We consider the problem of optimizing a linear function of a symmetric matrix X, subject to linear constraints on the matrix and the additional condition that X be positive semi-definite; we call this problem the semi-definite programming problem (SDP).
First we develop a duality theory for SDP which is structurally similar to the duality theory in linear programming and is a straightforward generalization of the latter. We also show that various eigenvalue optimization problems may be formulated as semi-definite programs.
Second, we develop two primal-dual interior-point methods for solving (SDP) which again is direct generalization of interior methods for linear programming. Furthermore, we argue that just about any linear programming interior point method may be modified in a mechanical way to yield a method for the SDP problems. The proofs of convergence or polynomial-time solvability may also be extended to the SDP problem.
Third, we study applications of SDP in combinatorial optimization problems. In particular, we apply our interior point methods to obtain good bounds for the maximum clique problem in perfect graphs, graph partitioning (and in particular graph bisection) problems, maximum induced multipartite subgraph problem, and the general 0-1 integer programming problems.
Index Terms
- Combinatorial optimization with interior point methods and semi-definite matrices
Recommendations
On handling cutting planes in interior-point methods for solving semi-definite relaxations of binary quadratic optimization problems
We describe an improved technique for handling large numbers of cutting planes when using an interior-point method for the solution of linear and semi-definite programming relaxations of combinatorial optimization problems. The approach combines an ...
On Interior-Point Warmstarts for Linear and Combinatorial Optimization
Despite the many advantages of interior-point algorithms over active-set methods for linear optimization, one of the remaining practical challenges is their current limitation to efficiently solve series of related problems by an effective warmstarting ...
Interior Point Methods in Semidefinite Programming with Applications to Combinatorial Optimization
This paper studies the semidefinite programming SDP problem, i.e., the optimization problem of a linear function of a symmetric matrix subject to linear equality constraints and the additional condition that the matrix be positive semidefinite. First the ...