Symbolic Optimization and Analysis of Numerical Dynamics Solutions

Abstract

A symbolic interpreter for the optimization and analysis of numerical dynamics solutions is presented. The interpreter is intended to address the order of magnitude performance disconnect between the numerical environments which support most general purpose dynamics implementations and the speed of heavily customized domain specific tools and/or symbolic implementations. Unlike other symbolic environments, the objective of this implementation is to be integrated and interchangeable within procedural numerical source code thereby enabling an auto-code feature within any software with only minor cosmetic modifications. Specifically, the interpreter generates efficient simulations via symbolic optimizations which include removal of trivial and identity operations, removal of redundant calculations, identification of constants and preprocessing steps, and culling of extraneous computations. The construction of the directed graph of symbolic computations also presents interesting opportunities for complexity analysis, optimal hybrid methods, and tailored fine grain parallel implementations.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2013
Accession Number
ADA583728

Entities

People

  • Amy Critchley
  • James Critchley
  • Paramsothy Jayakumar

Organizations

  • United States Army Tank Automotive Research, Development and Engineering Center

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Computations
  • Computer Programming
  • Computer Programs
  • Computers
  • Copyrights
  • Dynamics
  • Engineering
  • Equations
  • Equations Of Motion
  • Identification
  • Inversion
  • Language
  • Parallel Computing
  • Parallel Processing
  • Preprocessing
  • Simulations

Readers

  • Applied Combinatorial Optimization and Logic Circuit Design.
  • Computational Linguistics
  • Computational Modeling and Simulation