Towards the optimum by semidefinite and copositive programming
Semidefinite and copositive programming have attained an§important role in combinatorial optimization in the§last two decades.§There is a strong evidence that semidefinite and§copositive§approximation models are significantly stronger than§the purely§linear ones for many combinatorial problems. In some§cases the§copositive models give even the exact value of the§problem.§§§The first part of the book contains beside a survey of§standard results from linear algebra and conic§programming also a new§method to solve semidefinite programs, based on the§augmented§Lagrangian method. This method named the Boundary§point method§goes far beyond the reach of interior point methods§when the linear§constraints are nearly orthogonal.§§The second part demonstrates the application of§semidefinite and§copositive programming to the following NP-hard§problems from§combinatorial optimization: the bandwidth problem,§the quadratic§assignment problem, the min-cut problem and the§general graph§partitioning problem. The book also provides the§ideas how to extend the approach§to some other 0-1 problems, like the§ stability number problem and the balanced vertex§separator problem.