Better relaxations of classical discrete optimization problems.
A mathematical program is an optimization problem expressed as an objective function of multiple variables subject to set of constraints. When the optimization problem has specific structure, the problem class usually has a special name. A linear program is the optimization of a linear objective function subject to linear constraints. An integer program is a linear program where some of the variables must take only integer values. A semidefinite program is a linear program where the variables are arranged in a matrix and for all feasible solutions, this matrix must be positive semidefinite. There are general-purpose solvers for each of these classes of mathematical program. There are usually many ways to express a problem as a correct, say, linear program. However, equivalent formulations can have significantly different practical tractability. In this poster, we present new formulations for two classic discrete optimization problems, maximum cut (max cut) and the graphical traveling salesman problem (GTSP), that are significantly stronger, and hence more computationally tractable, than any previous formulations of their class. Both partially answer longstanding open theoretical questions in polyhedral combinatorics.
- Research Organization:
- Sandia National Laboratories (SNL), Albuquerque, NM, and Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC04-94AL85000
- OSTI ID:
- 947360
- Report Number(s):
- SAND2008-5337C; TRN: US200909%%170
- Resource Relation:
- Conference: Proposed for presentation at the Applied Mathematics Principal Investigators Meeting held October 15-17, 2008 in Argonne, IL.
- Country of Publication:
- United States
- Language:
- English
Similar Records
A mathematical programming approach to stochastic and dynamic optimization problems
TAS: 89 0227: TAS Recovery Act - Optimization and Control of Electric Power Systems: ARRA