The MOSEK Python optimizer API manual Version 7.0 (Revision 141)

Optimizer API for Python - Documentation - Mosek Optimizer API for Python - Documentation - Mosek

25.11.2015 Views

130 CHAPTER 10. PROBLEM FORMULATION AND SOLUTIONS 10.3 Semidefinite optimization Semidefinite optimization is an extension of conic quadratic optimization (see Section 10.2) allowing positive semidefinite matrix variables to be used in addition to the usual scalar variables. A semidefinite optimization problem can be written as minimize subject to l c i ≤ n−1 ∑ ∑p−1 〈 〉 c j x j + Cj , X j + c f j=0 n−1 j=0 p−1 ∑ ∑ 〈 〉 a ij x j + Aij , X j j=0 j=0 ≤ u c i, i = 0, . . . , m − 1 lj x ≤ x j ≤ u x j , j = 0, . . . , n − 1 x ∈ C, X j ∈ S r + j , j = 0, . . . , p − 1 (10.10) where the problem has p symmetric positive semidefinite variables X j ∈ S r + j of dimension r j with symmetric coefficient matrices C j ∈ S rj and A i,j ∈ S rj . We use standard notation for the matrix inner product, i.e., for U, V ∈ R m×n we have 〈U, V 〉 := m−1 ∑ i=0 n−1 ∑ U ij V ij . With semidefinite optimization we can model a wide range of problems as demonstrated in [7]. j=0 10.3.1 Duality for semidefinite optimization The dual problem corresponding to the semidefinite optimization problem (10.10) is given by maximize subject to (l c ) T s c l − (u c ) T s c u + (l x ) T s x l − (u x ) T s x u + c f c − A T y + s x u − s x l = s x n, m∑ C j − y i A ij = S j , j = 0, . . . , p − 1 i=0 s c l − s c u = y, s c l , s c u, s x l , s x u ≥ 0, s x n ∈ C ∗ , S j ∈ S r + j , j = 0, . . . , p − 1 (10.11) where A ∈ R m×n , A ij = a ij , which is similar to the dual problem for conic quadratic optimization (see Section 10.7), except for the addition of dual constraints m∑ (C j − y i A ij ) ∈ S r + j . Note that the dual of the dual problem is identical to the original primal problem. i=0

10.3. SEMIDEFINITE OPTIMIZATION 131 10.3.2 Infeasibility for semidefinite optimization In case MOSEK finds a problem to be infeasible it reports a certificate of the infeasibility. This works exactly as for linear problems (see Section 10.1.2). 10.3.2.1 Primal infeasible problems If the problem (10.10) is infeasible, MOSEK will report a certificate of primal infeasibility: The dual solution reported is a certificate of infeasibility, and the primal solution is undefined. A certificate of primal infeasibility is a feasible solution to the problem maximize subject to (l c ) T s c l − (u c ) T s c u + (l x ) T s x l − (u x ) T s x u A T y + s x l − s x u + s x n = 0, m−1 ∑ y i A ij + S j = 0, j = 0, . . . , p − 1 i=0 − y + s c l − s c u = 0, s c l , s c u, s x l , s x u ≥ 0, s x n ∈ C ∗ , S j ∈ S r + j , j = 0, . . . , p − 1 such that the objective value is strictly positive. (10.12) 10.3.2.2 Dual infeasible problems If the problem (10.11) is infeasible, MOSEK will report a certificate of dual infeasibility: The primal solution reported is the certificate of infeasibility, and the dual solution is undefined. A certificate of dual infeasibility is a feasible solution to the problem where minimize n−1 ∑ ∑p−1 〈 〉 c j x j + Cj , X j j=0 j=0 p−1 subject to ˆlc i ≤ ∑ ∑ 〈 〉 a ij x j + Aij , X j j=1 ˆlx j=0 ≤ û c i, i = 0, . . . , m − 1 ≤ x ≤ û x , x ∈ C, X j ∈ S + r j , j = 0, . . . , p − 1 (10.13) and { 0 if l c ˆlc i = i > −∞, − ∞ otherwise, { 0 if l x ˆlx j = j > −∞, − ∞ otherwise, { and û c 0 if u c i := i < ∞, ∞ otherwise, { 0 if u and û x x j := j < ∞, ∞ otherwise,

10.3. SEMIDEFINITE OPTIMIZATION 131<br />

10.3.2 Infeasibility for semidefinite optimization<br />

In case <strong>MOSEK</strong> finds a problem to be infeasible it reports a certificate of the infeasibility. This works<br />

exactly as for linear problems (see Section 10.1.2).<br />

10.3.2.1 Primal infeasible problems<br />

If the problem (10.10) is infeasible, <strong>MOSEK</strong> will report a certificate of primal infeasibility: <strong>The</strong> dual<br />

solution reported is a certificate of infeasibility, and the primal solution is undefined.<br />

A certificate of primal infeasibility is a feasible solution to the problem<br />

maximize<br />

subject to<br />

(l c ) T s c l − (u c ) T s c u + (l x ) T s x l − (u x ) T s x u<br />

A T y + s x l − s x u + s x n = 0,<br />

m−1<br />

∑<br />

y i A ij + S j = 0, j = 0, . . . , p − 1<br />

i=0<br />

− y + s c l − s c u = 0,<br />

s c l , s c u, s x l , s x u ≥ 0,<br />

s x n ∈ C ∗ , S j ∈ S r + j<br />

, j = 0, . . . , p − 1<br />

such that the objective value is strictly positive.<br />

(10.12)<br />

10.3.2.2 Dual infeasible problems<br />

If the problem (10.11) is infeasible, <strong>MOSEK</strong> will report a certificate of dual infeasibility: <strong>The</strong> primal<br />

solution reported is the certificate of infeasibility, and the dual solution is undefined.<br />

A certificate of dual infeasibility is a feasible solution to the problem<br />

where<br />

minimize<br />

n−1<br />

∑ ∑p−1<br />

〈 〉<br />

c j x j + Cj , X j<br />

j=0<br />

j=0<br />

p−1<br />

subject to ˆlc i ≤ ∑ ∑ 〈 〉<br />

a ij x j + Aij , X j<br />

j=1<br />

ˆlx<br />

j=0<br />

≤ û c i, i = 0, . . . , m − 1<br />

≤ x ≤ û x ,<br />

x ∈ C, X j ∈ S + r j<br />

, j = 0, . . . , p − 1<br />

(10.13)<br />

and<br />

{ 0 if l<br />

c<br />

ˆlc i = i > −∞,<br />

− ∞ otherwise,<br />

{ 0 if l<br />

x<br />

ˆlx j = j > −∞,<br />

− ∞ otherwise,<br />

{<br />

and û c 0 if u<br />

c<br />

i :=<br />

i < ∞,<br />

∞ otherwise,<br />

{ 0 if u<br />

and û x x<br />

j :=<br />

j < ∞,<br />

∞ otherwise,

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!