A Parallel implementation of approximate semidefinite programming for MAXCUT with experimental results
We have implemented the approximate semidefinite programming for MAXCUT given by Klein and Lu (STOC'96). Our implementation for an IBM SP2 reveals the impractibility of this algorithm, basically because it hides the use of very large and very small floating-point numbers that can not be directl...
| Autores: | , |
|---|---|
| Formato: | informe técnico |
| Fecha de publicación: | 1997 |
| País: | España |
| Recursos: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/96752 |
| Acesso em linha: | https://hdl.handle.net/2117/96752 |
| Access Level: | acceso abierto |
| Palavra-chave: | Approximate semidefinite programming MAXCUT Àrees temàtiques de la UPC::Informàtica::Programació |
| Resumo: | We have implemented the approximate semidefinite programming for MAXCUT given by Klein and Lu (STOC'96). Our implementation for an IBM SP2 reveals the impractibility of this algorithm, basically because it hides the use of very large and very small floating-point numbers that can not be directly handled by the current computer hardware. On the other side, our experimental results show that the algorithm converges to a good approximation of the optimal, though oscillating. |
|---|