Cost-based Partitioning for Distributed and Parallel Simulation of Decomposable Multiscale Constructive Models

  • Authors:

  • Sunwoo Park;C. Anthony Hunt;Bernard P. Zeigler

  • Affiliations:

  • BioSystems Group, Department of Biopharmaceutical SciencesUniversity of California, San Francisco 513 Parnassus Ave, San Francisco CA94143-0446, USA;BioSystems Group, Department of Biopharmaceutical Sciences,and Joint Graduate Group in Bioengineering University of California, Berkeleyand San Francisco 513 Parnassus Ave, San Francisco CA 94143- ...;Department of Electrical and Computer Engineering Universityof Arizona, 1230 E. Speedway Blvd Tucson, AZ 85721, USA

  • Venue:
  • Simulation
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a concise, generic, and configurable partitioning approach for decomposable, modular, and multiscale (or hierarchical) constructive models. A generic model partitioning (GMP) algorithm decomposes a given multiscale model to a set of partition blocks based on a cost modeling and analysis method in polynomial time. It minimizes model decompositions and constructs monotonically improved partitioning outcomes during the partitioning process. The cost modeling and analysis method enables translating subjective, domain-specific, and heterogeneous resource information to objective, domain-independent, and homogeneous cost information. By translating models to a homogeneous cost space and describing partitioning logics over the space, the proposed algorithm utilizes domain-specific knowledge to produce the best partitioning results without any modification of its programming logics. As a consequence of its clean separation between domain-specific partitioning requirements and goals, and generic partitioning logic, the proposed algorithm can be applied to a variety of partitioning problems in large-scale systems biology research utilizing distributed and parallel simulation. It is expected that the algorithm improves overall performance and efficiency of in silico experimentation of complex multiscale biological system models.