A scientific on-line database for efficient function approximation

Ivana Veljkovic, Paul E. Plassmann, Daniel C. Haworth

Research output: Chapter in Book/Report/Conference proceedingChapter

35 Scopus citations

Abstract

The use of a scientific, on-line database for function approximation is a technique that can significantly reduce the computational demands for problems requiring the frequent evaluation of computationally expensive functions. In this paper we present several new algorithms which significantly improve the performance of software implementations of this database approach. These algorithms are of two types-first, algorithms designed to improve the database retrieval rates; and, second, algorithms that seek to reduce the size of the database and subsequently reduce the cost of database queries. We have developed a software implementation of algorithms called DOLFA. We present experimental results which detail the performance of the DOLFA software for one representative combustion application. We observe significant improvements in cumulative time needed for database operations and memory requirements in a comparison of DOLFA to the function tabulation software system ISAT.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
EditorsVipin Kumar, Marina L. Gavrilova, Chih Jeng Kenneth Tan, Pierre L’Ecuyer, Chih Jeng Kenneth Tan
PublisherSpringer Verlag
Pages643-653
Number of pages11
ISBN (Print)3540401555, 9783540448396
DOIs
StatePublished - 2003

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2667
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'A scientific on-line database for efficient function approximation'. Together they form a unique fingerprint.

Cite this