SciPost logo

SciPost Submission Page

Fast counting with tensor networks

by Stefanos Kourtis, Claudio Chamon, Eduardo R. Mucciolo, Andrei E. Ruckenstein

This is not the latest submitted version.

This Submission thread is now published as

Submission summary

Authors (as registered SciPost users): Stefanos Kourtis
Submission information
Preprint Link: scipost_201908_00005v1  (pdf)
Date submitted: 2019-08-05 02:00
Submitted by: Kourtis, Stefanos
Submitted to: SciPost Physics
Ontological classification
Academic field: Physics
Specialties:
  • Artificial Intelligence
  • Computational Complexity
  • Data Structures and Algorithms
  • Condensed Matter Physics - Computational
  • Quantum Physics
  • Statistical and Soft Matter Physics
Approaches: Theoretical, Computational

Abstract

We introduce tensor network contraction algorithms for counting satisfying assignments of constraint satisfaction problems (#CSPs). We represent each arbitrary #CSP formula as a tensor network, whose full contraction yields the number of satisfying assignments of that formula, and use graph theoretical methods to determine favorable orders of contraction. We employ our heuristics for the solution of #P-hard counting boolean satisfiability (#SAT) problems, namely monotone #1-in-3SAT and #Cubic-Vertex-Cover, and find that they outperform state-of-the-art solvers by a significant margin.

Current status:
Has been resubmitted

Reports on this Submission

Report #2 by Anonymous (Referee 2) on 2019-10-17 (Invited Report)

  • Cite as: Anonymous, Report on arXiv:scipost_201908_00005v1, delivered 2019-10-17, doi: 10.21468/SciPost.Report.1238

Strengths

1- The paper is well-organized and well-written.
2- The paper is perfectly understandable for the tensor-network expert who is not familiar with CSPs, and (I assume) the other way around.
3- The benchmarks or "numerical experiments" are sound and the comparison with other state-of-the-art methods is convincing.

Weaknesses

none

Report

It would be good to clearly mention somewhere to what extent the two test cases are favourable to the tensor-network algorithms as compared to other techniques. The speed-ups that the authors found for these two cases, are these expected to be generic for a broad class of problems?

Requested changes

none

  • validity: high
  • significance: good
  • originality: ok
  • clarity: top
  • formatting: perfect
  • grammar: perfect

Report #1 by Anonymous (Referee 1) on 2019-8-25 (Invited Report)

Report

The paper seems to be written well and discussions are convincing. I recommend the publication.

  • validity: good
  • significance: good
  • originality: good
  • clarity: good
  • formatting: good
  • grammar: good

Login to report or comment