On isoperimetric profiles and computational complexity

Pavel Hrubeš, Amir Yehudayoff

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The isoperimetric profile of a graph is a function that measures, for an integer k, the size of the smallest edge boundary over all sets of vertices of size k. We observe a connection between isoperimetric profiles and computational complexity. We illustrate this connection by an example from communication complexity, but our main result is in algebraic complexity. We prove a sharp super-polynomial separation between monotone arithmetic circuits and monotone arithmetic branching programs. This shows that the classical simulation of arithmetic circuits by arithmetic branching programs by Valiant, Skyum, Berkowitz, and Rackoff (1983) cannot be improved, as long as it preserves monotonicity. A key ingredient in the proof is an accurate analysis of the isoperimetric profile of finite full binary trees. We show that the isoperimetric profile of a full binary tree constantly fluctuates between one and almost the depth of the tree.

Original languageEnglish
Title of host publication43rd International Colloquium on Automata, Languages, and Programming, ICALP 2016
EditorsYuval Rabani, Ioannis Chatzigiannakis, Davide Sangiorgi, Michael Mitzenmacher
Pages1
Number of pages12
ISBN (Electronic)9783959770132
DOIs
StatePublished - 1 Aug 2016
Event43rd International Colloquium on Automata, Languages, and Programming, ICALP 2016 - Rome, Italy
Duration: 12 Jul 201615 Jul 2016

Publication series

NameLeibniz International Proceedings in Informatics, LIPIcs
Volume55

Conference

Conference43rd International Colloquium on Automata, Languages, and Programming, ICALP 2016
Country/TerritoryItaly
CityRome
Period12/07/1615/07/16

Keywords

  • Communication complexity
  • Isoperimetry
  • Monotone computation
  • Separations

All Science Journal Classification (ASJC) codes

  • Software

Fingerprint

Dive into the research topics of 'On isoperimetric profiles and computational complexity'. Together they form a unique fingerprint.

Cite this