Macro tree transformations of linear size increase achieve cost-optimal parallelism

Akimasa Morihata

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

This paper studies parallel evaluation of tree transformations, in particular accumulative ones. Accumulation is a ubiquitous programming pattern. However, since accumulation usually imposes restrictions on evaluation orders, accumulative tree transformations appear to be unsuitable for parallel evaluation. We propose a parallel evaluation method for a large class of tree-to-tree recursive functions, which may contain accumulations, higher-order terms, and function compositions. Our parallel evaluation method achieves optimal parallel speedup if the transformation is of linear size increase, namely, the size of each output is linearly bounded by the size of the corresponding input. Our result is based on the theory of macro tree transducers and that of parallel tree contractions. The main contribution is to reveal a good collaboration between them.

Original languageEnglish
Title of host publicationProgramming Languages and Systems - 9th Asian Symposium, APLAS 2011, Proceedings
Pages204-219
Number of pages16
DOIs
Publication statusPublished - 2011 Dec 26
Event9th Asian Symposium on Programming Languages and Systems, APLAS 2011 - Kenting, Taiwan, Province of China
Duration: 2011 Dec 52011 Dec 7

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7078 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other9th Asian Symposium on Programming Languages and Systems, APLAS 2011
CountryTaiwan, Province of China
CityKenting
Period11/12/511/12/7

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Macro tree transformations of linear size increase achieve cost-optimal parallelism'. Together they form a unique fingerprint.

Cite this