Benchmarking using Frustrated Ising Problems with Planted Solutions

ORAL

Abstract

We introduce a method for generating families of benchmark problems that have a certain degree of tunable ``hardness,'' achieved by adjusting the amount of frustration. We construct such frustrated problems around ``planted solutions,'' representing a known ground state configuration. We construct such problems on the Chimera graph and use them to benchmark various optimization algorithms: simulated annealing (SA), simulated quantum annealing (SQA), the Shin-Smolin-Smith-Vazirani thermal rotor model (SSSV), and the Hamze-Freitas-Selby (HFS) algorithm, as well as the D-Wave device (DW2). We observe a universal hardness peak for all methods, and we observe that the optimal time for the numerical methods scales exponentially with the problem size.

Authors

  • Joshua Job

    • University of Southern California
  • Itay Hen

    • Information Sciences Institute
    • Univ of Southern California
    • Information Science Institute
  • Tameem Albash

    • University of Southern California
    • Univ of Southern California
  • Troels R{\O}nnow

    • ETH - Zurich and Nokia Technologies, Cambridge (UK)
    • ETH Zurich
    • Nokia
  • Matthias Troyer

    • Theoretische Physik, ETH
    • ETH - Zurich
    • ETH Zurich
  • Daniel Lidar

    • University of Southern California
    • Univ of Southern California