SDPs with upper bounds

Consider a standard form SDP with an added upper bound

(1)\[\begin{split}\begin{array}{ll} \mbox{minimize} & \mathbf{Tr}(BX) \\ \mbox{subject to} & \mathbf{Tr}(A_iX) + c_i = 0, \quad i = 1,\ldots, n \\ & 0 \preceq X \preceq I \end{array}\end{split}\]

with variable \(X \in \mathbf{S}^m\). The problem (1) can be reformulated by introducing a slack variable \(S\)

(2)\[\begin{split}\begin{array}{ll} \mbox{minimize} & \mathbf{Tr}(BX) \\ \mbox{subject to} & \mathbf{Tr}(A_iX) + c_i = 0, \quad i = 1,\ldots, n \\ & X + S = I \\ & X \succeq 0, \ S \succeq 0. \end{array}\end{split}\]

Documentation

A custom solver for SDPs with upper bounds is available as a Python module ubsdp.py. The module implements the following function:

ubsdp(c, A, B, pstart=None, dstart=None)

Solves the problem (2) using a custom KKT solver.

The input arguments are \(c \in \mathbf{R}^n\), a matrix \(A \in \mathbf{R}^{m^2 \times n}\), and a matrix \(B \in \mathbf{S}^m\). The columns of \(A\) are \(\mathbf{vec}(A_i)\) where \(A_i \in \mathbf{S}^m\) is the i’th data matrix.

Returns the solution \(X\).

Example

from cvxopt import matrix, normal, spdiag, misc, lapack
from ubsdp import ubsdp

m, n = 50, 50
A = normal(m**2, n)

# Z0 random positive definite with maximum e.v. less than 1.0.
Z0 = normal(m,m)
Z0 = Z0 * Z0.T
w = matrix(0.0, (m,1))
a = +Z0
lapack.syev(a, w, jobz = 'V')
wmax = max(w)
if wmax > 0.9:  w = (0.9/wmax) * w
Z0 = a * spdiag(w) * a.T

# c = -A'(Z0)
c = matrix(0.0, (n,1))
misc.sgemv(A, Z0, c, dims = {'l': 0, 'q': [], 's': [m]}, trans = 'T', alpha = -1.0)

# Z1 = I - Z0
Z1 = -Z0
Z1[::m+1] += 1.0

x0 = normal(n,1)
X0 = normal(m,m)
X0 = X0*X0.T
S0 = normal(m,m)
S0 = S0*S0.T
# B = A(x0) - X0 + S0
B = matrix(A*x0 - X0[:] + S0[:], (m,m))

X = ubsdp(c, A, B)