TY - GEN
T1 - Development of distributed computing framework for parallel multi-disciplinary optimization
AU - Chan, Yu
AU - Nan, Liu
AU - Pochiraju, Kishore
AU - Manoochehri, Souran
AU - Ko, K. H.
PY - 2006
Y1 - 2006
N2 - This paper discusses the detailed design and development of a web based parallel multi-disciplinary optimization (PMDO) framework in the distributed computing environment. This system consists of the HTTP server, the XML parser and the communication module based on TCP/IP, and is built around a computational kernel called the ACES kernel, which provides powerful computation and evaluation capabilities as well as optimization routines specifically designed for engineering purposes. We formulate and subdivide an optimization problem into several sub-problems such that one node is designated as a master which solves the overall problem and the others are distributed to client nodes each of which handles each independent sub-optimization problem. In the iteration to solve the overall optimization problem, the master forwards necessary data to each client for updated values which are the solutions of sub-optimization problems and collects the results from all the client nodes. The master continues the iteration until the optimum is reached. All optimization problems are represented in XML form and provided as input to the master and each client node.
AB - This paper discusses the detailed design and development of a web based parallel multi-disciplinary optimization (PMDO) framework in the distributed computing environment. This system consists of the HTTP server, the XML parser and the communication module based on TCP/IP, and is built around a computational kernel called the ACES kernel, which provides powerful computation and evaluation capabilities as well as optimization routines specifically designed for engineering purposes. We formulate and subdivide an optimization problem into several sub-problems such that one node is designated as a master which solves the overall problem and the others are distributed to client nodes each of which handles each independent sub-optimization problem. In the iteration to solve the overall optimization problem, the master forwards necessary data to each client for updated values which are the solutions of sub-optimization problems and collects the results from all the client nodes. The master continues the iteration until the optimum is reached. All optimization problems are represented in XML form and provided as input to the master and each client node.
UR - http://www.scopus.com/inward/record.url?scp=33751344619&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=33751344619&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:33751344619
SN - 079183784X
SN - 9780791837849
T3 - Proceedings of the ASME Design Engineering Technical Conference
BT - Proceedings of 2006 ASME International Design Engineering Technical Conferences and Computers and Information In Engineering Conference, DETC2006
T2 - 2006 ASME International Design Engineering Technical Conferences and Computers and Information In Engineering Conference, DETC2006
Y2 - 10 September 2006 through 13 September 2006
ER -