Invited Session Fri.3.H 1058

Friday, 15:15 - 16:45 h, Room: H 1058

Cluster 10: Implementations & software [...]

Parallel optimization software

 

Chair: Jeff Linderoth

 

 

Friday, 15:15 - 15:40 h, Room: H 1058, Talk 1

Katsuki Fujisawa
High-performance general solver for extremly large-scale semidefinite programming problems

Coauthors: Toshio Endo, Satoshi Matsuoka, Hitoshi Sato, Makoto Yamashita

 

Abstract:
Semidefinite Program (SDP) is one of the most important problems in
current research areas in optimization problems. It covers a wide range of applications such as combinatorial optimization, control theory, economics, quantum chemistry, sensor network location, data mining, etc.
Solving extremely large-scale SDPs has a significant importance for the current and future applications of SDPs. In 1995, Fujisawa et al. started the SDPA Project aimed for solving large-scale SDPs with numerical stability and accuracy. It is one of pioneers' code to solve general SDPs. The SDPARA is a parallel version of the SDPA on multiple processors and distributed memory, which replaces major bottleneck components of the SDPA by their parallel implementation. In particular, it has been successfully applied on quantum chemistry and combinatorial optimization, the SDPARA on a large-scale super computer called TSUBAME 2.0 in Tokyo Institute of Technology has succeeded to solve the largest SDP which has over one million constraints with high accuracy and make a new world record.

 

 

Friday, 15:45 - 16:10 h, Room: H 1058, Talk 2

Yuji Shinano
ParaSCIP and FiberSCIP - Parallel extensions of SCIP

Coauthors: Tobias Achterberg, Timo Berthold, Stefan Heinz, Thorsten Koch, Stefan Vigerske, Michael Winkler

 

Abstract:
SCIP is a powerful Mixed Integer Linear and Non-Linear Programming (MILP/MINLP) solver. We will present the implementation of two parallel extensions of SCIP. One is ParaSCIP, which is intended to run on a large scale distributed memory computing environment and the other is FiberSCIP, intended to run in shared memory computing environments. ParaSCIP has successfully been run on the HLRN II supercomputer utilizing up to 7,168 cores to solve a single difficult MILP. It has also been tested on a Fujitsu PRIMERGY RX200S5 using up to 512 cores. Even though ParaSCIP and FibreSCIP have different capabilities, they are realized using a single software: the Ubiquity Generator (UG) framework. The latest computational results using the both ParaSCIP and FiberSCIP will be presented.

 

 

Friday, 16:15 - 16:40 h, Room: H 1058, Talk 3

Cynthia Phillips
PICO's new hierarchical branch-and-bound system for massively parallel IP

Coauthors: Jonathan Eckstein, Ojas Parekh, John Siirola, Jean-Paul Watson

 

Abstract:
We will discuss the design, implementation, and large-scale parallel computational results for a new capability in the PICO (Parallel Integer and Combinatorial Optimizer) massively-parallel mixed-integer programming solver. We leverage the basic PICO ramp up system for automatic integer program decomposition and carefully manage runtime conditions to effectively run arbitrary black-box IP solvers on massively parallel systems. Our computational results use Sandia National Laboratories' "Red Sky'' system, which has more than 20,000 cores.

 

  Do you need Missouri Payday Loans as soon as possible? But at the same time, it acts only with sexual arousal. Cheap Viagra has a number of advantages in comparison with injections in the sexual organ or other procedures aimed at treatment of impotency.