The Layered Software Infrastructure for Solving Large-Scale Optimization Problems on the Grid (pp.129-144)
Authors: (A. Afanasiev, Yu. Evtushenko, M. Posypkin, Institution of Russian Academy of Sciences Institute for Systems Analysis, and others)
Abstract: The paper presents a hierarchical software infrastructure for solving large scale
optimization problems on the Grid. The proposed toolset support exact and heuristic search
strategies and runs on distributed systems consisting of different nodes ranging from PCs to
large publicly available supercomputers. It efficiently copes with difficulties arising in such
systems: the software diversity, unreliability of nodes and different ways of job submission.
The distinctive feature of our approach is the use of different communication packages on
different levels: on the top level we use ICE middleware coupled with TCP/IP sockets and
within a single computing element either MPI or POSIX Threads libraries are used. Such
approach imposes minimal requirements on the computing element software and efficiently
utilizes the communication facilities of each node by using native communication mechanism.
Developed infrastructure has been applied to molecular conformation problem that plays an
important role in computational chemistry. New results were obtained demonstrating that
general purposed optimization algorithm can efficiently cope with hard optimization problems
providing the sufficient computational resources are employed.