresult(address,accesstoken,token) -> (respavailable,resp,trm)


Request a solution from a remote job identified by the argument token. For other arguments see Task.asyncoptimize. If the solution is available it will be retrieved and loaded into the local task.

Parameters
• address (str) – Address of the OptServer. (input)

• accesstoken (str) – Access token. (input)

• token (str) – The task token. (input)

Return
• respavailable (bool) – Indicates if a remote response is available. If this is not true, resp and trm should be ignored.

• resp (mosek.rescode) – Is the response code from the remote solver.

• trm (mosek.rescode) – Is either rescode.ok or a termination response code.

Groups

Remote optimization

asyncoptimize(address,accesstoken) -> (token)


Offload the optimization task to an instance of OptServer specified by addr, which should be a valid URL, for example http://server:port or https://server:port. The call will exit immediately.

If the server requires authentication, the authentication token can be passed in the accesstoken argument.

If the server requires encryption, the keys can be passed using one of the solver parameters sparam.remote_tls_cert or sparam.remote_tls_cert_path.

The function returns a token which should be used in future calls to identify the task.

Parameters
• address (str) – Address of the OptServer. (input)

• accesstoken (str) – Access token. (input)

Return

token (str) – Returns the task token.

Groups

Remote optimization

asyncpoll(address,accesstoken,token) -> (respavailable,resp,trm)


Requests information about the status of the remote job identified by the argument token. For other arguments see Task.asyncoptimize.

Parameters
• address (str) – Address of the OptServer. (input)

• accesstoken (str) – Access token. (input)

• token (str) – The task token. (input)

Return
• respavailable (bool) – Indicates if a remote response is available. If this is not true, resp and trm should be ignored.

• resp (mosek.rescode) – Is the response code from the remote solver.

• trm (mosek.rescode) – Is either rescode.ok or a termination response code.

Groups

Remote optimization

asyncstop(address,accesstoken,token)


Request that the remote job identified by token is terminated. For other arguments see Task.asyncoptimize.

Parameters
• address (str) – Address of the OptServer. (input)

• accesstoken (str) – Access token. (input)

• token (str) – The task token. (input)

Groups

Remote optimization

basiscond() -> (nrmbasis,nrminvbasis)


If a basic solution is available and it defines a nonsingular basis, then this function computes the 1-norm estimate of the basis matrix and a 1-norm estimate for the inverse of the basis matrix. The 1-norm estimates are computed using the method outlined in [Ste98], pp. 388-391.

By definition the 1-norm condition number of a matrix $$B$$ is defined as

$\kappa_1(B) := \|B\|_1 \|B^{-1}\|_1.$

Moreover, the larger the condition number is the harder it is to solve linear equation systems involving $$B$$. Given estimates for $$\|B\|_1$$ and $$\|B^{-1}\|_1$$ it is also possible to estimate $$\kappa_1(B)$$.

Return
• nrmbasis (float) – An estimate for the 1-norm of the basis.

• nrminvbasis (float) – An estimate for the 1-norm of the inverse of the basis.

Groups

Solving systems with basis matrix

checkmem(file,line)


Checks the memory allocated by the task.

Parameters
• file (str) – File from which the function is called. (input)

• line (int) – Line in the file from which the function is called. (input)

Groups

System, memory and debugging

chgconbound(i,lower,finite,value)


Changes a bound for one constraint.

If lower is non-zero, then the lower bound is changed as follows:

$\begin{split}\mbox{new lower bound} = \left\{ \begin{array}{ll} - \infty, & \mathtt{finite}=0, \\ \mathtt{value} & \mbox{otherwise}. \end{array} \right.\end{split}$

Otherwise if lower is zero, then

$\begin{split}\mbox{new upper bound} = \left\{ \begin{array}{ll} \infty, & \mathtt{finite}=0, \\ \mathtt{value} & \mbox{otherwise}. \end{array} \right.\end{split}$

Please note that this function automatically updates the bound key for the bound, in particular, if the lower and upper bounds are identical, the bound key is changed to fixed.

Parameters
• i (int) – Index of the constraint for which the bounds should be changed. (input)

• lower (int) – If non-zero, then the lower bound is changed, otherwise the upper bound is changed. (input)

• finite (int) – If non-zero, then value is assumed to be finite. (input)

• value (float) – New value for the bound. (input)

Groups
chgvarbound(j,lower,finite,value)


Changes a bound for one variable.

If lower is non-zero, then the lower bound is changed as follows:

$\begin{split}\mbox{new lower bound} = \left\{ \begin{array}{ll} - \infty, & \mathtt{finite}=0, \\ \mathtt{value} & \mbox{otherwise}. \end{array} \right.\end{split}$

Otherwise if lower is zero, then

$\begin{split}\mbox{new upper bound} = \left\{ \begin{array}{ll} \infty, & \mathtt{finite}=0, \\ \mathtt{value} & \mbox{otherwise}. \end{array} \right.\end{split}$

Please note that this function automatically updates the bound key for the bound, in particular, if the lower and upper bounds are identical, the bound key is changed to fixed.

Parameters
• j (int) – Index of the variable for which the bounds should be changed. (input)

• lower (int) – If non-zero, then the lower bound is changed, otherwise the upper bound is changed. (input)

• finite (int) – If non-zero, then value is assumed to be finite. (input)

• value (float) – New value for the bound. (input)

Groups
commitchanges()


Commits all cached problem changes to the task. It is usually not necessary to call this function explicitly since changes will be committed automatically when required.

Groups

deletesolution(whichsol)


Undefine a solution and free the memory it uses.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Groups
dualsensitivity(subj,
leftpricej,
rightpricej,
leftrangej,
rightrangej)

dualsensitivity(subj) ->
(leftpricej,
rightpricej,
leftrangej,
rightrangej)


Calculates sensitivity information for objective coefficients. The indexes of the coefficients to analyze are

$\{\mathtt{subj}[i] ~|~ i = 0,\ldots,\mathtt{numj}-1\}$

The type of sensitivity analysis to perform (basis or optimal partition) is controlled by the parameter iparam.sensitivity_type.

For an example, please see Section Example: Sensitivity Analysis.

Parameters
• subj (int[]) – Indexes of objective coefficients to analyze. (input)

• leftpricej (float[]) – $$\mathtt{leftpricej}[j]$$ is the left shadow price for the coefficient with index $$\mathtt{subj[j]}$$. (output)

• rightpricej (float[]) – $$\mathtt{rightpricej}[j]$$ is the right shadow price for the coefficient with index $$\mathtt{subj[j]}$$. (output)

• leftrangej (float[]) – $$\mathtt{leftrangej}[j]$$ is the left range $$\beta_1$$ for the coefficient with index $$\mathtt{subj[j]}$$. (output)

• rightrangej (float[]) – $$\mathtt{rightrangej}[j]$$ is the right range $$\beta_2$$ for the coefficient with index $$\mathtt{subj[j]}$$. (output)

Return
• leftpricej (float[]) – $$\mathtt{leftpricej}[j]$$ is the left shadow price for the coefficient with index $$\mathtt{subj[j]}$$.

• rightpricej (float[]) – $$\mathtt{rightpricej}[j]$$ is the right shadow price for the coefficient with index $$\mathtt{subj[j]}$$.

• leftrangej (float[]) – $$\mathtt{leftrangej}[j]$$ is the left range $$\beta_1$$ for the coefficient with index $$\mathtt{subj[j]}$$.

• rightrangej (float[]) – $$\mathtt{rightrangej}[j]$$ is the right range $$\beta_2$$ for the coefficient with index $$\mathtt{subj[j]}$$.

Groups

Sensitivity analysis

emptyafebarfrow(afeidx)


Clears a row in $$\barF$$ i.e. sets $$\barF_{\mathrm{afeidx},*} = 0$$.

Parameters

afeidx (int) – Row index of $$\barF$$. (input)

Groups
emptyafebarfrowlist(afeidxlist)


Clears a number of rows in $$\barF$$ i.e. sets $$\barF_{i,*} = 0$$ for all indices $$i$$ in afeidxlist.

Parameters

afeidxlist (int[]) – Indices of rows in $$\barF$$ to clear. (input)

Groups
emptyafefcol(varidx)


Clears one column in the affine constraint matrix $$F$$, that is sets $$F_{*,\mathrm{varidx}}=0$$.

Parameters

varidx (int) – Index of a variable (column in $$F$$). (input)

Groups

Problem data - affine expressions

emptyafefcollist(varidx)


Clears a number of columns in $$F$$ i.e. sets $$F_{*,j} = 0$$ for all indices $$j$$ in varidx.

Parameters

varidx (int[]) – Indices of variables (columns) in $$F$$ to clear. (input)

Groups

Problem data - affine expressions

emptyafefrow(afeidx)


Clears one row in the affine constraint matrix $$F$$, that is sets $$F_{\mathrm{afeidx},*}=0$$.

Parameters

afeidx (int) – Index of a row in $$F$$. (input)

Groups

Problem data - affine expressions

emptyafefrowlist(afeidx)


Clears a number of rows in $$F$$ i.e. sets $$F_{i,*} = 0$$ for all indices $$i$$ in afeidx.

Parameters

afeidx (int[]) – Indices of rows in $$F$$ to clear. (input)

Groups

Problem data - affine expressions

evaluateacc(whichsol,accidx,activity)

evaluateacc(whichsol,accidx) -> (activity)


Evaluates the activity of an affine conic constraint.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• accidx (int) – The index of the affine conic constraint. (input)

• activity (float[]) – The activity of the affine conic constraint. The array should have length equal to the dimension of the constraint. (output)

Return

activity (float[]) – The activity of the affine conic constraint. The array should have length equal to the dimension of the constraint.

Groups
evaluateaccs(whichsol,activity)

evaluateaccs(whichsol) -> (activity)


Evaluates the activities of all affine conic constraints.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• activity (float[]) – The activity of affine conic constraints. The array should have length equal to the sum of dimensions of all affine conic constraints. (output)

Return

activity (float[]) – The activity of affine conic constraints. The array should have length equal to the sum of dimensions of all affine conic constraints.

Groups
generateaccnames(sub,fmt,dims,sp,namedaxisidxs,names)


Internal.

Parameters
• sub (int[]) – Indexes of the affine conic constraints. (input)

• fmt (str) – The variable name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups

Names

generatebarvarnames(subj,fmt,dims,sp,namedaxisidxs,names)


Generates systematic names for variables.

Parameters
• subj (int[]) – Indexes of the variables. (input)

• fmt (str) – The variable name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups
generateconenames(subk,fmt,dims,sp,namedaxisidxs,names)


Internal, deprecated.

Parameters
• subk (int[]) – Indexes of the cone. (input)

• fmt (str) – The cone name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups
generateconnames(subi,fmt,dims,sp,namedaxisidxs,names)


Generates systematic names for constraints.

Parameters
• subi (int[]) – Indexes of the constraints. (input)

• fmt (str) – The constraint name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups
generatedjcnames(sub,fmt,dims,sp,namedaxisidxs,names)


Internal.

Parameters
• sub (int[]) – Indexes of the disjunctive constraints. (input)

• fmt (str) – The variable name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups

Names

generatevarnames(subj,fmt,dims,sp,namedaxisidxs,names)


Generates systematic names for variables.

Parameters
• subj (int[]) – Indexes of the variables. (input)

• fmt (str) – The variable name formatting string. (input)

• dims (int[]) – Dimensions in the shape. (input)

• sp (int[]) – Items that should be named. (input)

• namedaxisidxs (int[]) – List if named index axes (input)

• names (str[]) – All axis names. (input)

Groups
getaccafeidxlist(accidx,afeidxlist)

getaccafeidxlist(accidx) -> (afeidxlist)


Obtains the list of affine expressions appearing in the affine conic constraint.

Parameters
• accidx (int) – Index of the affine conic constraint. (input)

• afeidxlist (int[]) – List of indexes of affine expressions appearing in the constraint. (output)

Return

afeidxlist (int[]) – List of indexes of affine expressions appearing in the constraint.

Groups
getaccb(accidx,b)

getaccb(accidx) -> (b)


Obtains the additional constant term vector appearing in the affine conic constraint.

Parameters
• accidx (int) – Index of the affine conic constraint. (input)

• b (float[]) – The vector b appearing in the constraint. (output)

Return

b (float[]) – The vector b appearing in the constraint.

Groups
getaccbarfblocktriplet(acc_afe,bar_var,blk_row,blk_col,blk_val) -> (numtrip)

getaccbarfblocktriplet() ->
(numtrip,
acc_afe,
bar_var,
blk_row,
blk_col,
blk_val)


Obtains $$\barF$$, implied by the ACCs, in block triplet form. If the AFEs passed to the ACCs were out of order, then this function can be used to obtain the barF as seen by the ACCs.

Parameters
• acc_afe (int[]) – Index of the AFE within the concatenated list of AFEs in ACCs. (output)

• bar_var (int[]) – Symmetric matrix variable index. (output)

• blk_row (int[]) – Block row index. (output)

• blk_col (int[]) – Block column index. (output)

• blk_val (float[]) – The numerical value associated with each block triplet. (output)

Return
• numtrip (int) – Number of elements in the block triplet form.

• acc_afe (int[]) – Index of the AFE within the concatenated list of AFEs in ACCs.

• bar_var (int[]) – Symmetric matrix variable index.

• blk_row (int[]) – Block row index.

• blk_col (int[]) – Block column index.

• blk_val (float[]) – The numerical value associated with each block triplet.

Groups
getaccbarfnumblocktriplets() -> (numtrip)


Obtains an upper bound on the number of elements in the block triplet form of $$\barF$$, as used within the ACCs.

Return

numtrip (int) – An upper bound on the number of elements in the block triplet form of $$\barF.$$, as used within the ACCs.

Groups
getaccdomain(accidx) -> (domidx)


Obtains the domain appearing in the affine conic constraint.

Parameters

accidx (int) – The index of the affine conic constraint. (input)

Return

domidx (int) – The index of domain in the affine conic constraint.

Groups
getaccdoty(whichsol,accidx,doty)

getaccdoty(whichsol,accidx) -> (doty)


Obtains the $$\dot{y}$$ vector for a solution (the dual values of an affine conic constraint).

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• accidx (int) – The index of the affine conic constraint. (input)

• doty (float[]) – The dual values for this affine conic constraint. The array should have length equal to the dimension of the constraint. (output)

Return

doty (float[]) – The dual values for this affine conic constraint. The array should have length equal to the dimension of the constraint.

Groups
getaccdotys(whichsol,doty)

getaccdotys(whichsol) -> (doty)


Obtains the $$\dot{y}$$ vector for a solution (the dual values of all affine conic constraint).

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• doty (float[]) – The dual values of affine conic constraints. The array should have length equal to the sum of dimensions of all affine conic constraints. (output)

Return

doty (float[]) – The dual values of affine conic constraints. The array should have length equal to the sum of dimensions of all affine conic constraints.

Groups
getaccfnumnz() -> (accfnnz)


If the AFEs are not added sequentially to the ACCs, then the present function gives the number of nonzero elements in the F matrix that would be implied by the ordering of AFEs within ACCs.

Return

accfnnz (int) – Number of non-zeros in $$F$$ implied by ACCs.

Groups
getaccftrip(frow,fcol,fval)

getaccftrip() -> (frow,fcol,fval)


Obtains the $$F$$ (that would be implied by the ordering of the AFEs within the ACCs) in triplet format.

Parameters
• frow (int[]) – Row indices of nonzeros in the implied F matrix. (output)

• fcol (int[]) – Column indices of nonzeros in the implied F matrix. (output)

• fval (float[]) – Values of nonzero entries in the implied F matrix. (output)

Return
• frow (int[]) – Row indices of nonzeros in the implied F matrix.

• fcol (int[]) – Column indices of nonzeros in the implied F matrix.

• fval (float[]) – Values of nonzero entries in the implied F matrix.

Groups
getaccgvector(g)

getaccgvector() -> (g)


If the AFEs are passed out of sequence to the ACCs, then this function can be used to obtain the vector $$g$$ of constant terms used within the ACCs.

Parameters

g (float[]) – The $$g$$ used within the ACCs as a dense vector. The length is sum of the dimensions of the ACCs. (output)

Return

g (float[]) – The $$g$$ used within the ACCs as a dense vector. The length is sum of the dimensions of the ACCs.

Groups
getaccn(accidx) -> (n)


Obtains the dimension of the affine conic constraint.

Parameters

accidx (int) – The index of the affine conic constraint. (input)

Return

n (int) – The dimension of the affine conic constraint (equal to the dimension of its domain).

Groups
getaccname(accidx) -> (name)


Obtains the name of an affine conic constraint.

Parameters

accidx (int) – Index of an affine conic constraint. (input)

Return

name (str) – Returns the required name.

Groups
getaccnamelen(accidx) -> (len)


Obtains the length of the name of an affine conic constraint.

Parameters

accidx (int) – Index of an affine conic constraint. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getaccntot() -> (n)


Obtains the total dimension of all affine conic constraints (the sum of all their dimensions).

Return

n (int) – The total dimension of all affine conic constraints.

Groups
getaccs(domidxlist,afeidxlist,b)

getaccs() -> (domidxlist,afeidxlist,b)


Obtains full data of all affine conic constraints. The output array domainidxlist must have at least length determined by Task.getnumacc. The output arrays afeidxlist and b must have at least length determined by Task.getaccntot.

Parameters
• domidxlist (int[]) – The list of domains appearing in all affine conic constraints. (output)

• afeidxlist (int[]) – The concatenation of index lists of affine expressions appearing in all affine conic constraints. (output)

• b (float[]) – The concatenation of vectors b appearing in all affine conic constraints. (output)

Return
• domidxlist (int[]) – The list of domains appearing in all affine conic constraints.

• afeidxlist (int[]) – The concatenation of index lists of affine expressions appearing in all affine conic constraints.

• b (float[]) – The concatenation of vectors b appearing in all affine conic constraints.

Groups
getacol(j,subj,valj) -> (nzj)

getacol(j) -> (nzj,subj,valj)


Obtains one column of $$A$$ in a sparse format.

Parameters
• j (int) – Index of the column. (input)

• subj (int[]) – Row indices of the non-zeros in the column obtained. (output)

• valj (float[]) – Numerical values in the column obtained. (output)

Return
• nzj (int) – Number of non-zeros in the column obtained.

• subj (int[]) – Row indices of the non-zeros in the column obtained.

• valj (float[]) – Numerical values in the column obtained.

Groups
getacolnumnz(i) -> (nzj)


Obtains the number of non-zero elements in one column of $$A$$.

Parameters

i (int) – Index of the column. (input)

Return

nzj (int) – Number of non-zeros in the $$j$$-th column of $$A$$.

Groups
getacolslice(first,last,ptrb,ptre,sub,val)

getacolslice(first,last) -> (ptrb,ptre,sub,val)


Obtains a sequence of columns from $$A$$ in sparse format.

Parameters
• first (int) – Index of the first column in the sequence. (input)

• last (int) – Index of the last column in the sequence plus one. (input)

• ptrb (int[]) – ptrb[t] is an index pointing to the first element in the $$t$$-th column obtained. (output)

• ptre (int[]) – ptre[t] is an index pointing to the last element plus one in the $$t$$-th column obtained. (output)

• sub (int[]) – Contains the row subscripts. (output)

• val (float[]) – Contains the coefficient values. (output)

Return
• ptrb (int[]) – ptrb[t] is an index pointing to the first element in the $$t$$-th column obtained.

• ptre (int[]) – ptre[t] is an index pointing to the last element plus one in the $$t$$-th column obtained.

• sub (int[]) – Contains the row subscripts.

• val (float[]) – Contains the coefficient values.

Groups
getacolslicenumnz(first,last) -> (numnz)


Obtains the number of non-zeros in a slice of columns of $$A$$.

Parameters
• first (int) – Index of the first column in the sequence. (input)

• last (int) – Index of the last column plus one in the sequence. (input)

Return

numnz (int) – Number of non-zeros in the slice.

Groups
getacolslicetrip(first,last,subi,subj,val)

getacolslicetrip(first,last) -> (subi,subj,val)


Obtains a sequence of columns from $$A$$ in sparse triplet format. The function returns the content of all columns whose index j satisfies first <= j < last. The triplets corresponding to nonzero entries are stored in the arrays subi, subj and val.

Parameters
• first (int) – Index of the first column in the sequence. (input)

• last (int) – Index of the last column in the sequence plus one. (input)

• subi (int[]) – Constraint subscripts. (output)

• subj (int[]) – Column subscripts. (output)

• val (float[]) – Values. (output)

Return
• subi (int[]) – Constraint subscripts.

• subj (int[]) – Column subscripts.

• val (float[]) – Values.

Groups
getafebarfblocktriplet(afeidx,barvaridx,subk,subl,valkl) -> (numtrip)

getafebarfblocktriplet() ->
(numtrip,
afeidx,
barvaridx,
subk,
subl,
valkl)


Obtains $$\barF$$ in block triplet form.

Parameters
• afeidx (int[]) – Constraint index. (output)

• barvaridx (int[]) – Symmetric matrix variable index. (output)

• subk (int[]) – Block row index. (output)

• subl (int[]) – Block column index. (output)

• valkl (float[]) – The numerical value associated with each block triplet. (output)

Return
• numtrip (int) – Number of elements in the block triplet form.

• afeidx (int[]) – Constraint index.

• barvaridx (int[]) – Symmetric matrix variable index.

• subk (int[]) – Block row index.

• subl (int[]) – Block column index.

• valkl (float[]) – The numerical value associated with each block triplet.

Groups
getafebarfnumblocktriplets() -> (numtrip)


Obtains an upper bound on the number of elements in the block triplet form of $$\barF$$.

Return

numtrip (int) – An upper bound on the number of elements in the block triplet form of $$\barF.$$

Groups
getafebarfnumrowentries(afeidx) -> (numentr)


Obtains the number of nonzero entries in one row of $$\barF$$, that is the number of $$j$$ such that $$\barF_{\mathrm{afeidx},j}$$ is not the zero matrix.

Parameters

afeidx (int) – Row index of $$\barF$$. (input)

Return

numentr (int) – Number of nonzero entries in a row of $$\barF$$.

Groups
getafebarfrow(afeidx,
barvaridx,
ptrterm,
numterm,
termidx,
termweight)

getafebarfrow(afeidx) ->
(barvaridx,
ptrterm,
numterm,
termidx,
termweight)


Obtains all nonzero entries in one row $$\barF_{\mathrm{afeidx},*}$$ of $$\barF$$. For every $$k$$ there is a nonzero entry $$\barF_{\mathrm{afeidx}, \mathrm{barvaridx}[k]}$$, which is represented as a weighted sum of $$\mathrm{numterm}[k]$$ terms. The indices in the matrix store $$E$$ and their weights for the $$k$$-th entry appear in the arrays termidx and termweight in positions

$\mathrm{ptrterm}[k],\ldots,\mathrm{ptrterm}[k]+\mathrm{numterm}[k]-1.$

The arrays should be long enough to accommodate the data; their required lengths can be obtained with Task.getafebarfrowinfo.

Parameters
• afeidx (int) – Row index of $$\barF$$. (input)

• barvaridx (int[]) – Semidefinite variable indices of nonzero entries in the row of $$\barF$$. (output)

• ptrterm (int[]) – Pointers to the start of each entry’s description. (output)

• numterm (int[]) – Number of terms in the weighted sum representation of each entry. (output)

• termidx (int[]) – Indices of semidefinite matrices from the matrix store $$E$$. (output)

• termweight (float[]) – Weights appearing in the weighted sum representations of all entries. (output)

Return
• barvaridx (int[]) – Semidefinite variable indices of nonzero entries in the row of $$\barF$$.

• ptrterm (int[]) – Pointers to the start of each entry’s description.

• numterm (int[]) – Number of terms in the weighted sum representation of each entry.

• termidx (int[]) – Indices of semidefinite matrices from the matrix store $$E$$.

• termweight (float[]) – Weights appearing in the weighted sum representations of all entries.

Groups
getafebarfrowinfo(afeidx) -> (numentr,numterm)


Obtains information about one row of $$\barF$$: the number of nonzero entries, that is the number of $$j$$ such that $$\barF_{\mathrm{afeidx},j}$$ is not the zero matrix, as well as the total number of terms in the representations of all these entries as weighted sums of matrices from $$E$$. This information provides the data sizes required for a call to Task.getafebarfrow.

Parameters

afeidx (int) – Row index of $$\barF$$. (input)

Return
• numentr (int) – Number of nonzero entries in a row of $$\barF$$.

• numterm (int) – Number of terms in the weighted sums representation of the row of $$\barF$$.

Groups
getafefnumnz() -> (numnz)


Obtains the total number of nonzeros in $$F$$.

Return

numnz (int) – Number of non-zeros in $$F$$.

Groups
getafefrow(afeidx,varidx,val) -> (numnz)

getafefrow(afeidx) -> (numnz,varidx,val)


Obtains one row of $$F$$ in sparse format.

Parameters
• afeidx (int) – Index of a row in $$F$$. (input)

• varidx (int[]) – Column indices of the non-zeros in the row obtained. (output)

• val (float[]) – Values of the non-zeros in the row obtained. (output)

Return
• numnz (int) – Number of non-zeros in the row obtained.

• varidx (int[]) – Column indices of the non-zeros in the row obtained.

• val (float[]) – Values of the non-zeros in the row obtained.

Groups
getafefrownumnz(afeidx) -> (numnz)


Obtains the number of nonzeros in one row of $$F$$.

Parameters

afeidx (int) – Index of a row in $$F$$. (input)

Return

numnz (int) – Number of non-zeros in row afeidx of $$F$$.

Groups
getafeftrip(afeidx,varidx,val)

getafeftrip() -> (afeidx,varidx,val)


Obtains the $$F$$ in triplet format.

Parameters
• afeidx (int[]) – Row indices of nonzeros. (output)

• varidx (int[]) – Column indices of nonzeros. (output)

• val (float[]) – Values of nonzero entries. (output)

Return
• afeidx (int[]) – Row indices of nonzeros.

• varidx (int[]) – Column indices of nonzeros.

• val (float[]) – Values of nonzero entries.

Groups
getafeg(afeidx) -> (g)


Obtains a single coefficient in $$g$$.

Parameters

afeidx (int) – Index of an element in $$g$$. (input)

Return

g (float) – The value of $$g_{\mathrm{afeidx}}$$.

Groups
getafegslice(first,last,g)

getafegslice(first,last) -> (g)


Obtains a sequence of elements from the vector $$g$$ of constant terms in the affine expressions list.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• g (float[]) – The slice $$g$$ as a dense vector. The length is last-first. (output)

Return

g (float[]) – The slice $$g$$ as a dense vector. The length is last-first.

Groups
getaij(i,j) -> (aij)


Obtains a single coefficient in $$A$$.

Parameters
• i (int) – Row index of the coefficient to be returned. (input)

• j (int) – Column index of the coefficient to be returned. (input)

Return

aij (float) – The required coefficient $$a_{i,j}$$.

Groups
getapiecenumnz(firsti,lasti,firstj,lastj) -> (numnz)


Obtains the number non-zeros in a rectangular piece of $$A$$, i.e. the number of elements in the set

$\{ (i,j)~:~ a_{i,j} \neq 0,~ \mathtt{firsti} \leq i \leq \mathtt{lasti}-1, ~\mathtt{firstj} \leq j \leq \mathtt{lastj}-1\}$

This function is not an efficient way to obtain the number of non-zeros in one row or column. In that case use the function Task.getarownumnz or Task.getacolnumnz.

Parameters
• firsti (int) – Index of the first row in the rectangular piece. (input)

• lasti (int) – Index of the last row plus one in the rectangular piece. (input)

• firstj (int) – Index of the first column in the rectangular piece. (input)

• lastj (int) – Index of the last column plus one in the rectangular piece. (input)

Return

numnz (int) – Number of non-zero $$A$$ elements in the rectangular piece.

Groups
getarow(i,subi,vali) -> (nzi)

getarow(i) -> (nzi,subi,vali)


Obtains one row of $$A$$ in a sparse format.

Parameters
• i (int) – Index of the row. (input)

• subi (int[]) – Column indices of the non-zeros in the row obtained. (output)

• vali (float[]) – Numerical values of the row obtained. (output)

Return
• nzi (int) – Number of non-zeros in the row obtained.

• subi (int[]) – Column indices of the non-zeros in the row obtained.

• vali (float[]) – Numerical values of the row obtained.

Groups
getarownumnz(i) -> (nzi)


Obtains the number of non-zero elements in one row of $$A$$.

Parameters

i (int) – Index of the row. (input)

Return

nzi (int) – Number of non-zeros in the $$i$$-th row of $$A$$.

Groups
getarowslice(first,last,ptrb,ptre,sub,val)

getarowslice(first,last) -> (ptrb,ptre,sub,val)


Obtains a sequence of rows from $$A$$ in sparse format.

Parameters
• first (int) – Index of the first row in the sequence. (input)

• last (int) – Index of the last row in the sequence plus one. (input)

• ptrb (int[]) – ptrb[t] is an index pointing to the first element in the $$t$$-th row obtained. (output)

• ptre (int[]) – ptre[t] is an index pointing to the last element plus one in the $$t$$-th row obtained. (output)

• sub (int[]) – Contains the column subscripts. (output)

• val (float[]) – Contains the coefficient values. (output)

Return
• ptrb (int[]) – ptrb[t] is an index pointing to the first element in the $$t$$-th row obtained.

• ptre (int[]) – ptre[t] is an index pointing to the last element plus one in the $$t$$-th row obtained.

• sub (int[]) – Contains the column subscripts.

• val (float[]) – Contains the coefficient values.

Groups
getarowslicenumnz(first,last) -> (numnz)


Obtains the number of non-zeros in a slice of rows of $$A$$.

Parameters
• first (int) – Index of the first row in the sequence. (input)

• last (int) – Index of the last row plus one in the sequence. (input)

Return

numnz (int) – Number of non-zeros in the slice.

Groups
getarowslicetrip(first,last,subi,subj,val)

getarowslicetrip(first,last) -> (subi,subj,val)


Obtains a sequence of rows from $$A$$ in sparse triplet format. The function returns the content of all rows whose index i satisfies first <= i < last. The triplets corresponding to nonzero entries are stored in the arrays subi, subj and val.

Parameters
• first (int) – Index of the first row in the sequence. (input)

• last (int) – Index of the last row in the sequence plus one. (input)

• subi (int[]) – Constraint subscripts. (output)

• subj (int[]) – Column subscripts. (output)

• val (float[]) – Values. (output)

Return
• subi (int[]) – Constraint subscripts.

• subj (int[]) – Column subscripts.

• val (float[]) – Values.

Groups
getatrip(subi,subj,val)

getatrip() -> (subi,subj,val)


Obtains $$A$$ in sparse triplet format. The triplets corresponding to nonzero entries are stored in the arrays subi, subj and val.

Parameters
• subi (int[]) – Constraint subscripts. (output)

• subj (int[]) – Column subscripts. (output)

• val (float[]) – Values. (output)

Return
• subi (int[]) – Constraint subscripts.

• subj (int[]) – Column subscripts.

• val (float[]) – Values.

Groups
getatruncatetol(tolzero)

getatruncatetol() -> (tolzero)


Obtains the tolerance value set with Task.putatruncatetol.

Parameters

tolzero (float[]) – All elements $$|a_{i,j}|$$ less than this tolerance is truncated to zero. (output)

Return

tolzero (float[]) – All elements $$|a_{i,j}|$$ less than this tolerance is truncated to zero.

Groups
getbarablocktriplet(subi,subj,subk,subl,valijkl) -> (num)

getbarablocktriplet() -> (num,subi,subj,subk,subl,valijkl)


Obtains $$\barA$$ in block triplet form.

Parameters
• subi (int[]) – Constraint index. (output)

• subj (int[]) – Symmetric matrix variable index. (output)

• subk (int[]) – Block row index. (output)

• subl (int[]) – Block column index. (output)

• valijkl (float[]) – The numerical value associated with each block triplet. (output)

Return
• num (int) – Number of elements in the block triplet form.

• subi (int[]) – Constraint index.

• subj (int[]) – Symmetric matrix variable index.

• subk (int[]) – Block row index.

• subl (int[]) – Block column index.

• valijkl (float[]) – The numerical value associated with each block triplet.

Groups
getbaraidx(idx,sub,weights) -> (i,j,num)

getbaraidx(idx) -> (i,j,num,sub,weights)


Obtains information about an element in $$\barA$$. Since $$\barA$$ is a sparse matrix of symmetric matrices, only the nonzero elements in $$\barA$$ are stored in order to save space. Now $$\barA$$ is stored vectorized i.e. as one long vector. This function makes it possible to obtain information such as the row index and the column index of a particular element of the vectorized form of $$\barA$$.

Please observe if one element of $$\barA$$ is inputted multiple times then it may be stored several times in vectorized form. In that case the element with the highest index is the one that is used.

Parameters
• idx (int) – Position of the element in the vectorized form. (input)

• sub (int[]) – A list indexes of the elements from symmetric matrix storage that appear in the weighted sum. (output)

• weights (float[]) – The weights associated with each term in the weighted sum. (output)

Return
• i (int) – Row index of the element at position idx.

• j (int) – Column index of the element at position idx.

• num (int) – Number of terms in weighted sum that forms the element.

• sub (int[]) – A list indexes of the elements from symmetric matrix storage that appear in the weighted sum.

• weights (float[]) – The weights associated with each term in the weighted sum.

Groups
getbaraidxij(idx) -> (i,j)


Obtains information about an element in $$\barA$$. Since $$\barA$$ is a sparse matrix of symmetric matrices, only the nonzero elements in $$\barA$$ are stored in order to save space. Now $$\barA$$ is stored vectorized i.e. as one long vector. This function makes it possible to obtain information such as the row index and the column index of a particular element of the vectorized form of $$\barA$$.

Please note that if one element of $$\barA$$ is inputted multiple times then it may be stored several times in vectorized form. In that case the element with the highest index is the one that is used.

Parameters

idx (int) – Position of the element in the vectorized form. (input)

Return
• i (int) – Row index of the element at position idx.

• j (int) – Column index of the element at position idx.

Groups
getbaraidxinfo(idx) -> (num)


Each nonzero element in $$\barA_{ij}$$ is formed as a weighted sum of symmetric matrices. Using this function the number of terms in the weighted sum can be obtained. See description of Task.appendsparsesymmat for details about the weighted sum.

Parameters

idx (int) – The internal position of the element for which information should be obtained. (input)

Return

num (int) – Number of terms in the weighted sum that form the specified element in $$\barA$$.

Groups
getbarasparsity(idxij) -> (numnz)

getbarasparsity() -> (numnz,idxij)


The matrix $$\barA$$ is assumed to be a sparse matrix of symmetric matrices. This implies that many of the elements in $$\barA$$ are likely to be zero matrices. Therefore, in order to save space, only nonzero elements in $$\barA$$ are stored on vectorized form. This function is used to obtain the sparsity pattern of $$\barA$$ and the position of each nonzero element in the vectorized form of $$\barA$$. From the index detailed information about each nonzero $$\barA_{i,j}$$ can be obtained using Task.getbaraidxinfo and Task.getbaraidx.

Parameters

idxij (int[]) – Position of each nonzero element in the vectorized form of $$\barA$$. (output)

Return
• numnz (int) – Number of nonzero elements in $$\barA$$.

• idxij (int[]) – Position of each nonzero element in the vectorized form of $$\barA$$.

Groups
getbarcblocktriplet(subj,subk,subl,valjkl) -> (num)

getbarcblocktriplet() -> (num,subj,subk,subl,valjkl)


Obtains $$\barC$$ in block triplet form.

Parameters
• subj (int[]) – Symmetric matrix variable index. (output)

• subk (int[]) – Block row index. (output)

• subl (int[]) – Block column index. (output)

• valjkl (float[]) – The numerical value associated with each block triplet. (output)

Return
• num (int) – Number of elements in the block triplet form.

• subj (int[]) – Symmetric matrix variable index.

• subk (int[]) – Block row index.

• subl (int[]) – Block column index.

• valjkl (float[]) – The numerical value associated with each block triplet.

Groups
getbarcidx(idx,sub,weights) -> (j,num)

getbarcidx(idx) -> (j,num,sub,weights)


Obtains information about an element in $$\barC$$.

Parameters
• idx (int) – Index of the element for which information should be obtained. (input)

• sub (int[]) – Elements appearing the weighted sum. (output)

• weights (float[]) – Weights of terms in the weighted sum. (output)

Return
• j (int) – Row index in $$\barC$$.

• num (int) – Number of terms in the weighted sum.

• sub (int[]) – Elements appearing the weighted sum.

• weights (float[]) – Weights of terms in the weighted sum.

Groups
getbarcidxinfo(idx) -> (num)


Obtains the number of terms in the weighted sum that forms a particular element in $$\barC$$.

Parameters

idx (int) – Index of the element for which information should be obtained. The value is an index of a symmetric sparse variable. (input)

Return

num (int) – Number of terms that appear in the weighted sum that forms the requested element.

Groups
getbarcidxj(idx) -> (j)


Obtains the row index of an element in $$\barC$$.

Parameters

idx (int) – Index of the element for which information should be obtained. (input)

Return

j (int) – Row index in $$\barC$$.

Groups
getbarcsparsity(idxj) -> (numnz)

getbarcsparsity() -> (numnz,idxj)


Internally only the nonzero elements of $$\barC$$ are stored in a vector. This function is used to obtain the nonzero elements of $$\barC$$ and their indexes in the internal vector representation (in idx). From the index detailed information about each nonzero $$\barC_j$$ can be obtained using Task.getbarcidxinfo and Task.getbarcidx.

Parameters

idxj (int[]) – Internal positions of the nonzeros elements in $$\barC$$. (output)

Return
• numnz (int) – Number of nonzero elements in $$\barC$$.

• idxj (int[]) – Internal positions of the nonzeros elements in $$\barC$$.

Groups
getbarsj(whichsol,j,barsj)

getbarsj(whichsol,j) -> (barsj)


Obtains the dual solution for a semidefinite variable. Only the lower triangular part of $$\barS_j$$ is returned because the matrix by construction is symmetric. The format is that the columns are stored sequentially in the natural order.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• j (int) – Index of the semidefinite variable. (input)

• barsj (float[]) – Value of $$\barS_j$$. (output)

Return

barsj (float[]) – Value of $$\barS_j$$.

Groups

Solution - semidefinite

getbarsslice(whichsol,first,last,slicesize,barsslice)

getbarsslice(whichsol,first,last,slicesize) -> (barsslice)


Obtains the dual solution for a sequence of semidefinite variables. The format is that matrices are stored sequentially, and in each matrix the columns are stored as in Task.getbarsj.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – Index of the first semidefinite variable in the slice. (input)

• last (int) – Index of the last semidefinite variable in the slice plus one. (input)

• slicesize (int) – Denotes the length of the array barsslice. (input)

• barsslice (float[]) – Dual solution values of symmetric matrix variables in the slice, stored sequentially. (output)

Return

barsslice (float[]) – Dual solution values of symmetric matrix variables in the slice, stored sequentially.

Groups

Solution - semidefinite

getbarvarname(i) -> (name)


Obtains the name of a semidefinite variable.

Parameters

i (int) – Index of the variable. (input)

Return

name (str) – The requested name is copied to this buffer.

Groups
getbarvarnameindex(somename) -> (asgn,index)


Obtains the index of semidefinite variable from its name.

Parameters

somename (str) – The name of the variable. (input)

Return
• asgn (int) – Non-zero if the name somename is assigned to some semidefinite variable.

• index (int) – The index of a semidefinite variable with the name somename (if one exists).

Groups
getbarvarnamelen(i) -> (len)


Obtains the length of the name of a semidefinite variable.

Parameters

i (int) – Index of the variable. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getbarxj(whichsol,j,barxj)

getbarxj(whichsol,j) -> (barxj)


Obtains the primal solution for a semidefinite variable. Only the lower triangular part of $$\barX_j$$ is returned because the matrix by construction is symmetric. The format is that the columns are stored sequentially in the natural order.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• j (int) – Index of the semidefinite variable. (input)

• barxj (float[]) – Value of $$\barX_j$$. (output)

Return

barxj (float[]) – Value of $$\barX_j$$.

Groups

Solution - semidefinite

getbarxslice(whichsol,first,last,slicesize,barxslice)

getbarxslice(whichsol,first,last,slicesize) -> (barxslice)


Obtains the primal solution for a sequence of semidefinite variables. The format is that matrices are stored sequentially, and in each matrix the columns are stored as in Task.getbarxj.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – Index of the first semidefinite variable in the slice. (input)

• last (int) – Index of the last semidefinite variable in the slice plus one. (input)

• slicesize (int) – Denotes the length of the array barxslice. (input)

• barxslice (float[]) – Solution values of symmetric matrix variables in the slice, stored sequentially. (output)

Return

barxslice (float[]) – Solution values of symmetric matrix variables in the slice, stored sequentially.

Groups

Solution - semidefinite

getc(c)

getc() -> (c)


Obtains all objective coefficients $$c$$.

Parameters

c (float[]) – Linear terms of the objective as a dense vector. The length is the number of variables. (output)

Return

c (float[]) – Linear terms of the objective as a dense vector. The length is the number of variables.

Groups
getcfix() -> (cfix)


Obtains the fixed term in the objective.

Return

cfix (float) – Fixed term in the objective.

Groups
getcj(j) -> (cj)


Obtains one coefficient of $$c$$.

Parameters

j (int) – Index of the variable for which the $$c$$ coefficient should be obtained. (input)

Return

cj (float) – The value of $$c_j$$.

Groups
getclist(subj,c)

getclist(subj) -> (c)


Obtains a sequence of elements in $$c$$.

Parameters
• subj (int[]) – A list of variable indexes. (input)

• c (float[]) – Linear terms of the requested list of the objective as a dense vector. (output)

Return

c (float[]) – Linear terms of the requested list of the objective as a dense vector.

Groups
getconbound(i) -> (bk,bl,bu)


Obtains bound information for one constraint.

Parameters

i (int) – Index of the constraint for which the bound information should be obtained. (input)

Return
• bk (mosek.boundkey) – Bound keys.

• bl (float) – Values for lower bounds.

• bu (float) – Values for upper bounds.

Groups
getconboundslice(first,last,bk,bl,bu)

getconboundslice(first,last) -> (bk,bl,bu)


Obtains bounds information for a slice of the constraints.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bk (mosek.boundkey[]) – Bound keys. (output)

• bl (float[]) – Values for lower bounds. (output)

• bu (float[]) – Values for upper bounds. (output)

Return
• bk (mosek.boundkey[]) – Bound keys.

• bl (float[]) – Values for lower bounds.

• bu (float[]) – Values for upper bounds.

Groups
getcone(k,submem) -> (ct,conepar,nummem)

getcone(k) -> (ct,conepar,nummem,submem)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters
• k (int) – Index of the cone. (input)

• submem (int[]) – Variable subscripts of the members in the cone. (output)

Return
• ct (mosek.conetype) – Specifies the type of the cone.

• conepar (float) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0.

• nummem (int) – Number of member variables in the cone.

• submem (int[]) – Variable subscripts of the members in the cone.

Groups
getconeinfo(k) -> (ct,conepar,nummem)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters

k (int) – Index of the cone. (input)

Return
• ct (mosek.conetype) – Specifies the type of the cone.

• conepar (float) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0.

• nummem (int) – Number of member variables in the cone.

Groups
getconename(i) -> (name)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters

i (int) – Index of the cone. (input)

Return

name (str) – The required name.

Groups
getconenameindex(somename) -> (asgn,index)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Checks whether the name somename has been assigned to any cone. If it has been assigned to a cone, then the index of the cone is reported.

Parameters

somename (str) – The name which should be checked. (input)

Return
• asgn (int) – Is non-zero if the name somename is assigned to some cone.

• index (int) – If the name somename is assigned to some cone, then index is the index of the cone.

Groups
getconenamelen(i) -> (len)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters

i (int) – Index of the cone. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getconname(i) -> (name)


Obtains the name of a constraint.

Parameters

i (int) – Index of the constraint. (input)

Return

name (str) – The required name.

Groups
getconnameindex(somename) -> (asgn,index)


Checks whether the name somename has been assigned to any constraint. If so, the index of the constraint is reported.

Parameters

somename (str) – The name which should be checked. (input)

Return
• asgn (int) – Is non-zero if the name somename is assigned to some constraint.

• index (int) – If the name somename is assigned to a constraint, then index is the index of the constraint.

Groups
getconnamelen(i) -> (len)


Obtains the length of the name of a constraint.

Parameters

i (int) – Index of the constraint. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getcslice(first,last,c)

getcslice(first,last) -> (c)


Obtains a sequence of elements in $$c$$.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• c (float[]) – Linear terms of the requested slice of the objective as a dense vector. The length is last-first. (output)

Return

c (float[]) – Linear terms of the requested slice of the objective as a dense vector. The length is last-first.

Groups
getdimbarvarj(j) -> (dimbarvarj)


Obtains the dimension of a symmetric matrix variable.

Parameters

j (int) – Index of the semidefinite variable whose dimension is requested. (input)

Return

dimbarvarj (int) – The dimension of the $$j$$-th semidefinite variable.

Groups
getdjcafeidxlist(djcidx,afeidxlist)

getdjcafeidxlist(djcidx) -> (afeidxlist)


Obtains the list of affine expression indexes in a disjunctive constraint.

Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• afeidxlist (int[]) – List of affine expression indexes. (output)

Return

afeidxlist (int[]) – List of affine expression indexes.

Groups
getdjcb(djcidx,b)

getdjcb(djcidx) -> (b)


Obtains the optional constant term vector of a disjunctive constraint.

Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• b (float[]) – The vector b. (output)

Return

b (float[]) – The vector b.

Groups
getdjcdomainidxlist(djcidx,domidxlist)

getdjcdomainidxlist(djcidx) -> (domidxlist)


Obtains the list of domain indexes in a disjunctive constraint.

Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• domidxlist (int[]) – List of term sizes. (output)

Return

domidxlist (int[]) – List of term sizes.

Groups
getdjcname(djcidx) -> (name)


Obtains the name of a disjunctive constraint.

Parameters

djcidx (int) – Index of a disjunctive constraint. (input)

Return

name (str) – Returns the required name.

Groups
getdjcnamelen(djcidx) -> (len)


Obtains the length of the name of a disjunctive constraint.

Parameters

djcidx (int) – Index of a disjunctive constraint. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getdjcnumafe(djcidx) -> (numafe)


Obtains the number of affine expressions in the disjunctive constraint.

Parameters

djcidx (int) – Index of the disjunctive constraint. (input)

Return

numafe (int) – Number of affine expressions in the disjunctive constraint.

Groups
getdjcnumafetot() -> (numafetot)


Obtains the total number of affine expressions in all disjunctive constraints.

Return

numafetot (int) – Number of affine expressions in all disjunctive constraints.

Groups
getdjcnumdomain(djcidx) -> (numdomain)


Obtains the number of domains in the disjunctive constraint.

Parameters

djcidx (int) – Index of the disjunctive constraint. (input)

Return

numdomain (int) – Number of domains in the disjunctive constraint.

Groups
getdjcnumdomaintot() -> (numdomaintot)


Obtains the total number of domains in all disjunctive constraints.

Return

numdomaintot (int) – Number of domains in all disjunctive constraints.

Groups
getdjcnumterm(djcidx) -> (numterm)


Obtains the number terms in the disjunctive constraint.

Parameters

djcidx (int) – Index of the disjunctive constraint. (input)

Return

numterm (int) – Number of terms in the disjunctive constraint.

Groups
getdjcnumtermtot() -> (numtermtot)


Obtains the total number of terms in all disjunctive constraints.

Return

numtermtot (int) – Total number of terms in all disjunctive constraints.

Groups
getdjcs(domidxlist,
afeidxlist,
b,
termsizelist,
numterms)

getdjcs() ->
(domidxlist,
afeidxlist,
b,
termsizelist,
numterms)


Obtains full data of all disjunctive constraints. The output arrays must have minimal lengths determined by the following methods: domainidxlist by Task.getdjcnumdomaintot, afeidxlist and b by Task.getdjcnumafetot, termsizelist by Task.getdjcnumtermtot and numterms by Task.getnumdomain.

Parameters
• domidxlist (int[]) – The concatenation of index lists of domains appearing in all disjunctive constraints. (output)

• afeidxlist (int[]) – The concatenation of index lists of affine expressions appearing in all disjunctive constraints. (output)

• b (float[]) – The concatenation of vectors b appearing in all disjunctive constraints. (output)

• termsizelist (int[]) – The concatenation of lists of term sizes appearing in all disjunctive constraints. (output)

• numterms (int[]) – The number of terms in each of the disjunctive constraints. (output)

Return
• domidxlist (int[]) – The concatenation of index lists of domains appearing in all disjunctive constraints.

• afeidxlist (int[]) – The concatenation of index lists of affine expressions appearing in all disjunctive constraints.

• b (float[]) – The concatenation of vectors b appearing in all disjunctive constraints.

• termsizelist (int[]) – The concatenation of lists of term sizes appearing in all disjunctive constraints.

• numterms (int[]) – The number of terms in each of the disjunctive constraints.

Groups
getdjctermsizelist(djcidx,termsizelist)

getdjctermsizelist(djcidx) -> (termsizelist)


Obtains the list of term sizes in a disjunctive constraint.

Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• termsizelist (int[]) – List of term sizes. (output)

Return

termsizelist (int[]) – List of term sizes.

Groups
getdomainn(domidx) -> (n)


Obtains the dimension of the domain.

Parameters

domidx (int) – Index of the domain. (input)

Return

n (int) – Dimension of the domain.

Groups
getdomainname(domidx) -> (name)


Obtains the name of a domain.

Parameters

domidx (int) – Index of a domain. (input)

Return

name (str) – Returns the required name.

Groups
getdomainnamelen(domidx) -> (len)


Obtains the length of the name of a domain.

Parameters

domidx (int) – Index of a domain. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getdomaintype(domidx) -> (domtype)


Returns the type of the domain.

Parameters

domidx (int) – Index of the domain. (input)

Return

domtype (mosek.domaintype) – The type of the domain.

Groups
getdouinf(whichdinf) -> (dvalue)


Obtains a double information item from the task information database.

Parameters

whichdinf (mosek.dinfitem) – Specifies a double information item. (input)

Return

dvalue (float) – The value of the required double information item.

Groups

Information items and statistics

getdouparam(param) -> (parvalue)


Obtains the value of a double parameter.

Parameters

param (mosek.dparam) – Which parameter. (input)

Return

parvalue (float) – Parameter value.

Groups

Parameters

getdualobj(whichsol) -> (dualobj)


Computes the dual objective value associated with the solution. Note that if the solution is a primal infeasibility certificate, then the fixed term in the objective value is not included.

Moreover, since there is no dual solution associated with an integer solution, an error will be reported if the dual objective value is requested for the integer solution.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return

dualobj (float) – Objective value corresponding to the dual solution.

Groups
getdualsolutionnorms(whichsol) ->
(nrmy,
nrmslc,
nrmsuc,
nrmslx,
nrmsux,
nrmsnx,
nrmbars)


Compute norms of the dual solution.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return
• nrmy (float) – The norm of the $$y$$ vector.

• nrmslc (float) – The norm of the $$s_l^c$$ vector.

• nrmsuc (float) – The norm of the $$s_u^c$$ vector.

• nrmslx (float) – The norm of the $$s_l^x$$ vector.

• nrmsux (float) – The norm of the $$s_u^x$$ vector.

• nrmsnx (float) – The norm of the $$s_n^x$$ vector.

• nrmbars (float) – The norm of the $$\barS$$ vector.

Groups

Solution information

getdviolacc(whichsol,accidxlist,viol)

getdviolacc(whichsol,accidxlist) -> (viol)


Let $$(s_n^x)^*$$ be the value of variable $$(s_n^x)$$ for the specified solution. For simplicity let us assume that $$s_n^x$$ is a member of a quadratic cone, then the violation is computed as follows

$\begin{split}\left\{ \begin{array}{ll} \max(0,(\|s_n^x\|_{2:n}^*-(s_n^x)_1^*) / \sqrt{2}, & (s_n^x)^* \geq -\|(s_n^x)_{2:n}^*\|, \\ \|(s_n^x)^*\|, & \mbox{otherwise.} \end{array} \right.\end{split}$

Both when the solution is a certificate of primal infeasibility or when it is a dual feasible solution the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• accidxlist (int[]) – An array of indexes of conic constraints. (input)

• viol (float[]) – viol[k] is the violation of the dual solution associated with the conic constraint sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation of the dual solution associated with the conic constraint sub[k].

Groups

Solution information

getdviolbarvar(whichsol,sub,viol)

getdviolbarvar(whichsol,sub) -> (viol)


Let $$(\barS_j)^*$$ be the value of variable $$\barS_j$$ for the specified solution. Then the dual violation of the solution associated with variable $$\barS_j$$ is given by

$\max(-\lambda_{\min}(\barS_j),\ 0.0).$

Both when the solution is a certificate of primal infeasibility and when it is dual feasible solution the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of $$\barX$$ variables. (input)

• viol (float[]) – viol[k] is the violation of the solution for the constraint $$\barS_{\mathtt{sub}[k]} \in \PSD$$. (output)

Return

viol (float[]) – viol[k] is the violation of the solution for the constraint $$\barS_{\mathtt{sub}[k]} \in \PSD$$.

Groups

Solution information

getdviolcon(whichsol,sub,viol)

getdviolcon(whichsol,sub) -> (viol)


The violation of the dual solution associated with the $$i$$-th constraint is computed as follows

$\max( \rho( (s_l^c)_i^*,(b_l^c)_i ),\ \rho( (s_u^c)_i^*, -(b_u^c)_i ),\ |-y_i+(s_l^c)_i^*-(s_u^c)_i^*| )$

where

$\begin{split}\rho(x,l) = \left\{ \begin{array}{ll} -x, & l > -\infty , \\ |x|, & \mbox{otherwise}.\\ \end{array} \right.\end{split}$

Both when the solution is a certificate of primal infeasibility or it is a dual feasible solution the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of constraints. (input)

• viol (float[]) – viol[k] is the violation of dual solution associated with the constraint sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation of dual solution associated with the constraint sub[k].

Groups

Solution information

getdviolcones(whichsol,sub,viol)

getdviolcones(whichsol,sub) -> (viol)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Let $$(s_n^x)^*$$ be the value of variable $$(s_n^x)$$ for the specified solution. For simplicity let us assume that $$s_n^x$$ is a member of a quadratic cone, then the violation is computed as follows

$\begin{split}\left\{ \begin{array}{ll} \max(0,(\|s_n^x\|_{2:n}^*-(s_n^x)_1^*) / \sqrt{2}, & (s_n^x)^* \geq -\|(s_n^x)_{2:n}^*\|, \\ \|(s_n^x)^*\|, & \mbox{otherwise.} \end{array} \right.\end{split}$

Both when the solution is a certificate of primal infeasibility or when it is a dual feasible solution the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of conic constraints. (input)

• viol (float[]) – viol[k] is the violation of the dual solution associated with the conic constraint sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation of the dual solution associated with the conic constraint sub[k].

Groups

Solution information

getdviolvar(whichsol,sub,viol)

getdviolvar(whichsol,sub) -> (viol)


The violation of the dual solution associated with the $$j$$-th variable is computed as follows

$\max \left(\rho((s_l^x)_j^*,(b_l^x)_j),\ \rho((s_u^x)_j^*,-(b_u^x)_j),\ |\sum_{i=\idxbeg}^{\idxend{numcon}} a_{ij} y_i+(s_l^x)_j^*-(s_u^x)_j^* - \tau c_j| \right)$

where

$\begin{split}\rho(x,l) = \left\{ \begin{array}{ll} -x, & l > -\infty , \\ |x|, & \mbox{otherwise} \end{array} \right.\end{split}$

and $$\tau=0$$ if the solution is a certificate of primal infeasibility and $$\tau=1$$ otherwise. The formula for computing the violation is only shown for the linear case but is generalized appropriately for the more general problems. Both when the solution is a certificate of primal infeasibility or when it is a dual feasible solution the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of $$x$$ variables. (input)

• viol (float[]) – viol[k] is the violation of dual solution associated with the variable sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation of dual solution associated with the variable sub[k].

Groups

Solution information

getinfeasiblesubproblem(whichsol) -> (inftask)


Given the solution is a certificate of primal or dual infeasibility then a primal or dual infeasible subproblem is obtained respectively. The subproblem tends to be much smaller than the original problem and hence it is easier to locate the infeasibility inspecting the subproblem than the original problem.

For the procedure to be useful it is important to assign meaningful names to constraints, variables etc. in the original task because those names will be duplicated in the subproblem.

The function is only applicable to linear and conic quadratic optimization problems.

For more information see Sec. 8.3 (Debugging infeasibility) and Sec. 14.2 (Automatic Repair of Infeasible Problems).

Parameters

whichsol (mosek.soltype) – Which solution to use when determining the infeasible subproblem. (input)

Return

inftask (mosek.Task) – A new task containing the infeasible subproblem.

Groups

Infeasibility diagnostic

getintinf(whichiinf) -> (ivalue)


Obtains an integer information item from the task information database.

Parameters

whichiinf (mosek.iinfitem) – Specifies an integer information item. (input)

Return

ivalue (int) – The value of the required integer information item.

Groups

Information items and statistics

getintparam(param) -> (parvalue)


Obtains the value of an integer parameter.

Parameters

param (mosek.iparam) – Which parameter. (input)

Return

parvalue (int) – Parameter value.

Groups

Parameters

getlenbarvarj(j) -> (lenbarvarj)


Obtains the length of the $$j$$-th semidefinite variable i.e. the number of elements in the lower triangular part.

Parameters

j (int) – Index of the semidefinite variable whose length if requested. (input)

Return

lenbarvarj (int) – Number of scalar elements in the lower triangular part of the semidefinite variable.

Groups
getlintinf(whichliinf) -> (ivalue)


Obtains a long integer information item from the task information database.

Parameters

whichliinf (mosek.liinfitem) – Specifies a long information item. (input)

Return

ivalue (int) – The value of the required long integer information item.

Groups

Information items and statistics

getmaxnumanz() -> (maxnumanz)


Obtains number of preallocated non-zeros in $$A$$. When this number of non-zeros is reached MOSEK will automatically allocate more space for $$A$$.

Return

maxnumanz (int) – Number of preallocated non-zero linear matrix elements.

Groups
getmaxnumbarvar() -> (maxnumbarvar)


Obtains maximum number of symmetric matrix variables for which space is currently preallocated.

Return

maxnumbarvar (int) – Maximum number of symmetric matrix variables for which space is currently preallocated.

Groups
getmaxnumcon() -> (maxnumcon)


Obtains the number of preallocated constraints in the optimization task. When this number of constraints is reached MOSEK will automatically allocate more space for constraints.

Return

maxnumcon (int) – Number of preallocated constraints in the optimization task.

Groups
getmaxnumcone() -> (maxnumcone)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Obtains the number of preallocated cones in the optimization task. When this number of cones is reached MOSEK will automatically allocate space for more cones.

Return

maxnumcone (int) – Number of preallocated conic constraints in the optimization task.

Groups
getmaxnumqnz() -> (maxnumqnz)


Obtains the number of preallocated non-zeros for $$Q$$ (both objective and constraints). When this number of non-zeros is reached MOSEK will automatically allocate more space for $$Q$$.

Return

maxnumqnz (int) – Number of non-zero elements preallocated in quadratic coefficient matrices.

Groups
getmaxnumvar() -> (maxnumvar)


Obtains the number of preallocated variables in the optimization task. When this number of variables is reached MOSEK will automatically allocate more space for variables.

Return

maxnumvar (int) – Number of preallocated variables in the optimization task.

Groups
getmemusage() -> (meminuse,maxmemuse)


Return
• meminuse (int) – Amount of memory currently used by the task.

• maxmemuse (int) – Maximum amount of memory used by the task until now.

Groups

System, memory and debugging

getnumacc() -> (num)


Obtains the number of affine conic constraints.

Return

num (int) – The number of affine conic constraints.

Groups
getnumafe() -> (numafe)


Obtains the number of affine expressions.

Return

numafe (int) – Number of affine expressions.

Groups
getnumanz() -> (numanz)


Obtains the number of non-zeros in $$A$$.

Return

numanz (int) – Number of non-zero elements in the linear constraint matrix.

Groups
getnumanz64() -> (numanz)


Obtains the number of non-zeros in $$A$$.

Return

numanz (int) – Number of non-zero elements in the linear constraint matrix.

Groups
getnumbarablocktriplets() -> (num)


Obtains an upper bound on the number of elements in the block triplet form of $$\barA$$.

Return

num (int) – An upper bound on the number of elements in the block triplet form of $$\barA.$$

Groups
getnumbaranz() -> (nz)


Get the number of nonzero elements in $$\barA$$.

Return

nz (int) – The number of nonzero block elements in $$\barA$$ i.e. the number of $$\barA_{ij}$$ elements that are nonzero.

Groups
getnumbarcblocktriplets() -> (num)


Obtains an upper bound on the number of elements in the block triplet form of $$\barC$$.

Return

num (int) – An upper bound on the number of elements in the block triplet form of $$\barC.$$

Groups
getnumbarcnz() -> (nz)


Obtains the number of nonzero elements in $$\barC$$.

Return

nz (int) – The number of nonzeros in $$\barC$$ i.e. the number of elements $$\barC_j$$ that are nonzero.

Groups
getnumbarvar() -> (numbarvar)


Obtains the number of semidefinite variables.

Return

numbarvar (int) – Number of semidefinite variables in the problem.

Groups
getnumcon() -> (numcon)


Obtains the number of constraints.

Return

numcon (int) – Number of constraints.

Groups
getnumcone() -> (numcone)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Return

numcone (int) – Number of conic constraints.

Groups
getnumconemem(k) -> (nummem)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters

k (int) – Index of the cone. (input)

Return

nummem (int) – Number of member variables in the cone.

Groups
getnumdjc() -> (num)


Obtains the number of disjunctive constraints.

Return

num (int) – The number of disjunctive constraints.

Groups
getnumdomain() -> (numdomain)


Obtain the number of domains defined.

Return

numdomain (int) – Number of domains in the task.

Groups
getnumintvar() -> (numintvar)


Obtains the number of integer-constrained variables.

Return

numintvar (int) – Number of integer variables.

Groups
getnumparam(partype) -> (numparam)


Obtains the number of parameters of a given type.

Parameters

partype (mosek.parametertype) – Parameter type. (input)

Return

numparam (int) – The number of parameters of type partype.

Groups
getnumqconknz(k) -> (numqcnz)


Obtains the number of non-zero quadratic terms in a constraint.

Parameters

k (int) – Index of the constraint for which the number quadratic terms should be obtained. (input)

Return

numqcnz (int) – Number of quadratic terms.

Groups
getnumqobjnz() -> (numqonz)


Obtains the number of non-zero quadratic terms in the objective.

Return

numqonz (int) – Number of non-zero elements in the quadratic objective terms.

Groups
getnumsymmat() -> (num)


Obtains the number of symmetric matrices stored in the vector $$E$$.

Return

num (int) – The number of symmetric sparse matrices.

Groups
getnumvar() -> (numvar)


Obtains the number of variables.

Return

numvar (int) – Number of variables.

Groups
getobjname() -> (objname)


Obtains the name assigned to the objective function.

Return

objname (str) – Assigned the objective name.

Groups
getobjnamelen() -> (len)


Obtains the length of the name assigned to the objective function.

Return

len (int) – Assigned the length of the objective name.

Groups
getobjsense() -> (sense)


Gets the objective sense of the task.

Return

sense (mosek.objsense) – The returned objective sense.

Groups

Problem data - linear part

getpowerdomainalpha(domidx,alpha)

getpowerdomainalpha(domidx) -> (alpha)


Obtains the exponent vector $$\alpha$$ of a primal or dual power cone domain.

Parameters
• domidx (int) – Index of the domain. (input)

• alpha (float[]) – The vector $$\alpha$$. (output)

Return

alpha (float[]) – The vector $$\alpha$$.

Groups
getpowerdomaininfo(domidx) -> (n,nleft)


Obtains structural information about a primal or dual power cone domain.

Parameters

domidx (int) – Index of the domain. (input)

Return
• n (int) – Dimension of the domain.

• nleft (int) – Number of variables on the left hand side.

Groups
getprimalobj(whichsol) -> (primalobj)


Computes the primal objective value for the desired solution. Note that if the solution is an infeasibility certificate, then the fixed term in the objective is not included.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return

primalobj (float) – Objective value corresponding to the primal solution.

Groups
getprimalsolutionnorms(whichsol) -> (nrmxc,nrmxx,nrmbarx)


Compute norms of the primal solution.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return
• nrmxc (float) – The norm of the $$x^c$$ vector.

• nrmxx (float) – The norm of the $$x$$ vector.

• nrmbarx (float) – The norm of the $$\barX$$ vector.

Groups

Solution information

getprobtype() -> (probtype)


Obtains the problem type.

Return

probtype (mosek.problemtype) – The problem type.

Groups

getprosta(whichsol) -> (problemsta)


Obtains the problem status.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return

problemsta (mosek.prosta) – Problem status.

Groups

Solution information

getpviolacc(whichsol,accidxlist,viol)

getpviolacc(whichsol,accidxlist) -> (viol)


Computes the primal solution violation for a set of affine conic constraints. Let $$x^*$$ be the value of the variable $$x$$ for the specified solution. For simplicity let us assume that $$x$$ is a member of a quadratic cone, then the violation is computed as follows

$\begin{split}\left\{ \begin{array}{ll} \max(0,\|x_{2:n}\|-x_1) / \sqrt{2}, & x_1 \geq -\|x_{2:n}\|, \\ \|x\|, & \mbox{otherwise.} \end{array} \right.\end{split}$

Both when the solution is a certificate of dual infeasibility or when it is primal feasible the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• accidxlist (int[]) – An array of indexes of conic constraints. (input)

• viol (float[]) – viol[k] is the violation of the solution associated with the affine conic constraint number accidxlist[k]. (output)

Return

viol (float[]) – viol[k] is the violation of the solution associated with the affine conic constraint number accidxlist[k].

Groups

Solution information

getpviolbarvar(whichsol,sub,viol)

getpviolbarvar(whichsol,sub) -> (viol)


Computes the primal solution violation for a set of semidefinite variables. Let $$(\barX_j)^*$$ be the value of the variable $$\barX_j$$ for the specified solution. Then the primal violation of the solution associated with variable $$\barX_j$$ is given by

$\max(-\lambda_{\min}(\barX_j),\ 0.0).$

Both when the solution is a certificate of dual infeasibility or when it is primal feasible the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of $$\barX$$ variables. (input)

• viol (float[]) – viol[k] is how much the solution violates the constraint $$\barX_{\mathtt{sub}[k]} \in \PSD$$. (output)

Return

viol (float[]) – viol[k] is how much the solution violates the constraint $$\barX_{\mathtt{sub}[k]} \in \PSD$$.

Groups

Solution information

getpviolcon(whichsol,sub,viol)

getpviolcon(whichsol,sub) -> (viol)


Computes the primal solution violation for a set of constraints. The primal violation of the solution associated with the $$i$$-th constraint is given by

$\max(\tau l_i^c - (x_i^c)^*,\ (x_i^c)^* - \tau u_i^c),\ |\sum_{j=\idxbeg}^{\idxend{numvar}} a_{ij} x_j^* - x_i^c|)$

where $$\tau=0$$ if the solution is a certificate of dual infeasibility and $$\tau=1$$ otherwise. Both when the solution is a certificate of dual infeasibility and when it is primal feasible the violation should be small. The above formula applies for the linear case but is appropriately generalized in other cases.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of constraints. (input)

• viol (float[]) – viol[k] is the violation associated with the solution for the constraint sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation associated with the solution for the constraint sub[k].

Groups

Solution information

getpviolcones(whichsol,sub,viol)

getpviolcones(whichsol,sub) -> (viol)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Computes the primal solution violation for a set of conic constraints. Let $$x^*$$ be the value of the variable $$x$$ for the specified solution. For simplicity let us assume that $$x$$ is a member of a quadratic cone, then the violation is computed as follows

$\begin{split}\left\{ \begin{array}{ll} \max(0,\|x_{2:n}\|-x_1) / \sqrt{2}, & x_1 \geq -\|x_{2:n}\|, \\ \|x\|, & \mbox{otherwise.} \end{array} \right.\end{split}$

Both when the solution is a certificate of dual infeasibility or when it is primal feasible the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of conic constraints. (input)

• viol (float[]) – viol[k] is the violation of the solution associated with the conic constraint number sub[k]. (output)

Return

viol (float[]) – viol[k] is the violation of the solution associated with the conic constraint number sub[k].

Groups

Solution information

getpvioldjc(whichsol,djcidxlist,viol)

getpvioldjc(whichsol,djcidxlist) -> (viol)


Computes the primal solution violation for a set of disjunctive constraints. For a single DJC the violation is defined as

$\mathrm{viol}\left(\bigvee_{i=1}^t \bigwedge_{j=1}^{s_i} T_{i,j}\right) = \min_{i=1,\ldots,t}\left(\max_{j=1,\ldots,s_j}(\mathrm{viol}(T_{i,j}))\right)$

where the violation of each simple term $$T_{i,j}$$ is defined as for an ordinary linear constraint.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• djcidxlist (int[]) – An array of indexes of disjunctive constraints. (input)

• viol (float[]) – viol[k] is the violation of the solution associated with the disjunctive constraint number djcidxlist[k]. (output)

Return

viol (float[]) – viol[k] is the violation of the solution associated with the disjunctive constraint number djcidxlist[k].

Groups

Solution information

getpviolvar(whichsol,sub,viol)

getpviolvar(whichsol,sub) -> (viol)


Computes the primal solution violation associated to a set of variables. Let $$x_j^*$$ be the value of $$x_j$$ for the specified solution. Then the primal violation of the solution associated with variable $$x_j$$ is given by

$\max( \tau l_j^x - x_j^*,\ x_j^* - \tau u_j^x,\ 0).$

where $$\tau=0$$ if the solution is a certificate of dual infeasibility and $$\tau=1$$ otherwise. Both when the solution is a certificate of dual infeasibility and when it is primal feasible the violation should be small.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sub (int[]) – An array of indexes of $$x$$ variables. (input)

• viol (float[]) – viol[k] is the violation associated with the solution for the variable $$x_\mathtt{sub[k]}$$. (output)

Return

viol (float[]) – viol[k] is the violation associated with the solution for the variable $$x_\mathtt{sub[k]}$$.

Groups

Solution information

getqconk(k,qcsubi,qcsubj,qcval) -> (numqcnz)

getqconk(k) -> (numqcnz,qcsubi,qcsubj,qcval)


Obtains all the quadratic terms in a constraint. The quadratic terms are stored sequentially in qcsubi, qcsubj, and qcval.

Parameters
• k (int) – Which constraint. (input)

• qcsubi (int[]) – Row subscripts for quadratic constraint matrix. (output)

• qcsubj (int[]) – Column subscripts for quadratic constraint matrix. (output)

• qcval (float[]) – Quadratic constraint coefficient values. (output)

Return
• numqcnz (int) – Number of quadratic terms.

• qcsubi (int[]) – Row subscripts for quadratic constraint matrix.

• qcsubj (int[]) – Column subscripts for quadratic constraint matrix.

• qcval (float[]) – Quadratic constraint coefficient values.

Groups
getqobj(qosubi,qosubj,qoval) -> (numqonz)

getqobj() -> (numqonz,qosubi,qosubj,qoval)


Obtains the quadratic terms in the objective. The required quadratic terms are stored sequentially in qosubi, qosubj, and qoval.

Parameters
• qosubi (int[]) – Row subscripts for quadratic objective coefficients. (output)

• qosubj (int[]) – Column subscripts for quadratic objective coefficients. (output)

• qoval (float[]) – Quadratic objective coefficient values. (output)

Return
• numqonz (int) – Number of non-zero elements in the quadratic objective terms.

• qosubi (int[]) – Row subscripts for quadratic objective coefficients.

• qosubj (int[]) – Column subscripts for quadratic objective coefficients.

• qoval (float[]) – Quadratic objective coefficient values.

Groups
getqobjij(i,j) -> (qoij)


Obtains one coefficient $$q_{ij}^o$$ in the quadratic term of the objective.

Parameters
• i (int) – Row index of the coefficient. (input)

• j (int) – Column index of coefficient. (input)

Return

qoij (float) – The required coefficient.

Groups
getreducedcosts(whichsol,first,last,redcosts)

getreducedcosts(whichsol,first,last) -> (redcosts)


Computes the reduced costs for a slice of variables and returns them in the array redcosts i.e.

(15.2)$\mathtt{redcosts}[j-\mathtt{first}] = (s_l^x)_j-(s_u^x)_j, ~j=\mathtt{first},\ldots,\mathtt{last}-1$
Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – The index of the first variable in the sequence. (input)

• last (int) – The index of the last variable in the sequence plus 1. (input)

• redcosts (float[]) – The reduced costs for the required slice of variables. (output)

Return

redcosts (float[]) – The reduced costs for the required slice of variables.

Groups

Solution - dual

getskc(whichsol,skc)

getskc(whichsol) -> (skc)


Obtains the status keys for the constraints.

Parameters
Return

skc (mosek.stakey[]) – Status keys for the constraints.

Groups

Solution information

getskcslice(whichsol,first,last,skc)

getskcslice(whichsol,first,last) -> (skc)


Obtains the status keys for a slice of the constraints.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (output)

Return

skc (mosek.stakey[]) – Status keys for the constraints.

Groups

Solution information

getskn(whichsol,skn)

getskn(whichsol) -> (skn)


Obtains the status keys for the conic constraints.

Parameters
Return

skn (mosek.stakey[]) – Status keys for the conic constraints.

Groups

Solution information

getskx(whichsol,skx)

getskx(whichsol) -> (skx)


Obtains the status keys for the scalar variables.

Parameters
Return

skx (mosek.stakey[]) – Status keys for the variables.

Groups

Solution information

getskxslice(whichsol,first,last,skx)

getskxslice(whichsol,first,last) -> (skx)


Obtains the status keys for a slice of the scalar variables.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• skx (mosek.stakey[]) – Status keys for the variables. (output)

Return

skx (mosek.stakey[]) – Status keys for the variables.

Groups

Solution information

getslc(whichsol,slc)

getslc(whichsol) -> (slc)


Obtains the $$s_l^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (output)

Return

slc (float[]) – Dual variables corresponding to the lower bounds on the constraints.

Groups

Solution - dual

getslcslice(whichsol,first,last,slc)

getslcslice(whichsol,first,last) -> (slc)


Obtains a slice of the $$s_l^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (output)

Return

slc (float[]) – Dual variables corresponding to the lower bounds on the constraints.

Groups

Solution - dual

getslx(whichsol,slx)

getslx(whichsol) -> (slx)


Obtains the $$s_l^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (output)

Return

slx (float[]) – Dual variables corresponding to the lower bounds on the variables.

Groups

Solution - dual

getslxslice(whichsol,first,last,slx)

getslxslice(whichsol,first,last) -> (slx)


Obtains a slice of the $$s_l^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (output)

Return

slx (float[]) – Dual variables corresponding to the lower bounds on the variables.

Groups

Solution - dual

getsnx(whichsol,snx)

getsnx(whichsol) -> (snx)


Obtains the $$s_n^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (output)

Return

snx (float[]) – Dual variables corresponding to the conic constraints on the variables.

Groups

Solution - dual

getsnxslice(whichsol,first,last,snx)

getsnxslice(whichsol,first,last) -> (snx)


Obtains a slice of the $$s_n^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (output)

Return

snx (float[]) – Dual variables corresponding to the conic constraints on the variables.

Groups

Solution - dual

getsolsta(whichsol) -> (solutionsta)


Obtains the solution status.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return

solutionsta (mosek.solsta) – Solution status.

Groups

Solution information

getsolution(whichsol,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx) -> (problemsta,solutionsta)

getsolution(whichsol) ->
(problemsta,
solutionsta,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx)


Obtains the complete solution.

Consider the case of linear programming. The primal problem is given by

$\begin{split}\begin{array}{lccccl} \mbox{minimize} & & & c^T x+c^f & & \\ \mbox{subject to} & l^c & \leq & A x & \leq & u^c, \\ & l^x & \leq & x & \leq & u^x. \\ \end{array}\end{split}$

and the corresponding dual problem is

$\begin{split}\begin{array}{lccl} \mbox{maximize} & (l^c)^T s_l^c - (u^c)^T s_u^c & \\ & + (l^x)^T s_l^x - (u^x)^T s_u^x + c^f & \\ \mbox{subject to} & A^T y + s_l^x - s_u^x & = & c, \\ & -y + s_l^c - s_u^c & = & 0, \\ & s_l^c,s_u^c,s_l^x,s_u^x \geq 0. & & \\ \end{array}\end{split}$

A conic optimization problem has the same primal variables as in the linear case. Recall that the dual of a conic optimization problem is given by:

$\begin{split}\begin{array}{lccccc} \mbox{maximize} & (l^c)^T s_l^c - (u^c)^T s_u^c & & \\ & +(l^x)^T s_l^x - (u^x)^T s_u^x + c^f & & \\ \mbox{subject to} & A^T y + s_l^x - s_u^x + s_n^x & = & c, \\ & -y + s_l^c - s_u^c & = & 0, \\ & s_l^c,s_u^c,s_l^x,s_u^x & \geq & 0, \\ & s_n^x \in \K^* & & \\ \end{array}\end{split}$

The mapping between variables and arguments to the function is as follows:

• xx : Corresponds to variable $$x$$ (also denoted $$x^x$$).

• xc : Corresponds to $$x^c:=Ax$$.

• y : Corresponds to variable $$y$$.

• slc: Corresponds to variable $$s_l^c$$.

• suc: Corresponds to variable $$s_u^c$$.

• slx: Corresponds to variable $$s_l^x$$.

• sux: Corresponds to variable $$s_u^x$$.

• snx: Corresponds to variable $$s_n^x$$.

The meaning of the values returned by this function depend on the solution status returned in the argument solsta. The most important possible values of solsta are:

In order to retrieve the primal and dual values of semidefinite variables see Task.getbarxj and Task.getbarsj.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (output)

• skx (mosek.stakey[]) – Status keys for the variables. (output)

• skn (mosek.stakey[]) – Status keys for the conic constraints. (output)

• xc (float[]) – Primal constraint solution. (output)

• xx (float[]) – Primal variable solution. (output)

• y (float[]) – Vector of dual variables corresponding to the constraints. (output)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (output)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (output)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (output)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (output)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (output)

Return
• problemsta (mosek.prosta) – Problem status.

• solutionsta (mosek.solsta) – Solution status.

• skc (mosek.stakey[]) – Status keys for the constraints.

• skx (mosek.stakey[]) – Status keys for the variables.

• skn (mosek.stakey[]) – Status keys for the conic constraints.

• xc (float[]) – Primal constraint solution.

• xx (float[]) – Primal variable solution.

• y (float[]) – Vector of dual variables corresponding to the constraints.

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints.

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints.

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables.

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables.

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables.

Groups
getsolutioninfo(whichsol) ->
(pobj,
pviolcon,
pviolvar,
pviolbarvar,
pviolcone,
pviolitg,
dobj,
dviolcon,
dviolvar,
dviolbarvar,
dviolcone)


Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return
• pobj (float) – The primal objective value as computed by Task.getprimalobj.

• pviolcon (float) – Maximal primal violation of the solution associated with the $$x^c$$ variables where the violations are computed by Task.getpviolcon.

• pviolvar (float) – Maximal primal violation of the solution for the $$x$$ variables where the violations are computed by Task.getpviolvar.

• pviolbarvar (float) – Maximal primal violation of solution for the $$\barX$$ variables where the violations are computed by Task.getpviolbarvar.

• pviolcone (float) – Maximal primal violation of solution for the conic constraints where the violations are computed by Task.getpviolcones.

• pviolitg (float) – Maximal violation in the integer constraints. The violation for an integer variable $$x_j$$ is given by $$\min(x_j-\lfloor x_j \rfloor,\lceil x_j \rceil - x_j)$$. This number is always zero for the interior-point and basic solutions.

• dobj (float) – Dual objective value as computed by Task.getdualobj.

• dviolcon (float) – Maximal violation of the dual solution associated with the $$x^c$$ variable as computed by Task.getdviolcon.

• dviolvar (float) – Maximal violation of the dual solution associated with the $$x$$ variable as computed by Task.getdviolvar.

• dviolbarvar (float) – Maximal violation of the dual solution associated with the $$\barS$$ variable as computed by Task.getdviolbarvar.

• dviolcone (float) – Maximal violation of the dual solution associated with the dual conic constraints as computed by Task.getdviolcones.

Groups

Solution information

getsolutioninfonew(whichsol) ->
(pobj,
pviolcon,
pviolvar,
pviolbarvar,
pviolcone,
pviolacc,
pvioldjc,
pviolitg,
dobj,
dviolcon,
dviolvar,
dviolbarvar,
dviolcone,
dviolacc)


Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return
• pobj (float) – The primal objective value as computed by Task.getprimalobj.

• pviolcon (float) – Maximal primal violation of the solution associated with the $$x^c$$ variables where the violations are computed by Task.getpviolcon.

• pviolvar (float) – Maximal primal violation of the solution for the $$x$$ variables where the violations are computed by Task.getpviolvar.

• pviolbarvar (float) – Maximal primal violation of solution for the $$\barX$$ variables where the violations are computed by Task.getpviolbarvar.

• pviolcone (float) – Maximal primal violation of solution for the conic constraints where the violations are computed by Task.getpviolcones.

• pviolacc (float) – Maximal primal violation of solution for the affine conic constraints where the violations are computed by Task.getpviolacc.

• pvioldjc (float) – Maximal primal violation of solution for the disjunctive constraints where the violations are computed by Task.getpvioldjc.

• pviolitg (float) – Maximal violation in the integer constraints. The violation for an integer variable $$x_j$$ is given by $$\min(x_j-\lfloor x_j \rfloor,\lceil x_j \rceil - x_j)$$. This number is always zero for the interior-point and basic solutions.

• dobj (float) – Dual objective value as computed by Task.getdualobj.

• dviolcon (float) – Maximal violation of the dual solution associated with the $$x^c$$ variable as computed by Task.getdviolcon.

• dviolvar (float) – Maximal violation of the dual solution associated with the $$x$$ variable as computed by Task.getdviolvar.

• dviolbarvar (float) – Maximal violation of the dual solution associated with the $$\barS$$ variable as computed by Task.getdviolbarvar.

• dviolcone (float) – Maximal violation of the dual solution associated with the dual conic constraints as computed by Task.getdviolcones.

• dviolacc (float) – Maximal violation of the dual solution associated with the affine conic constraints as computed by Task.getdviolacc.

Groups

Solution information

getsolutionnew(whichsol,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx,
doty) -> (problemsta,solutionsta)

getsolutionnew(whichsol) ->
(problemsta,
solutionsta,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx,
doty)


Obtains the complete solution. See Task.getsolution for further information.

In order to retrieve the primal and dual values of semidefinite variables see Task.getbarxj and Task.getbarsj.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (output)

• skx (mosek.stakey[]) – Status keys for the variables. (output)

• skn (mosek.stakey[]) – Status keys for the conic constraints. (output)

• xc (float[]) – Primal constraint solution. (output)

• xx (float[]) – Primal variable solution. (output)

• y (float[]) – Vector of dual variables corresponding to the constraints. (output)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (output)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (output)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (output)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (output)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (output)

• doty (float[]) – Dual variables corresponding to affine conic constraints. (output)

Return
• problemsta (mosek.prosta) – Problem status.

• solutionsta (mosek.solsta) – Solution status.

• skc (mosek.stakey[]) – Status keys for the constraints.

• skx (mosek.stakey[]) – Status keys for the variables.

• skn (mosek.stakey[]) – Status keys for the conic constraints.

• xc (float[]) – Primal constraint solution.

• xx (float[]) – Primal variable solution.

• y (float[]) – Vector of dual variables corresponding to the constraints.

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints.

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints.

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables.

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables.

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables.

• doty (float[]) – Dual variables corresponding to affine conic constraints.

Groups
getsolutionslice(whichsol,solitem,first,last,values)

getsolutionslice(whichsol,solitem,first,last) -> (values)


Obtains a slice of one item from the solution. The format of the solution is exactly as in Task.getsolution. The parameter solitem determines which of the solution vectors should be returned.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• solitem (mosek.solitem) – Which part of the solution is required. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• values (float[]) – The values in the required sequence are stored sequentially in values. (output)

Return

values (float[]) – The values in the required sequence are stored sequentially in values.

Groups
getsparsesymmat(idx,subi,subj,valij)

getsparsesymmat(idx) -> (subi,subj,valij)


Get a single symmetric matrix from the matrix store.

Parameters
• idx (int) – Index of the matrix to retrieve. (input)

• subi (int[]) – Row subscripts of the matrix non-zero elements. (output)

• subj (int[]) – Column subscripts of the matrix non-zero elements. (output)

• valij (float[]) – Coefficients of the matrix non-zero elements. (output)

Return
• subi (int[]) – Row subscripts of the matrix non-zero elements.

• subj (int[]) – Column subscripts of the matrix non-zero elements.

• valij (float[]) – Coefficients of the matrix non-zero elements.

Groups
getstrparam(param) -> (len,parvalue)


Obtains the value of a string parameter.

Parameters

param (mosek.sparam) – Which parameter. (input)

Return
• len (int) – The length of the parameter value.

• parvalue (str) – Parameter value.

Groups
getstrparamlen(param) -> (len)


Obtains the length of a string parameter.

Parameters

param (mosek.sparam) – Which parameter. (input)

Return

len (int) – The length of the parameter value.

Groups
getsuc(whichsol,suc)

getsuc(whichsol) -> (suc)


Obtains the $$s_u^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (output)

Return

suc (float[]) – Dual variables corresponding to the upper bounds on the constraints.

Groups

Solution - dual

getsucslice(whichsol,first,last,suc)

getsucslice(whichsol,first,last) -> (suc)


Obtains a slice of the $$s_u^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (output)

Return

suc (float[]) – Dual variables corresponding to the upper bounds on the constraints.

Groups

Solution - dual

getsux(whichsol,sux)

getsux(whichsol) -> (sux)


Obtains the $$s_u^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (output)

Return

sux (float[]) – Dual variables corresponding to the upper bounds on the variables.

Groups

Solution - dual

getsuxslice(whichsol,first,last,sux)

getsuxslice(whichsol,first,last) -> (sux)


Obtains a slice of the $$s_u^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (output)

Return

sux (float[]) – Dual variables corresponding to the upper bounds on the variables.

Groups

Solution - dual

getsymmatinfo(idx) -> (dim,nz,mattype)


MOSEK maintains a vector denoted by $$E$$ of symmetric data matrices. This function makes it possible to obtain important information about a single matrix in $$E$$.

Parameters

idx (int) – Index of the matrix for which information is requested. (input)

Return
• dim (int) – Returns the dimension of the requested matrix.

• nz (int) – Returns the number of non-zeros in the requested matrix.

• mattype (mosek.symmattype) – Returns the type of the requested matrix.

Groups
gettaskname() -> (taskname)


Obtains the name assigned to the task.

Return

taskname (str) – Returns the task name.

Groups
gettasknamelen() -> (len)


Obtains the length the task name.

Return

len (int) – Returns the length of the task name.

Groups
getvarbound(i) -> (bk,bl,bu)


Obtains bound information for one variable.

Parameters

i (int) – Index of the variable for which the bound information should be obtained. (input)

Return
• bk (mosek.boundkey) – Bound keys.

• bl (float) – Values for lower bounds.

• bu (float) – Values for upper bounds.

Groups
getvarboundslice(first,last,bk,bl,bu)

getvarboundslice(first,last) -> (bk,bl,bu)


Obtains bounds information for a slice of the variables.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bk (mosek.boundkey[]) – Bound keys. (output)

• bl (float[]) – Values for lower bounds. (output)

• bu (float[]) – Values for upper bounds. (output)

Return
• bk (mosek.boundkey[]) – Bound keys.

• bl (float[]) – Values for lower bounds.

• bu (float[]) – Values for upper bounds.

Groups
getvarname(j) -> (name)


Obtains the name of a variable.

Parameters

j (int) – Index of a variable. (input)

Return

name (str) – Returns the required name.

Groups
getvarnameindex(somename) -> (asgn,index)


Checks whether the name somename has been assigned to any variable. If so, the index of the variable is reported.

Parameters

somename (str) – The name which should be checked. (input)

Return
• asgn (int) – Is non-zero if the name somename is assigned to a variable.

• index (int) – If the name somename is assigned to a variable, then index is the index of the variable.

Groups
getvarnamelen(i) -> (len)


Obtains the length of the name of a variable.

Parameters

i (int) – Index of a variable. (input)

Return

len (int) – Returns the length of the indicated name.

Groups
getvartype(j) -> (vartype)


Gets the variable type of one variable.

Parameters

j (int) – Index of the variable. (input)

Return

vartype (mosek.variabletype) – Variable type of the $$j$$-th variable.

Groups
getvartypelist(subj,vartype)

getvartypelist(subj) -> (vartype)


Obtains the variable type of one or more variables. Upon return vartype[k] is the variable type of variable subj[k].

Parameters
• subj (int[]) – A list of variable indexes. (input)

• vartype (mosek.variabletype[]) – The variables types corresponding to the variables specified by subj. (output)

Return

vartype (mosek.variabletype[]) – The variables types corresponding to the variables specified by subj.

Groups
getxc(whichsol,xc)

getxc(whichsol) -> (xc)


Obtains the $$x^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• xc (float[]) – Primal constraint solution. (output)

Return

xc (float[]) – Primal constraint solution.

Groups

Solution - primal

getxcslice(whichsol,first,last,xc)

getxcslice(whichsol,first,last) -> (xc)


Obtains a slice of the $$x^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• xc (float[]) – Primal constraint solution. (output)

Return

xc (float[]) – Primal constraint solution.

Groups

Solution - primal

getxx(whichsol,xx)

getxx(whichsol) -> (xx)


Obtains the $$x^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• xx (float[]) – Primal variable solution. (output)

Return

xx (float[]) – Primal variable solution.

Groups

Solution - primal

getxxslice(whichsol,first,last,xx)

getxxslice(whichsol,first,last) -> (xx)


Obtains a slice of the $$x^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• xx (float[]) – Primal variable solution. (output)

Return

xx (float[]) – Primal variable solution.

Groups

Solution - primal

gety(whichsol,y)

gety(whichsol) -> (y)


Obtains the $$y$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (output)

Return

y (float[]) – Vector of dual variables corresponding to the constraints.

Groups

Solution - dual

getyslice(whichsol,first,last,y)

getyslice(whichsol,first,last) -> (y)


Obtains a slice of the $$y$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (output)

Return

y (float[]) – Vector of dual variables corresponding to the constraints.

Groups

Solution - dual

infeasibilityreport(whichstream,whichsol)


TBD

Parameters
Groups

Infeasibility diagnostic

initbasissolve(basis)

initbasissolve() -> (basis)


Prepare a task for use with the Task.solvewithbasis function.

This function should be called

If the basis is singular i.e. not invertible, then the error rescode.err_basis_singular is reported.

Parameters

basis (int[]) – The array of basis indexes to use. The array is interpreted as follows: If $$\mathtt{basis}[i] \leq \idxend{numcon}$$, then $$x_{\mathtt{basis}[i]}^c$$ is in the basis at position $$i$$, otherwise $$x_{\mathtt{basis}[i]-\mathtt{numcon}}$$ is in the basis at position $$i$$. (output)

Return

basis (int[]) – The array of basis indexes to use. The array is interpreted as follows: If $$\mathtt{basis}[i] \leq \idxend{numcon}$$, then $$x_{\mathtt{basis}[i]}^c$$ is in the basis at position $$i$$, otherwise $$x_{\mathtt{basis}[i]-\mathtt{numcon}}$$ is in the basis at position $$i$$.

Groups

Solving systems with basis matrix

inputdata(maxnumcon,
maxnumvar,
c,
cfix,
aptrb,
aptre,
asub,
aval,
bkc,
blc,
buc,
bkx,
blx,
bux)


Input the linear part of an optimization problem.

The non-zeros of $$A$$ are inputted column-wise in the format described in Section Column or Row Ordered Sparse Matrix.

For an explained code example see Section Linear Optimization and Section Matrix Formats.

Parameters
• maxnumcon (int) – Number of preallocated constraints in the optimization task. (input)

• maxnumvar (int) – Number of preallocated variables in the optimization task. (input)

• c (float[]) – Linear terms of the objective as a dense vector. The length is the number of variables. (input)

• cfix (float) – Fixed term in the objective. (input)

• aptrb (int[]) – Row or column start pointers. (input)

• aptre (int[]) – Row or column end pointers. (input)

• asub (int[]) – Coefficient subscripts. (input)

• aval (float[]) – Coefficient values. (input)

• bkc (mosek.boundkey[]) – Bound keys for the constraints. (input)

• blc (float[]) – Lower bounds for the constraints. (input)

• buc (float[]) – Upper bounds for the constraints. (input)

• bkx (mosek.boundkey[]) – Bound keys for the variables. (input)

• blx (float[]) – Lower bounds for the variables. (input)

• bux (float[]) – Upper bounds for the variables. (input)

Groups
isdouparname(parname) -> (param)


Checks whether parname is a valid double parameter name.

Parameters

parname (str) – Parameter name. (input)

Return

param (mosek.dparam) – Returns the parameter corresponding to the name, if one exists.

Groups
isintparname(parname) -> (param)


Checks whether parname is a valid integer parameter name.

Parameters

parname (str) – Parameter name. (input)

Return

param (mosek.iparam) – Returns the parameter corresponding to the name, if one exists.

Groups
isstrparname(parname) -> (param)


Checks whether parname is a valid string parameter name.

Parameters

parname (str) – Parameter name. (input)

Return

param (mosek.sparam) – Returns the parameter corresponding to the name, if one exists.

Groups
linkfiletostream(whichstream,filename,append)


Directs all output from a task stream whichstream to a file filename.

Parameters
• whichstream (mosek.streamtype) – Index of the stream. (input)

• filename (str) – A valid file name. (input)

• append (int) – If this argument is 0 the output file will be overwritten, otherwise it will be appended to. (input)

Groups

Logging

onesolutionsummary(whichstream,whichsol)


Prints a short summary of a specified solution.

Parameters
Groups
optimize() -> (trmcode)


Calls the optimizer. Depending on the problem type and the selected optimizer this will call one of the optimizers in MOSEK. By default the interior point optimizer will be selected for continuous problems. The optimizer may be selected manually by setting the parameter iparam.optimizer.

Return

trmcode (mosek.rescode) – Is either rescode.ok or a termination response code.

Groups

Optimization

optimizermt(address,accesstoken) -> (trmcode)


Offload the optimization task to an instance of OptServer specified by addr, which should be a valid URL, for example http://server:port or https://server:port. The call will block until a result is available or the connection closes.

If the server requires authentication, the authentication token can be passed in the accesstoken argument.

If the server requires encryption, the keys can be passed using one of the solver parameters sparam.remote_tls_cert or sparam.remote_tls_cert_path.

Parameters
• address (str) – Address of the OptServer. (input)

• accesstoken (str) – Access token. (input)

Return

trmcode (mosek.rescode) – Is either rescode.ok or a termination response code.

Groups

Remote optimization

optimizersummary(whichstream)


Prints a short summary with optimizer statistics from last optimization.

Parameters

whichstream (mosek.streamtype) – Index of the stream. (input)

Groups

Logging

primalrepair(wlc,wuc,wlx,wux)


The function repairs a primal infeasible optimization problem by adjusting the bounds on the constraints and variables where the adjustment is computed as the minimal weighted sum of relaxations to the bounds on the constraints and variables. Observe the function only repairs the problem but does not solve it. If an optimal solution is required the problem should be optimized after the repair.

The function is applicable to linear and conic problems possibly with integer variables.

Observe that when computing the minimal weighted relaxation the termination tolerance specified by the parameters of the task is employed. For instance the parameter iparam.mio_mode can be used to make MOSEK ignore the integer constraints during the repair which usually leads to a much faster repair. However, the drawback is of course that the repaired problem may not have an integer feasible solution.

Note the function modifies the task in place. If this is not desired, then apply the function to a cloned task.

Parameters
• wlc (float[]) – $$(w_l^c)_i$$ is the weight associated with relaxing the lower bound on constraint $$i$$. If the weight is negative, then the lower bound is not relaxed. Moreover, if the argument is NULL, then all the weights are assumed to be $$1$$. (input)

• wuc (float[]) – $$(w_u^c)_i$$ is the weight associated with relaxing the upper bound on constraint $$i$$. If the weight is negative, then the upper bound is not relaxed. Moreover, if the argument is NULL, then all the weights are assumed to be $$1$$. (input)

• wlx (float[]) – $$(w_l^x)_j$$ is the weight associated with relaxing the lower bound on variable $$j$$. If the weight is negative, then the lower bound is not relaxed. Moreover, if the argument is NULL, then all the weights are assumed to be $$1$$. (input)

• wux (float[]) – $$(w_l^x)_i$$ is the weight associated with relaxing the upper bound on variable $$j$$. If the weight is negative, then the upper bound is not relaxed. Moreover, if the argument is NULL, then all the weights are assumed to be $$1$$. (input)

Groups

Infeasibility diagnostic

primalsensitivity(subi,
marki,
subj,
markj,
leftpricei,
rightpricei,
leftrangei,
rightrangei,
leftpricej,
rightpricej,
leftrangej,
rightrangej)

primalsensitivity(subi,marki,subj,markj) ->
(leftpricei,
rightpricei,
leftrangei,
rightrangei,
leftpricej,
rightpricej,
leftrangej,
rightrangej)


Calculates sensitivity information for bounds on variables and constraints. For details on sensitivity analysis, the definitions of shadow price and linearity interval and an example see Section Sensitivity Analysis.

The type of sensitivity analysis to be performed (basis or optimal partition) is controlled by the parameter iparam.sensitivity_type.

Parameters
• subi (int[]) – Indexes of constraints to analyze. (input)

• marki (mosek.mark[]) – The value of marki[i] indicates for which bound of constraint subi[i] sensitivity analysis is performed. If marki[i] = mark.up the upper bound of constraint subi[i] is analyzed, and if marki[i] = mark.lo the lower bound is analyzed. If subi[i] is an equality constraint, either mark.lo or mark.up can be used to select the constraint for sensitivity analysis. (input)

• subj (int[]) – Indexes of variables to analyze. (input)

• markj (mosek.mark[]) – The value of markj[j] indicates for which bound of variable subj[j] sensitivity analysis is performed. If markj[j] = mark.up the upper bound of variable subj[j] is analyzed, and if markj[j] = mark.lo the lower bound is analyzed. If subj[j] is a fixed variable, either mark.lo or mark.up can be used to select the bound for sensitivity analysis. (input)

• leftpricei (float[]) – leftpricei[i] is the left shadow price for the bound marki[i] of constraint subi[i]. (output)

• rightpricei (float[]) – rightpricei[i] is the right shadow price for the bound marki[i] of constraint subi[i]. (output)

• leftrangei (float[]) – leftrangei[i] is the left range $$\beta_1$$ for the bound marki[i] of constraint subi[i]. (output)

• rightrangei (float[]) – rightrangei[i] is the right range $$\beta_2$$ for the bound marki[i] of constraint subi[i]. (output)

• leftpricej (float[]) – leftpricej[j] is the left shadow price for the bound markj[j] of variable subj[j]. (output)

• rightpricej (float[]) – rightpricej[j] is the right shadow price for the bound markj[j] of variable subj[j]. (output)

• leftrangej (float[]) – leftrangej[j] is the left range $$\beta_1$$ for the bound markj[j] of variable subj[j]. (output)

• rightrangej (float[]) – rightrangej[j] is the right range $$\beta_2$$ for the bound markj[j] of variable subj[j]. (output)

Return
• leftpricei (float[]) – leftpricei[i] is the left shadow price for the bound marki[i] of constraint subi[i].

• rightpricei (float[]) – rightpricei[i] is the right shadow price for the bound marki[i] of constraint subi[i].

• leftrangei (float[]) – leftrangei[i] is the left range $$\beta_1$$ for the bound marki[i] of constraint subi[i].

• rightrangei (float[]) – rightrangei[i] is the right range $$\beta_2$$ for the bound marki[i] of constraint subi[i].

• leftpricej (float[]) – leftpricej[j] is the left shadow price for the bound markj[j] of variable subj[j].

• rightpricej (float[]) – rightpricej[j] is the right shadow price for the bound markj[j] of variable subj[j].

• leftrangej (float[]) – leftrangej[j] is the left range $$\beta_1$$ for the bound markj[j] of variable subj[j].

• rightrangej (float[]) – rightrangej[j] is the right range $$\beta_2$$ for the bound markj[j] of variable subj[j].

Groups

Sensitivity analysis

putacc(accidx,domidx,afeidxlist,b)


Puts an affine conic constraint. This method overwrites an existing affine conic constraint number accidx with new data specified in the same format as in Task.appendacc.

Parameters
• accidx (int) – Affine conic constraint index. (input)

• domidx (int) – Domain index. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

putaccb(accidx,b)


Updates an existing affine conic constraint number accidx by putting a new vector $$b$$.

Parameters
• accidx (int) – Affine conic constraint index. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

putaccbj(accidx,j,bj)


Sets one value $$b[j]$$ in the $$b$$ vector for the affine conic constraint number accidx.

Parameters
• accidx (int) – Affine conic constraint index. (input)

• j (int) – The index of an element in b to change. (input)

• bj (float) – The new value of $$b[j]$$. (input)

Groups

Problem data - affine conic constraints

putaccdoty(whichsol,accidx,doty)

putaccdoty(whichsol,accidx) -> (doty)


Puts the $$\dot{y}$$ vector for a solution (the dual values of an affine conic constraint).

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• accidx (int) – The index of the affine conic constraint. (input)

• doty (float[]) – The dual values for this affine conic constraint. The array should have length equal to the dimension of the constraint. (output)

Return

doty (float[]) – The dual values for this affine conic constraint. The array should have length equal to the dimension of the constraint.

Groups
putacclist(accidxs,domidxs,afeidxlist,b)


Puts affine conic constraints. This method overwrites existing affine conic constraints whose numbers are provided in the list accidxs with new data which is a concatenation of individual constraint descriptions in the same format as in Task.appendacc (see also Task.appendaccs).

Parameters
• accidxs (int[]) – Affine conic constraint indices. (input)

• domidxs (int[]) – Domain indices. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, can be NULL. (input)

Groups

Problem data - affine conic constraints

putaccname(accidx,name)


Sets the name of an affine conic constraint.

Parameters
• accidx (int) – Index of the affine conic constraint. (input)

• name (str) – The name of the affine conic constraint. (input)

Groups
putacol(j,subj,valj)


Change one column of the linear constraint matrix $$A$$. Resets all the elements in column $$j$$ to zero and then sets

$a_{\mathtt{subj}[k],\mathtt{j}} = \mathtt{valj}[k], \quad k=0,\ldots,\mathtt{nzj}-1.$
Parameters
• j (int) – Index of a column in $$A$$. (input)

• subj (int[]) – Row indexes of non-zero values in column $$j$$ of $$A$$. (input)

• valj (float[]) – New non-zero values of column $$j$$ in $$A$$. (input)

Groups

Problem data - linear part

putacollist(sub,ptrb,ptre,asub,aval)


Change a set of columns in the linear constraint matrix $$A$$ with data in sparse triplet format. The requested columns are set to zero and then updated with:

$\begin{split}\begin{array}{rl} \mathtt{for} & i=\idxbeg,\ldots,\idxend{num}\\ & a_{\mathtt{asub}[k],\mathtt{sub}[i]} = \mathtt{aval}[k],\quad k=\mathtt{ptrb}[i],\ldots,\mathtt{ptre}[i]-1. \end{array}\end{split}$
Parameters
• sub (int[]) – Indexes of columns that should be replaced, no duplicates. (input)

• ptrb (int[]) – Array of pointers to the first element in each column. (input)

• ptre (int[]) – Array of pointers to the last element plus one in each column. (input)

• asub (int[]) – Row indexes of new elements. (input)

• aval (float[]) – Coefficient values. (input)

Groups

Problem data - linear part

putacolslice(first,last,ptrb,ptre,asub,aval)


Change a slice of columns in the linear constraint matrix $$A$$ with data in sparse triplet format. The requested columns are set to zero and then updated with:

$\begin{split}\begin{array}{rl} \mathtt{for} & i=\mathtt{first},\ldots,\mathtt{last}-1\\ & a_{\mathtt{asub}[k],i} = \mathtt{aval}[k],\quad k=\mathtt{ptrb}[i-\mathtt{first}],\ldots,\mathtt{ptre}[i-\mathtt{first}]-1. \end{array}\end{split}$
Parameters
• first (int) – First column in the slice. (input)

• last (int) – Last column plus one in the slice. (input)

• ptrb (int[]) – Array of pointers to the first element in each column. (input)

• ptre (int[]) – Array of pointers to the last element plus one in each column. (input)

• asub (int[]) – Row indexes of new elements. (input)

• aval (float[]) – Coefficient values. (input)

Groups

Problem data - linear part

putafebarfblocktriplet(numtrip,afeidx,barvaridx,subk,subl,valkl)


Inputs the $$\barF$$ matrix data in block triplet form.

Parameters
• numtrip (int) – Number of elements in the block triplet form. (input)

• afeidx (int[]) – Constraint index. (input)

• barvaridx (int[]) – Symmetric matrix variable index. (input)

• subk (int[]) – Block row index. (input)

• subl (int[]) – Block column index. (input)

• valkl (float[]) – The numerical value associated with each block triplet. (input)

Groups
putafebarfentry(afeidx,barvaridx,termidx,termweight)


This function sets one entry $$\barF_{ij}$$ where $$i=\mathrm{afeidx}$$ is the row index in the store of affine expressions and $$j=\mathrm{barvaridx}$$ is the index of a symmetric variable. That is, the expression

$\langle \barF_{ij}, \barX_j\rangle$

will be added to the $$i$$-th affine expression.

The matrix $$\barF_{ij}$$ is specified as a weighted sum of symmetric matrices from the symmetric matrix storage $$E$$, so $$\barF_{ij}$$ is a symmetric matrix, precisely:

$\barF_{\mathrm{afeidx},\mathrm{barvaridx}} = \sum_{k} \mathrm{termweight}[k] \cdot E_{\mathrm{termidx}[k]}.$

By default all elements in $$\barF$$ are 0, so only non-zero elements need be added. Setting the same entry again will overwrite the earlier entry.

The symmetric matrices from $$E$$ are defined separately using the function Task.appendsparsesymmat.

Parameters
• afeidx (int) – Row index of $$\barF$$. (input)

• barvaridx (int) – Semidefinite variable index. (input)

• termidx (int[]) – Indices in $$E$$ of the matrices appearing in the weighted sum for the $$\barF$$ entry being specified. (input)

• termweight (float[]) – termweight[k] is the coefficient of the termidx[k]-th element of $$E$$ in the weighted sum the $$\barF$$ entry being specified. (input)

Groups
putafebarfentrylist(afeidx,
barvaridx,
numterm,
ptrterm,
termidx,
termweight)


This function sets a list of entries in $$\barF$$. Each entry should be described as in Task.putafebarfentry and all those descriptions should be combined (for example concatenated) in the input to this method. That means the $$k$$-th entry set will have row index afeidx[k], symmetric variable index barvaridx[k] and the description of this term consists of indices in $$E$$ and weights appearing in positions

$\mathrm{ptrterm}[k],\ldots,\mathrm{ptrterm}[k] + \mathrm{lenterm}[k] - 1$

in the corresponding arrays termidx and termweight. See Task.putafebarfentry for details.

Parameters
• afeidx (int[]) – Row indexes of $$\barF$$. (input)

• barvaridx (int[]) – Semidefinite variable indexes. (input)

• numterm (int[]) – The number of terms in the weighted sums that form each entry. (input)

• ptrterm (int[]) – The pointer to the beginning of the description of each entry. (input)

• termidx (int[]) – Concatenated lists of indices in $$E$$ of the matrices appearing in the weighted sums for the $$\barF$$ being specified. (input)

• termweight (float[]) – Concatenated lists of weights appearing in the weighted sums forming the $$\barF$$ elements being specified. (input)

Groups
putafebarfrow(afeidx,
barvaridx,
numterm,
ptrterm,
termidx,
termweight)


This function inputs one row in $$\barF$$. It first clears the row, i.e. sets $$\barF_{\mathrm{afeidx},*}=0$$ and then sets the new entries. Each entry should be described as in Task.putafebarfentry and all those descriptions should be combined (for example concatenated) in the input to this method. That means the $$k$$-th entry set will have row index afeidx, symmetric variable index barvaridx[k] and the description of this term consists of indices in $$E$$ and weights appearing in positions

$\mathrm{ptrterm}[k],\ldots,\mathrm{ptrterm}[k] + \mathrm{numterm}[k] - 1$

in the corresponding arrays termidx and termweight. See Task.putafebarfentry for details.

Parameters
• afeidx (int) – Row index of $$\barF$$. (input)

• barvaridx (int[]) – Semidefinite variable indexes. (input)

• numterm (int[]) – The number of terms in the weighted sums that form each entry. (input)

• ptrterm (int[]) – The pointer to the beginning of the description of each entry. (input)

• termidx (int[]) – Concatenated lists of indices in $$E$$ of the matrices appearing in the weighted sums for the $$\barF$$ entries in the row. (input)

• termweight (float[]) – Concatenated lists of weights appearing in the weighted sums forming the $$\barF$$ entries in the row. (input)

Groups
putafefcol(varidx,afeidx,val)


Change one column of the matrix $$F$$ of affine expressions. Resets all the elements in column varidx to zero and then sets

$F_{\mathtt{afeidx}[k],\mathtt{varidx}} = \mathtt{val}[k], \quad k=0,\ldots,\mathtt{numnz}-1.$
Parameters
• varidx (int) – Index of a column in $$F$$. (input)

• afeidx (int[]) – Row indexes of non-zero values in the column of $$F$$. (input)

• val (float[]) – New non-zero values in the column of $$F$$. (input)

Groups

Problem data - affine expressions

putafefentry(afeidx,varidx,value)


Replaces one entry in the affine expression store $$F$$, that is it sets:

$F_{\mathrm{afeidx}, \mathrm{varidx}} = \mathrm{value}.$
Parameters
• afeidx (int) – Row index in $$F$$. (input)

• varidx (int) – Column index in $$F$$. (input)

• value (float) – Value of $$F_{\mathrm{afeidx},\mathrm{varidx}}$$. (input)

Groups

Problem data - affine expressions

putafefentrylist(afeidx,varidx,val)


Replaces a number of entries in the affine expression store $$F$$, that is it sets:

$F_{\mathrm{afeidxs}[k], \mathrm{varidx}[k]} = \mathrm{val}[k]$

for all $$k$$.

Parameters
• afeidx (int[]) – Row indices in $$F$$. (input)

• varidx (int[]) – Column indices in $$F$$. (input)

• val (float[]) – Values of the entries in $$F$$. (input)

Groups

Problem data - affine expressions

putafefrow(afeidx,varidx,val)


Change one row of the matrix $$F$$ of affine expressions. Resets all the elements in row afeidx to zero and then sets

$F_{\mathtt{afeidx},\mathtt{varidx}[k]} = \mathtt{val}[k], \quad k=0,\ldots,\mathtt{numnz}-1.$
Parameters
• afeidx (int) – Index of a row in $$F$$. (input)

• varidx (int[]) – Column indexes of non-zero values in the row of $$F$$. (input)

• val (float[]) – New non-zero values in the row of $$F$$. (input)

Groups

Problem data - affine expressions

putafefrowlist(afeidx,numnzrow,ptrrow,varidx,val)


Clears and then changes a number of rows of the matrix $$F$$ of affine expressions. The $$k$$-th of the rows to be changed has index $$i = \mathrm{afeidx}[k]$$, contains $$\mathrm{numnzrow}[k]$$ nonzeros and its description as in Task.putafefrow starts in position $$\mathrm{ptrrow}[k]$$ of the arrays varidx and val. Formally, the row with index $$i$$ is cleared and then set as:

$F_{i,\mathrm{varidx}[\mathrm{ptrrow}[k]+j]} = \mathrm{val}[\mathrm{ptrrow}[k] + j], \quad j=0,\ldots,\mathrm{numnzrow}[k]-1.$
Parameters
• afeidx (int[]) – Indices of rows in $$F$$. (input)

• numnzrow (int[]) – Number of non-zeros in each of the modified rows of $$F$$. (input)

• ptrrow (int[]) – Pointer to the first nonzero in each row of $$F$$. (input)

• varidx (int[]) – Column indexes of non-zero values. (input)

• val (float[]) – New non-zero values in the rows of $$F$$. (input)

Groups

Problem data - affine expressions

putafeg(afeidx,g)


Change one element of the vector $$g$$ in affine expressions i.e.

$g_{\mathtt{afeidx}} = \mathtt{gi}.$
Parameters
• afeidx (int) – Index of an entry in $$g$$. (input)

• g (float) – New value for $$g_{\mathrm{afeidx}}$$. (input)

Groups

Problem data - affine expressions

putafeglist(afeidx,g)


Changes a list of elements of the vector $$g$$ in affine expressions i.e. for all $$k$$ it sets

$g_{\mathrm{afeidx}[k]} = \mathrm{glist}[k].$
Parameters
• afeidx (int[]) – Indices of entries in $$g$$. (input)

• g (float[]) – New values for $$g$$. (input)

Groups

Problem data - affine expressions

putafegslice(first,last,slice)


Modifies a slice in the vector $$g$$ of constant terms in affine expressions using the principle

$g_{\mathtt{j}} = \mathtt{slice[j-first]}, \quad j=\mathrm{first},..,\idxend{\mathrm{last}}$
Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• slice (float[]) – The slice of $$g$$ as a dense vector. The length is last-first. (input)

Groups

Problem data - affine expressions

putaij(i,j,aij)


Changes a coefficient in the linear coefficient matrix $$A$$ using the method

$a_{i,j} = \mathtt{aij}.$
Parameters
• i (int) – Constraint (row) index. (input)

• j (int) – Variable (column) index. (input)

• aij (float) – New coefficient for $$a_{i,j}$$. (input)

Groups

Problem data - linear part

putaijlist(subi,subj,valij)


Changes one or more coefficients in $$A$$ using the method

$a_{\mathtt{subi[k]},\mathtt{subj[k]}} = \mathtt{valij[k]}, \quad k=\idxbeg,\ldots,\idxend{num}.$

Duplicates are not allowed.

Parameters
• subi (int[]) – Constraint (row) indices. (input)

• subj (int[]) – Variable (column) indices. (input)

• valij (float[]) – New coefficient values for $$a_{i,j}$$. (input)

Groups

Problem data - linear part

putarow(i,subi,vali)


Change one row of the linear constraint matrix $$A$$. Resets all the elements in row $$i$$ to zero and then sets

$a_{\mathtt{i},\mathtt{subi}[k]} = \mathtt{vali}[k], \quad k=0,\ldots,\mathtt{nzi}-1.$
Parameters
• i (int) – Index of a row in $$A$$. (input)

• subi (int[]) – Column indexes of non-zero values in row $$i$$ of $$A$$. (input)

• vali (float[]) – New non-zero values of row $$i$$ in $$A$$. (input)

Groups

Problem data - linear part

putarowlist(sub,ptrb,ptre,asub,aval)


Change a set of rows in the linear constraint matrix $$A$$ with data in sparse triplet format. The requested rows are set to zero and then updated with:

$\begin{split}\begin{array}{rl} \mathtt{for} & i=\idxbeg,\ldots,\idxend{num} \\ & a_{\mathtt{sub}[i],\mathtt{asub}[k]} = \mathtt{aval}[k],\quad k=\mathtt{ptrb}[i],\ldots,\mathtt{ptre}[i]-1. \end{array}\end{split}$
Parameters
• sub (int[]) – Indexes of rows that should be replaced, no duplicates. (input)

• ptrb (int[]) – Array of pointers to the first element in each row. (input)

• ptre (int[]) – Array of pointers to the last element plus one in each row. (input)

• asub (int[]) – Column indexes of new elements. (input)

• aval (float[]) – Coefficient values. (input)

Groups

Problem data - linear part

putarowslice(first,last,ptrb,ptre,asub,aval)


Change a slice of rows in the linear constraint matrix $$A$$ with data in sparse triplet format. The requested rows are set to zero and then updated with:

$\begin{split}\begin{array}{rl} \mathtt{for} & i=\mathtt{first},\ldots,\mathtt{last}-1 \\ & a_{i,\mathtt{asub}[k]} = \mathtt{aval}[k],\quad k=\mathtt{ptrb}[i-\mathtt{first}],\ldots,\mathtt{ptre}[i-\mathtt{first}]-1. \end{array}\end{split}$
Parameters
• first (int) – First row in the slice. (input)

• last (int) – Last row plus one in the slice. (input)

• ptrb (int[]) – Array of pointers to the first element in each row. (input)

• ptre (int[]) – Array of pointers to the last element plus one in each row. (input)

• asub (int[]) – Column indexes of new elements. (input)

• aval (float[]) – Coefficient values. (input)

Groups

Problem data - linear part

putatruncatetol(tolzero)


Truncates (sets to zero) all elements in $$A$$ that satisfy

$|a_{i,j}| \leq \mathtt{tolzero}.$
Parameters

tolzero (float) – Truncation tolerance. (input)

Groups

Problem data - linear part

putbarablocktriplet(num,subi,subj,subk,subl,valijkl)


Inputs the $$\barA$$ matrix in block triplet form.

Parameters
• num (int) – Number of elements in the block triplet form. (input)

• subi (int[]) – Constraint index. (input)

• subj (int[]) – Symmetric matrix variable index. (input)

• subk (int[]) – Block row index. (input)

• subl (int[]) – Block column index. (input)

• valijkl (float[]) – The numerical value associated with each block triplet. (input)

Groups

Problem data - semidefinite

putbaraij(i,j,sub,weights)


This function sets one element in the $$\barA$$ matrix.

Each element in the $$\barA$$ matrix is a weighted sum of symmetric matrices from the symmetric matrix storage $$E$$, so $$\barA_{ij}$$ is a symmetric matrix. By default all elements in $$\barA$$ are 0, so only non-zero elements need be added. Setting the same element again will overwrite the earlier entry.

The symmetric matrices from $$E$$ are defined separately using the function Task.appendsparsesymmat.

Parameters
• i (int) – Row index of $$\barA$$. (input)

• j (int) – Column index of $$\barA$$. (input)

• sub (int[]) – Indices in $$E$$ of the matrices appearing in the weighted sum for $$\barA_{ij}$$. (input)

• weights (float[]) – weights[k] is the coefficient of the sub[k]-th element of $$E$$ in the weighted sum forming $$\barA_{ij}$$. (input)

Groups

Problem data - semidefinite

putbaraijlist(subi,
subj,
alphaptrb,
alphaptre,
matidx,
weights)


This function sets a list of elements in the $$\barA$$ matrix.

Each element in the $$\barA$$ matrix is a weighted sum of symmetric matrices from the symmetric matrix storage $$E$$, so $$\barA_{ij}$$ is a symmetric matrix. By default all elements in $$\barA$$ are 0, so only non-zero elements need be added. Setting the same element again will overwrite the earlier entry.

The symmetric matrices from $$E$$ are defined separately using the function Task.appendsparsesymmat.

Parameters
• subi (int[]) – Row index of $$\barA$$. (input)

• subj (int[]) – Column index of $$\barA$$. (input)

• alphaptrb (int[]) – Start entries for terms in the weighted sum that forms $$\barA_{ij}$$. (input)

• alphaptre (int[]) – End entries for terms in the weighted sum that forms $$\barA_{ij}$$. (input)

• matidx (int[]) – Indices in $$E$$ of the matrices appearing in the weighted sum for $$\barA_{ij}$$. (input)

• weights (float[]) – weights[k] is the coefficient of the sub[k]-th element of $$E$$ in the weighted sum forming $$\barA_{ij}$$. (input)

Groups

Problem data - semidefinite

putbararowlist(subi,
ptrb,
ptre,
subj,
nummat,
matidx,
weights)


This function replaces a list of rows in the $$\barA$$ matrix.

Parameters
• subi (int[]) – Row indexes of $$\barA$$. (input)

• ptrb (int[]) – Start of rows in $$\barA$$. (input)

• ptre (int[]) – End of rows in $$\barA$$. (input)

• subj (int[]) – Column index of $$\barA$$. (input)

• nummat (int[]) – Number of entries in weighted sum of matrixes. (input)

• matidx (int[]) – Matrix indexes for weighted sum of matrixes. (input)

• weights (float[]) – Weights for weighted sum of matrixes. (input)

Groups

Problem data - semidefinite

putbarcblocktriplet(num,subj,subk,subl,valjkl)


Inputs the $$\barC$$ matrix in block triplet form.

Parameters
• num (int) – Number of elements in the block triplet form. (input)

• subj (int[]) – Symmetric matrix variable index. (input)

• subk (int[]) – Block row index. (input)

• subl (int[]) – Block column index. (input)

• valjkl (float[]) – The numerical value associated with each block triplet. (input)

Groups

Problem data - semidefinite

putbarcj(j,sub,weights)


This function sets one entry in the $$\barC$$ vector.

Each element in the $$\barC$$ vector is a weighted sum of symmetric matrices from the symmetric matrix storage $$E$$, so $$\barC_{j}$$ is a symmetric matrix. By default all elements in $$\barC$$ are 0, so only non-zero elements need be added. Setting the same element again will overwrite the earlier entry.

The symmetric matrices from $$E$$ are defined separately using the function Task.appendsparsesymmat.

Parameters
• j (int) – Index of the element in $$\barC$$ that should be changed. (input)

• sub (int[]) – Indices in $$E$$ of matrices appearing in the weighted sum for $$\barC_j$$ (input)

• weights (float[]) – weights[k] is the coefficient of the sub[k]-th element of $$E$$ in the weighted sum forming $$\barC_j$$. (input)

Groups
putbarsj(whichsol,j,barsj)


Sets the dual solution for a semidefinite variable.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• j (int) – Index of the semidefinite variable. (input)

• barsj (float[]) – Value of $$\barS_j$$. Format as in Task.getbarsj. (input)

Groups

Solution - semidefinite

putbarvarname(j,name)


Sets the name of a semidefinite variable.

Parameters
• j (int) – Index of the variable. (input)

• name (str) – The variable name. (input)

Groups
putbarxj(whichsol,j,barxj)


Sets the primal solution for a semidefinite variable.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• j (int) – Index of the semidefinite variable. (input)

• barxj (float[]) – Value of $$\barX_j$$. Format as in Task.getbarxj. (input)

Groups

Solution - semidefinite

putcfix(cfix)


Replaces the fixed term in the objective by a new one.

Parameters

cfix (float) – Fixed term in the objective. (input)

Groups
putcj(j,cj)


Modifies one coefficient in the linear objective vector $$c$$, i.e.

$c_{\mathtt{j}} = \mathtt{cj}.$

If the absolute value exceeds dparam.data_tol_c_huge an error is generated. If the absolute value exceeds dparam.data_tol_cj_large, a warning is generated, but the coefficient is inputted as specified.

Parameters
• j (int) – Index of the variable for which $$c$$ should be changed. (input)

• cj (float) – New value of $$c_j$$. (input)

Groups
putclist(subj,val)


Modifies the coefficients in the linear term $$c$$ in the objective using the principle

$c_{\mathtt{subj[t]}} = \mathtt{val[t]}, \quad t=\idxbeg,\ldots,\idxend{num}.$

If a variable index is specified multiple times in subj only the last entry is used. Data checks are performed as in Task.putcj.

Parameters
• subj (int[]) – Indices of variables for which the coefficient in $$c$$ should be changed. (input)

• val (float[]) – New numerical values for coefficients in $$c$$ that should be modified. (input)

Groups
putconbound(i,bkc,blc,buc)


Changes the bounds for one constraint.

If the bound value specified is numerically larger than dparam.data_tol_bound_inf it is considered infinite and the bound key is changed accordingly. If a bound value is numerically larger than dparam.data_tol_bound_wrn, a warning will be displayed, but the bound is inputted as specified.

Parameters
• i (int) – Index of the constraint. (input)

• bkc (mosek.boundkey) – New bound key. (input)

• blc (float) – New lower bound. (input)

• buc (float) – New upper bound. (input)

Groups
putconboundlist(sub,bkc,blc,buc)


Changes the bounds for a list of constraints. If multiple bound changes are specified for a constraint, then only the last change takes effect. Data checks are performed as in Task.putconbound.

Parameters
• sub (int[]) – List of constraint indexes. (input)

• bkc (mosek.boundkey[]) – Bound keys for the constraints. (input)

• blc (float[]) – Lower bounds for the constraints. (input)

• buc (float[]) – Upper bounds for the constraints. (input)

Groups
putconboundlistconst(sub,bkc,blc,buc)


Changes the bounds for one or more constraints. Data checks are performed as in Task.putconbound.

Parameters
• sub (int[]) – List of constraint indexes. (input)

• bkc (mosek.boundkey) – New bound key for all constraints in the list. (input)

• blc (float) – New lower bound for all constraints in the list. (input)

• buc (float) – New upper bound for all constraints in the list. (input)

Groups
putconboundslice(first,last,bkc,blc,buc)


Changes the bounds for a slice of the constraints. Data checks are performed as in Task.putconbound.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bkc (mosek.boundkey[]) – Bound keys for the constraints. (input)

• blc (float[]) – Lower bounds for the constraints. (input)

• buc (float[]) – Upper bounds for the constraints. (input)

Groups
putconboundsliceconst(first,last,bkc,blc,buc)


Changes the bounds for a slice of the constraints. Data checks are performed as in Task.putconbound.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bkc (mosek.boundkey) – New bound key for all constraints in the slice. (input)

• blc (float) – New lower bound for all constraints in the slice. (input)

• buc (float) – New upper bound for all constraints in the slice. (input)

Groups
putcone(k,ct,conepar,submem)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters
• k (int) – Index of the cone. (input)

• ct (mosek.conetype) – Specifies the type of the cone. (input)

• conepar (float) – For the power cone it denotes the exponent alpha. For other cone types it is unused and can be set to 0. (input)

• submem (int[]) – Variable subscripts of the members in the cone. (input)

Groups

Problem data - cones (deprecated)

putconename(j,name)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Parameters
• j (int) – Index of the cone. (input)

• name (str) – The name of the cone. (input)

Groups
putconname(i,name)


Sets the name of a constraint.

Parameters
• i (int) – Index of the constraint. (input)

• name (str) – The name of the constraint. (input)

Groups
putconsolutioni(i,whichsol,sk,x,sl,su)


Sets the primal and dual solution information for a single constraint.

Parameters
• i (int) – Index of the constraint. (input)

• whichsol (mosek.soltype) – Selects a solution. (input)

• sk (mosek.stakey) – Status key of the constraint. (input)

• x (float) – Primal solution value of the constraint. (input)

• sl (float) – Solution value of the dual variable associated with the lower bound. (input)

• su (float) – Solution value of the dual variable associated with the upper bound. (input)

Groups
putcslice(first,last,slice)


Modifies a slice in the linear term $$c$$ in the objective using the principle

$c_{\mathtt{j}} = \mathtt{slice[j-first]}, \quad j=first,..,\idxend{last}$

Data checks are performed as in Task.putcj.

Parameters
• first (int) – First element in the slice of $$c$$. (input)

• last (int) – Last element plus 1 of the slice in $$c$$ to be changed. (input)

• slice (float[]) – New numerical values for coefficients in $$c$$ that should be modified. (input)

Groups
putdjc(djcidx,
domidxlist,
afeidxlist,
b,
termsizelist)


Inputs a disjunctive constraint. The constraint has the form

$T_1\ \mathrm{or}\ T_2\ \mathrm{or}\ \cdots\ \mathrm{or}\ T_{\mathrm{numterms}}$

For each $$i=1,\ldots\mathrm{numterms}$$ the $$i$$-th clause (term) $$T_i$$ has the form a sequence of affine expressions belongs to a product of domains, where the number of domains is $$\mathrm{termsizelist}[i]$$ and the number of affine expressions is equal to the sum of dimensions of all domains appearing in $$T_i$$.

All the domains and all the affine expressions appearing in the above description are arranged sequentially in the lists domidxlist and afeidxlist, respectively. In particular, the length of domidxlist must be equal to the sum of elements of termsizelist, and the length of afeidxlist must be equal to the sum of dimensions of all the domains appearing in domidxlist.

The elements of domidxlist are indexes of domains previously defined with one of the append...domain functions.

The elements of afeidxlist are indexes to the store of affine expressions, i.e. the $$k$$-th affine expression appearing in the disjunctive constraint is going to be

$F_{\mathrm{afeidxlist}[k],:}x + g_{\mathrm{afeidxlist}[k]}$

If an optional vector b of the same length as afeidxlist is specified then the $$k$$-th affine expression appearing in the disjunctive constraint will be taken as

$F_{\mathrm{afeidxlist}[k],:}x + g_{\mathrm{afeidxlist}[k]} - b_k$
Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• domidxlist (int[]) – List of domain indexes. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. (input)

• termsizelist (int[]) – List of term sizes. (input)

Groups

Problem data - disjunctive constraints

putdjcname(djcidx,name)


Sets the name of a disjunctive constraint.

Parameters
• djcidx (int) – Index of the disjunctive constraint. (input)

• name (str) – The name of the disjunctive constraint. (input)

Groups
putdjcslice(idxfirst,
idxlast,
domidxlist,
afeidxlist,
b,
termsizelist,
termsindjc)


Inputs a slice of disjunctive constraints.

The array termsindjc should have length $$\mathrm{idxlast}-\mathrm{idxfirst}$$ and contain the number of terms in consecutive constraints forming the slice.

The rest of the input consists of concatenated descriptions of individual constraints, where each constraint is described as in Task.putdjc.

Parameters
• idxfirst (int) – Index of the first disjunctive constraint in the slice. (input)

• idxlast (int) – Index of the last disjunctive constraint in the slice plus 1. (input)

• domidxlist (int[]) – List of domain indexes. (input)

• afeidxlist (int[]) – List of affine expression indexes. (input)

• b (float[]) – The vector of constant terms added to affine expressions. Optional, may be NULL. (input)

• termsizelist (int[]) – List of term sizes. (input)

• termsindjc (int[]) – Number of terms in each of the disjunctive constraints in the slice. (input)

Groups

Problem data - disjunctive constraints

putdomainname(domidx,name)


Sets the name of a domain.

Parameters
• domidx (int) – Index of the domain. (input)

• name (str) – The name of the domain. (input)

Groups
putdouparam(param,parvalue)


Sets the value of a double parameter.

Parameters
• param (mosek.dparam) – Which parameter. (input)

• parvalue (float) – Parameter value. (input)

Groups

Parameters

putintparam(param,parvalue)


Sets the value of an integer parameter.

Parameters
• param (mosek.iparam) – Which parameter. (input)

• parvalue (int) – Parameter value. (input)

Groups

Parameters

putmaxnumacc(maxnumacc)


Sets the number of preallocated affine conic constraints in the optimization task. When this number is reached MOSEK will automatically allocate more space. It is never mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Parameters

maxnumacc (int) – Number of preallocated affine conic constraints. (input)

Groups
putmaxnumafe(maxnumafe)


Sets the number of preallocated affine expressions in the optimization task. When this number is reached MOSEK will automatically allocate more space for affine expressions. It is never mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Parameters

maxnumafe (int) – Number of preallocated affine expressions. (input)

Groups
putmaxnumanz(maxnumanz)


Sets the number of preallocated non-zero entries in $$A$$.

MOSEK stores only the non-zero elements in the linear coefficient matrix $$A$$ and it cannot predict how much storage is required to store $$A$$. Using this function it is possible to specify the number of non-zeros to preallocate for storing $$A$$.

If the number of non-zeros in the problem is known, it is a good idea to set maxnumanz slightly larger than this number, otherwise a rough estimate can be used. In general, if $$A$$ is inputted in many small chunks, setting this value may speed up the data input phase.

It is not mandatory to call this function, since MOSEK will reallocate internal structures whenever it is necessary.

The function call has no effect if both maxnumcon and maxnumvar are zero.

Parameters

maxnumanz (int) – Number of preallocated non-zeros in $$A$$. (input)

Groups
putmaxnumbarvar(maxnumbarvar)


Sets the number of preallocated symmetric matrix variables in the optimization task. When this number of variables is reached MOSEK will automatically allocate more space for variables.

It is not mandatory to call this function. It only gives a hint about the amount of data to preallocate for efficiency reasons.

Please note that maxnumbarvar must be larger than the current number of symmetric matrix variables in the task.

Parameters

maxnumbarvar (int) – Number of preallocated symmetric matrix variables. (input)

Groups
putmaxnumcon(maxnumcon)


Sets the number of preallocated constraints in the optimization task. When this number of constraints is reached MOSEK will automatically allocate more space for constraints.

It is never mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Please note that maxnumcon must be larger than the current number of constraints in the task.

Parameters

maxnumcon (int) – Number of preallocated constraints in the optimization task. (input)

Groups
putmaxnumcone(maxnumcone)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Sets the number of preallocated conic constraints in the optimization task. When this number of conic constraints is reached MOSEK will automatically allocate more space for conic constraints.

It is not mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Please note that maxnumcon must be larger than the current number of conic constraints in the task.

Parameters

maxnumcone (int) – Number of preallocated conic constraints in the optimization task. (input)

Groups
putmaxnumdjc(maxnumdjc)


Sets the number of preallocated disjunctive constraints in the optimization task. When this number is reached MOSEK will automatically allocate more space. It is never mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Parameters

maxnumdjc (int) – Number of preallocated disjunctive constraints in the task. (input)

Groups
putmaxnumdomain(maxnumdomain)


Sets the number of preallocated domains in the optimization task. When this number is reached MOSEK will automatically allocate more space. It is never mandatory to call this function, since MOSEK will reallocate any internal structures whenever it is required.

Parameters

maxnumdomain (int) – Number of preallocated domains. (input)

Groups
putmaxnumqnz(maxnumqnz)


Sets the number of preallocated non-zero entries in quadratic terms.

MOSEK stores only the non-zero elements in $$Q$$. Therefore, MOSEK cannot predict how much storage is required to store $$Q$$. Using this function it is possible to specify the number non-zeros to preallocate for storing $$Q$$ (both objective and constraints).

It may be advantageous to reserve more non-zeros for $$Q$$ than actually needed since it may improve the internal efficiency of MOSEK, however, it is never worthwhile to specify more than the double of the anticipated number of non-zeros in $$Q$$.

It is not mandatory to call this function, since MOSEK will reallocate internal structures whenever it is necessary.

Parameters

maxnumqnz (int) – Number of non-zero elements preallocated in quadratic coefficient matrices. (input)

Groups
putmaxnumvar(maxnumvar)


Sets the number of preallocated variables in the optimization task. When this number of variables is reached MOSEK will automatically allocate more space for variables.

It is not mandatory to call this function. It only gives a hint about the amount of data to preallocate for efficiency reasons.

Please note that maxnumvar must be larger than the current number of variables in the task.

Parameters

maxnumvar (int) – Number of preallocated variables in the optimization task. (input)

Groups
putnadouparam(paramname,parvalue)


Sets the value of a named double parameter.

Parameters
• paramname (str) – Name of a parameter. (input)

• parvalue (float) – Parameter value. (input)

Groups

Parameters

putnaintparam(paramname,parvalue)


Sets the value of a named integer parameter.

Parameters
• paramname (str) – Name of a parameter. (input)

• parvalue (int) – Parameter value. (input)

Groups

Parameters

putnastrparam(paramname,parvalue)


Sets the value of a named string parameter.

Parameters
• paramname (str) – Name of a parameter. (input)

• parvalue (str) – Parameter value. (input)

Groups

Parameters

putobjname(objname)


Assigns a new name to the objective.

Parameters

objname (str) – Name of the objective. (input)

Groups
putobjsense(sense)


Sets the objective sense of the task.

Parameters

sense (mosek.objsense) – The objective sense of the task. The values objsense.maximize and objsense.minimize mean that the problem is maximized or minimized respectively. (input)

Groups
putoptserverhost(host)


Specify an OptServer URL for remote calls. The URL should contain protocol, host and port in the form http://server:port or https://server:port. If the URL is set using this function, all subsequent calls to any MOSEK function that involves synchronous optimization will be sent to the specified OptServer instead of being executed locally. Passing NULL deactivates this redirection.

Has the same effect as setting the parameter sparam.remote_optserver_host.

Parameters

host (str) – A URL specifying the optimization server to be used. (input)

Groups

Remote optimization

putparam(parname,parvalue)


Checks if parname is valid parameter name. If it is, the parameter is assigned the value specified by parvalue.

Parameters
• parname (str) – Parameter name. (input)

• parvalue (str) – Parameter value. (input)

Groups

Parameters

putqcon(qcsubk,qcsubi,qcsubj,qcval)


Replace all quadratic entries in the constraints. The list of constraints has the form

$l_k^c \leq \half \sum_{i=0}^{\idxend{numvar}} \sum_{j=0}^{\idxend{numvar}} q_{ij}^k x_i x_j + \sum_{j=0}^{\idxend{numvar}} a_{kj} x_j \leq u_k^c, ~\ k=0,\ldots,m-1.$

This function sets all the quadratic terms to zero and then performs the update:

$q_{\mathtt{qcsubi[t]},\mathtt{qcsubj[t]}}^{\mathtt{qcsubk[t]}} = q_{\mathtt{\mathtt{qcsubj[t]},qcsubi[t]}}^{\mathtt{qcsubk[t]}} = q_{\mathtt{\mathtt{qcsubj[t]},qcsubi[t]}}^{\mathtt{qcsubk[t]}} + \mathtt{qcval[t]},$

for $$t=\idxbeg,\ldots,\idxend{numqcnz}$$.

• For large problems it is essential for the efficiency that the function Task.putmaxnumqnz is employed to pre-allocate space.

• Only the lower triangular parts should be specified because the $$Q$$ matrices are symmetric. Specifying entries where $$i < j$$ will result in an error.

• Only non-zero elements should be specified.

• The order in which the non-zero elements are specified is insignificant.

• Duplicate elements are added together as shown above. Hence, it is usually not recommended to specify the same entry multiple times.

For a code example see Section Quadratic Optimization

Parameters
• qcsubk (int[]) – Constraint subscripts for quadratic coefficients. (input)

• qcsubi (int[]) – Row subscripts for quadratic constraint matrix. (input)

• qcsubj (int[]) – Column subscripts for quadratic constraint matrix. (input)

• qcval (float[]) – Quadratic constraint coefficient values. (input)

Groups

putqconk(k,qcsubi,qcsubj,qcval)


Replaces all the quadratic entries in one constraint. This function performs the same operations as Task.putqcon but only with respect to constraint number k and it does not modify the other constraints. See the description of Task.putqcon for definitions and important remarks.

Parameters
• k (int) – The constraint in which the new $$Q$$ elements are inserted. (input)

• qcsubi (int[]) – Row subscripts for quadratic constraint matrix. (input)

• qcsubj (int[]) – Column subscripts for quadratic constraint matrix. (input)

• qcval (float[]) – Quadratic constraint coefficient values. (input)

Groups

putqobj(qosubi,qosubj,qoval)


Replace all quadratic terms in the objective. If the objective has the form

$\half \sum_{i=0}^{\idxend{numvar}} \sum_{j=0}^{\idxend{numvar}} q_{ij}^o x_i x_j + \sum_{j=0}^{\idxend{numvar}} c_{j} x_j + c^f$

then this function sets all the quadratic terms to zero and then performs the update:

$q_{\mathtt{qosubi[t]},\mathtt{qosubj[t]}}^{o} = q_{\mathtt{\mathtt{qosubj[t]},qosubi[t]}}^{o} = q_{\mathtt{\mathtt{qosubj[t]},qosubi[t]}}^{o} + \mathtt{qoval[t]},$

for $$t=\idxbeg,\ldots,\idxend{numqonz}$$.

See the description of Task.putqcon for important remarks and example.

Parameters
• qosubi (int[]) – Row subscripts for quadratic objective coefficients. (input)

• qosubj (int[]) – Column subscripts for quadratic objective coefficients. (input)

• qoval (float[]) – Quadratic objective coefficient values. (input)

Groups
putqobjij(i,j,qoij)


Replaces one coefficient in the quadratic term in the objective. The function performs the assignment

$q_{ij}^o = q_{ji}^o = \mathtt{qoij}.$

Only the elements in the lower triangular part are accepted. Setting $$q_{ij}$$ with $$j>i$$ will cause an error.

Please note that replacing all quadratic elements one by one is more computationally expensive than replacing them all at once. Use Task.putqobj instead whenever possible.

Parameters
• i (int) – Row index for the coefficient to be replaced. (input)

• j (int) – Column index for the coefficient to be replaced. (input)

• qoij (float) – The new value for $$q_{ij}^o$$. (input)

Groups
putskc(whichsol,skc)


Sets the status keys for the constraints.

Parameters
Groups

Solution information

putskcslice(whichsol,first,last,skc)


Sets the status keys for a slice of the constraints.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (input)

Groups

Solution information

putskx(whichsol,skx)


Sets the status keys for the scalar variables.

Parameters
Groups

Solution information

putskxslice(whichsol,first,last,skx)


Sets the status keys for a slice of the variables.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• skx (mosek.stakey[]) – Status keys for the variables. (input)

Groups

Solution information

putslc(whichsol,slc)


Sets the $$s_l^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (input)

Groups

Solution - dual

putslcslice(whichsol,first,last,slc)


Sets a slice of the $$s_l^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (input)

Groups

Solution - dual

putslx(whichsol,slx)


Sets the $$s_l^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (input)

Groups

Solution - dual

putslxslice(whichsol,first,last,slx)


Sets a slice of the $$s_l^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (input)

Groups

Solution - dual

putsnx(whichsol,sux)


Sets the $$s_n^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (input)

Groups

Solution - dual

putsnxslice(whichsol,first,last,snx)


Sets a slice of the $$s_n^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (input)

Groups

Solution - dual

putsolution(whichsol,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx)


Inserts a solution into the task.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (input)

• skx (mosek.stakey[]) – Status keys for the variables. (input)

• skn (mosek.stakey[]) – Status keys for the conic constraints. (input)

• xc (float[]) – Primal constraint solution. (input)

• xx (float[]) – Primal variable solution. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (input)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (input)

Groups
putsolutionnew(whichsol,
skc,
skx,
skn,
xc,
xx,
y,
slc,
suc,
slx,
sux,
snx,
doty)


Inserts a solution into the task.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• skc (mosek.stakey[]) – Status keys for the constraints. (input)

• skx (mosek.stakey[]) – Status keys for the variables. (input)

• skn (mosek.stakey[]) – Status keys for the conic constraints. (input)

• xc (float[]) – Primal constraint solution. (input)

• xx (float[]) – Primal variable solution. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (input)

• slc (float[]) – Dual variables corresponding to the lower bounds on the constraints. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (input)

• slx (float[]) – Dual variables corresponding to the lower bounds on the variables. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (input)

• snx (float[]) – Dual variables corresponding to the conic constraints on the variables. (input)

• doty (float[]) – Dual variables corresponding to affine conic constraints. (input)

Groups
putsolutionyi(i,whichsol,y)


Inputs the dual variable of a solution.

Parameters
• i (int) – Index of the dual variable. (input)

• whichsol (mosek.soltype) – Selects a solution. (input)

• y (float) – Solution value of the dual variable. (input)

Groups
putstrparam(param,parvalue)


Sets the value of a string parameter.

Parameters
• param (mosek.sparam) – Which parameter. (input)

• parvalue (str) – Parameter value. (input)

Groups

Parameters

putsuc(whichsol,suc)


Sets the $$s_u^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (input)

Groups

Solution - dual

putsucslice(whichsol,first,last,suc)


Sets a slice of the $$s_u^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• suc (float[]) – Dual variables corresponding to the upper bounds on the constraints. (input)

Groups

Solution - dual

putsux(whichsol,sux)


Sets the $$s_u^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (input)

Groups

Solution - dual

putsuxslice(whichsol,first,last,sux)


Sets a slice of the $$s_u^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• sux (float[]) – Dual variables corresponding to the upper bounds on the variables. (input)

Groups

Solution - dual

puttaskname(taskname)


Assigns a new name to the task.

Parameters

taskname (str) – Name assigned to the task. (input)

Groups
putvarbound(j,bkx,blx,bux)


Changes the bounds for one variable.

If the bound value specified is numerically larger than dparam.data_tol_bound_inf it is considered infinite and the bound key is changed accordingly. If a bound value is numerically larger than dparam.data_tol_bound_wrn, a warning will be displayed, but the bound is inputted as specified.

Parameters
• j (int) – Index of the variable. (input)

• bkx (mosek.boundkey) – New bound key. (input)

• blx (float) – New lower bound. (input)

• bux (float) – New upper bound. (input)

Groups
putvarboundlist(sub,bkx,blx,bux)


Changes the bounds for one or more variables. If multiple bound changes are specified for a variable, then only the last change takes effect. Data checks are performed as in Task.putvarbound.

Parameters
• sub (int[]) – List of variable indexes. (input)

• bkx (mosek.boundkey[]) – Bound keys for the variables. (input)

• blx (float[]) – Lower bounds for the variables. (input)

• bux (float[]) – Upper bounds for the variables. (input)

Groups
putvarboundlistconst(sub,bkx,blx,bux)


Changes the bounds for one or more variables. Data checks are performed as in Task.putvarbound.

Parameters
• sub (int[]) – List of variable indexes. (input)

• bkx (mosek.boundkey) – New bound key for all variables in the list. (input)

• blx (float) – New lower bound for all variables in the list. (input)

• bux (float) – New upper bound for all variables in the list. (input)

Groups
putvarboundslice(first,last,bkx,blx,bux)


Changes the bounds for a slice of the variables. Data checks are performed as in Task.putvarbound.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bkx (mosek.boundkey[]) – Bound keys for the variables. (input)

• blx (float[]) – Lower bounds for the variables. (input)

• bux (float[]) – Upper bounds for the variables. (input)

Groups
putvarboundsliceconst(first,last,bkx,blx,bux)


Changes the bounds for a slice of the variables. Data checks are performed as in Task.putvarbound.

Parameters
• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• bkx (mosek.boundkey) – New bound key for all variables in the slice. (input)

• blx (float) – New lower bound for all variables in the slice. (input)

• bux (float) – New upper bound for all variables in the slice. (input)

Groups
putvarname(j,name)


Sets the name of a variable.

Parameters
• j (int) – Index of the variable. (input)

• name (str) – The variable name. (input)

Groups
putvarsolutionj(j,whichsol,sk,x,sl,su,sn)


Sets the primal and dual solution information for a single variable.

Parameters
• j (int) – Index of the variable. (input)

• whichsol (mosek.soltype) – Selects a solution. (input)

• sk (mosek.stakey) – Status key of the variable. (input)

• x (float) – Primal solution value of the variable. (input)

• sl (float) – Solution value of the dual variable associated with the lower bound. (input)

• su (float) – Solution value of the dual variable associated with the upper bound. (input)

• sn (float) – Solution value of the dual variable associated with the conic constraint. (input)

Groups
putvartype(j,vartype)


Sets the variable type of one variable.

Parameters
• j (int) – Index of the variable. (input)

• vartype (mosek.variabletype) – The new variable type. (input)

Groups

Problem data - variables

putvartypelist(subj,vartype)


Sets the variable type for one or more variables. If the same index is specified multiple times in subj only the last entry takes effect.

Parameters
• subj (int[]) – A list of variable indexes for which the variable type should be changed. (input)

• vartype (mosek.variabletype[]) – A list of variable types that should be assigned to the variables specified by subj. (input)

Groups

Problem data - variables

putxc(whichsol,xc)

putxc(whichsol) -> (xc)


Sets the $$x^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• xc (float[]) – Primal constraint solution. (output)

Return

xc (float[]) – Primal constraint solution.

Groups

Solution - primal

putxcslice(whichsol,first,last,xc)


Sets a slice of the $$x^c$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• xc (float[]) – Primal constraint solution. (input)

Groups

Solution - primal

putxx(whichsol,xx)


Sets the $$x^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• xx (float[]) – Primal variable solution. (input)

Groups

Solution - primal

putxxslice(whichsol,first,last,xx)


Sets a slice of the $$x^x$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• xx (float[]) – Primal variable solution. (input)

Groups

Solution - primal

puty(whichsol,y)


Sets the $$y$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (input)

Groups

Solution - primal

putyslice(whichsol,first,last,y)


Sets a slice of the $$y$$ vector for a solution.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• first (int) – First index in the sequence. (input)

• last (int) – Last index plus 1 in the sequence. (input)

• y (float[]) – Vector of dual variables corresponding to the constraints. (input)

Groups

Solution - dual

readbsolution(filename,compress)


Parameters
Groups

Input/Output

readdata(filename)


Reads an optimization problem and associated data from a file.

Parameters

filename (str) – A valid file name. (input)

Groups

Input/Output

readdataformat(filename,format,compress)


Reads an optimization problem and associated data from a file.

Parameters
Groups

Input/Output

readjsonstring(data)


Load task data from a JSON string, replacing any data that already exists in the task object. All problem data, parameters and other settings are resorted, but if the string contains solutions, the solution status after loading a file is set to unknown, even if it is optimal or otherwise well-defined.

Parameters

data (str) – Problem data in text format. (input)

Groups

Input/Output

readlpstring(data)


Parameters

data (str) – Problem data in text format. (input)

Groups

Input/Output

readopfstring(data)


Parameters

data (str) – Problem data in text format. (input)

Groups

Input/Output

readparamfile(filename)


Reads MOSEK parameters from a file. Data is read from the file filename if it is a nonempty string. Otherwise data is read from the file specified by sparam.param_read_file_name.

Parameters

filename (str) – A valid file name. (input)

Groups
readptfstring(data)


Load task data from a PTF string, replacing any data that already exists in the task object. All problem data, parameters and other settings are resorted, but if the string contains solutions, the solution status after loading a file is set to unknown, even if it is optimal or otherwise well-defined.

Parameters

data (str) – Problem data in text format. (input)

Groups

Input/Output

readsolution(whichsol,filename)


Reads a solution file and inserts it as a specified solution in the task. Data is read from the file filename if it is a nonempty string. Otherwise data is read from one of the files specified by sparam.bas_sol_file_name, sparam.itr_sol_file_name or sparam.int_sol_file_name depending on which solution is chosen.

Parameters
• whichsol (mosek.soltype) – Selects a solution. (input)

• filename (str) – A valid file name. (input)

Groups

Input/Output

readsolutionfile(filename)


Read solution file in format determined by the filename

Parameters

filename (str) – A valid file name. (input)

Groups

Input/Output

readsummary(whichstream)


Prints a short summary of last file that was read.

Parameters

whichstream (mosek.streamtype) – Index of the stream. (input)

Groups
readtask(filename)


Load task data from a file, replacing any data that already exists in the task object. All problem data, parameters and other settings are resorted, but if the file contains solutions, the solution status after loading a file is set to unknown, even if it was optimal or otherwise well-defined when the file was dumped.

See section The Task Format for a description of the Task format.

Parameters

filename (str) – A valid file name. (input)

Groups

Input/Output

removebarvars(subset)


The function removes a subset of the symmetric matrices from the optimization task. This implies that the remaining symmetric matrices are renumbered.

Parameters

subset (int[]) – Indexes of symmetric matrices which should be removed. (input)

Groups

Problem data - semidefinite

removecones(subset)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Removes a number of conic constraints from the problem. This implies that the remaining conic constraints are renumbered. In general, it is much more efficient to remove a cone with a high index than a low index.

Parameters

subset (int[]) – Indexes of cones which should be removed. (input)

Groups

Problem data - cones (deprecated)

removecons(subset)


The function removes a subset of the constraints from the optimization task. This implies that the remaining constraints are renumbered.

Parameters

subset (int[]) – Indexes of constraints which should be removed. (input)

Groups
removevars(subset)


The function removes a subset of the variables from the optimization task. This implies that the remaining variables are renumbered.

Parameters

subset (int[]) – Indexes of variables which should be removed. (input)

Groups
resizetask(maxnumcon,
maxnumvar,
maxnumcone,
maxnumanz,
maxnumqnz)


Sets the amount of preallocated space assigned for each type of data in an optimization task.

It is never mandatory to call this function, since it only gives a hint about the amount of data to preallocate for efficiency reasons.

Please note that the procedure is destructive in the sense that all existing data stored in the task is destroyed.

Parameters
• maxnumcon (int) – New maximum number of constraints. (input)

• maxnumvar (int) – New maximum number of variables. (input)

• maxnumcone (int) – New maximum number of cones. (input)

• maxnumanz (int) – New maximum number of non-zeros in $$A$$. (input)

• maxnumqnz (int) – New maximum number of non-zeros in all $$Q$$ matrices. (input)

Groups

sensitivityreport(whichstream)


Reads a sensitivity format file from a location given by sparam.sensitivity_file_name and writes the result to the stream whichstream. If sparam.sensitivity_res_file_name is set to a non-empty string, then the sensitivity report is also written to a file of this name.

Parameters

whichstream (mosek.streamtype) – Index of the stream. (input)

Groups

Sensitivity analysis

set_InfoCallback(callback)


Receive callbacks with solver status and information during optimization.

For example:

task.set_InfoCallback(lambda code,dinf,iinf,liinf: print("Called from: {0}".format(code)))

Parameters

callback (callbackfunc) – The callback function. (input)

set_Progress(callback)


For example:

task.set_Progress(lambda code: print("Called from: {0}".format(code)))

Parameters

callback (progresscallbackfunc) – The callback function. (input)

set_Stream(whichstream, callback)


Directs all output from a task stream to a callback function.

Parameters
setdefaults()


Resets all the parameters to their default values.

Groups

Parameters

solutiondef(whichsol) -> (isdef)


Checks whether a solution is defined.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Return

isdef (bool) – Is non-zero if the requested solution is defined.

Groups

Solution information

solutionsummary(whichstream)


Prints a short summary of the current solutions.

Parameters

whichstream (mosek.streamtype) – Index of the stream. (input)

Groups
solvewithbasis(transp,numnz,sub,val) -> (numnzout)


If a basic solution is available, then exactly $$numcon$$ basis variables are defined. These $$numcon$$ basis variables are denoted the basis. Associated with the basis is a basis matrix denoted $$B$$. This function solves either the linear equation system

(15.3)$B \barX = b$

or the system

(15.4)$B^T \barX = b$

for the unknowns $$\barX$$, with $$b$$ being a user-defined vector. In order to make sense of the solution $$\barX$$ it is important to know the ordering of the variables in the basis because the ordering specifies how $$B$$ is constructed. When calling Task.initbasissolve an ordering of the basis variables is obtained, which can be used to deduce how MOSEK has constructed $$B$$. Indeed if the $$k$$-th basis variable is variable $$x_j$$ it implies that

$B_{i,k} = A_{i,j}, ~i=\idxbeg,\ldots,\idxend{numcon}.$

Otherwise if the $$k$$-th basis variable is variable $$x_j^c$$ it implies that

$\begin{split}B_{i,k} = \left\{ \begin{array}{ll} -1, & i = j, \\ 0 , & i \neq j. \\ \end{array} \right.\end{split}$

The function Task.initbasissolve must be called before a call to this function. Please note that this function exploits the sparsity in the vector $$b$$ to speed up the computations.

Parameters
• transp (int) – If this argument is zero, then (15.3) is solved, if non-zero then (15.4) is solved. (input)

• numnz (int) – The number of non-zeros in $$b$$. (input)

• sub (int[]) – As input it contains the positions of non-zeros in $$b$$. As output it contains the positions of the non-zeros in $$\barX$$. It must have room for $$numcon$$ elements. (input/output)

• val (float[]) – As input it is the vector $$b$$ as a dense vector (although the positions of non-zeros are specified in sub it is required that $$\mathtt{val}[i] = 0$$ when $$b[i] = 0$$). As output val is the vector $$\barX$$ as a dense vector. It must have length $$numcon$$. (input/output)

Return

numnzout (int) – The number of non-zeros in $$\barX$$.

Groups

Solving systems with basis matrix

strtoconetype(str) -> (conetype)


NOTE: This interface to conic optimization is deprecated and will be removed in a future major release. Conic problems should be specified using the affine conic constraints interface (ACC), see Sec. 6.2 (From Linear to Conic Optimization) for details.

Obtains cone type code corresponding to a cone type string.

Parameters

str (str) – String corresponding to the cone type code conetype. (input)

Return

conetype (mosek.conetype) – The cone type corresponding to the string str.

Groups

Names

strtosk(str) -> (sk)


Obtains the status key corresponding to an abbreviation string.

Parameters

str (str) – A status key abbreviation string. (input)

Return

sk (mosek.stakey) – Status key corresponding to the string.

Groups

Names

toconic()


This function tries to reformulate a given Quadratically Constrained Quadratic Optimization problem (QCQO) as a Conic Quadratic Optimization problem (CQO). The first step of the reformulation is to convert the quadratic term of the objective function, if any, into a constraint. Then the following steps are repeated for each quadratic constraint:

• a conic constraint is added along with a suitable number of auxiliary variables and constraints;

• the original quadratic constraint is not removed, but all its coefficients are zeroed out.

Note that the reformulation preserves all the original variables.

The conversion is performed in-place, i.e. the task passed as argument is modified on exit. That also means that if the reformulation fails, i.e. the given QCQP is not representable as a CQO, then the task has an undefined state. In some cases, users may want to clone the task to ensure a clean copy is preserved.

Groups

updatesolutioninfo(whichsol)


Update the information items related to the solution.

Parameters

whichsol (mosek.soltype) – Selects a solution. (input)

Groups

Information items and statistics

writebsolution(filename,compress)


Write a binary dump of the task solution.

Parameters
Groups

Input/Output

writedata(filename)


Writes problem data associated with the optimization task to a file in one of the supported formats. See Section Supported File Formats for the complete list.

The data file format is determined by the file name extension. To write in compressed format append the extension .gz. E.g to write a gzip compressed MPS file use the extension mps.gz.

Please note that MPS, LP and OPF files require all variables to have unique names. If a task contains no names, it is possible to write the file with automatically generated anonymous names by setting the iparam.write_generic_names parameter to onoffkey.on.

Data is written to the file filename if it is a nonempty string. Otherwise data is written to the file specified by sparam.data_file_name.

Parameters

filename (str) – A valid file name. (input)

Groups

Input/Output

writedatastream(format,stream)


Writes problem data associated with the optimization task to a stream in one of the supported formats.

The stream should have a write method accepting a byte-like object, typically an instance of io.BytesIO will be suitable. Examples:

    task.writedatastream(dataformat.ptf, compresstype.none, sys.stdout.buffer)

writejsonsol(filename)

filename (str) – A valid file name. (input)