PHYSBO is a Python library for researchers mainly in the materials science field to perform fast and scalable Bayesian optimization based on COMBO (Common Bayesian Optimization). Users can search the candidate with the largest objective function value from candidates listed in advance by using machine learning prediction. PHYSBO can handle a larger amount of data compared with standard implementations such as scikit-learn.

Homepage

Installed latest version

1.1.1 (2023/2/1)

How to use PHYSBO on the ISSP supercomputer 

    • In the following, replace $MA_ROOT with the following directory.
      • System B
        • /home/issp/materiapps/oneapi_compiler_classic-2023.0.0–openmpi-4.1.5
      • System C
        • /home/issp/materiapps/PrgEnv-intel_8.3.3
    • The directory where sample scripts and input files are stored.
      • Sample scripts: $MA_ROOT/physbo/sample_jobscript/physbo.sh
      • Input files: $MA_ROOT/physbo/physbo-1.1.1-1/examples/
    • How to run PHYSBO
      1. Preparing the computing environments 
        $ source $MA_ROOT/physbo/physbovars.sh
        This will pass the PATH to python3 of the latest version of PHYSBO installed. 
      2. Preparing  input files
        $ cp -rf $MA_ROOT/physbo/physbo-1.1.1-1/examples .
        $ cd examples
      3. Submitting job

        • A sample script is provided in $MA_ROOT/physbo/sample_jobscript/physbo.sh. Copy the script to the current folder and submit the job.
        • For System B

          $ cp $MA_ROOT/physbo/sample_jobscript/physbo.sh .
          $ sbatch physbo.sh
        • For System C

          $ cp $MA_ROOT/physbo/sample_jobscript/physbo.sh .
          $ sbatch physbo.sh
      4. Check the results
        The result is output to the file ( for system B slurm-xxxxx.out, system C physbo.sh.oxxxxx where xxxxxx is the job number). For more information on PHYSBO, see the manual.

    Measurement of the number of times used

    • When you use PHYSBO, please call the following command: 
      $ /home/issp/materiapps/tool/bin/issp-ucount physbo
      Then, the number of times PHYSBO is used on the ISSP supercomputer is measured.
    • We measure the number of uses of PASUMS software on the ISSP supercomputer. This becomes an important indicator for evaluating the significance of the project, and we appreciate your cooperation. We can also count the number of users who use versions of the software that are not pre-installed, so we would appreciate your cooperation in counting the number of users to help us understand how the software is being used (for details, please click here).

    Contact

    • How to use PHYSBO (e.g., create inputs), functions, etc.
      PHYSBO on GitHub: https://github.com/issp-center-dev/PHYSBO
    • How to Execute on ISSP Supercomputer System
      Software Consultation Service for ISSP Supercomputer
      center-apps__at__issp.u-tokyo.ac.jp (please replace __at__ by @)