/gmid2-public

Code for paper Submodel Decomposition Bounds for Influence Diagrams (AAAI 2021)

Primary LanguagePythonMIT LicenseMIT

gmid2

  • graphical models for influence diagrams
  • Junkyu Lee, Radu Marinescu, and Rina Dechter. "Submodel Decomposition Bounds for Influence Diagrams", AAAI 2021.

Dependencies

$ conda create env --name gmid2 python=3.6
$ conda install numpy, networkx, sortedcontainers
  • numpy version: 1.18
  • networkx: 2.4

Benchmark Problems in UAI file format

  • benchmark data set under /gmid2/gmid2/benchmarks
  • file formats
    • .uai: same as uai file
    • .id: identity of variables and functions
    • .pvo: partial variable elimination ordering
    • .vo: variable elimination ordering

Usage

  • use scripts under /gmid2/gmid2/scripts
$ PRJ_PATH=<path to gmid2>
$ python st_bte_bw.py $PRJ_PATH/benchmarks/synthetic/mpd1-4_2_2_5 2
$ python st_wmbmm_bw.py $PRJ_PATH/benchmarks/synthetic/mpd1-4_2_2_5 2