/MMLPS

Primary LanguagePython

Microkinetics-Guided Machine Learning Pathway Search (MMLPS)

Requirements:

  • conda (recommand)
  • python 3 (py3.9 tested)
  • numpy
  • scipy
  • mpmath
  • pprint
  • networkx
  • MongoDB and pymongo (optional)

Usage

  1. Add the $XXX/pymodule to your $PYTHONPATH

  2. Run a single branch of MMLPS:

    The following directory structure should be prepared, as shown in the example

    job1
├── sourcedir/
    ├──input
    ├──input_split
    ├──CuZnCHO.pot
├── custom_para.py
├── pathsample.py
├── start.arc
    • input: Parameters for SSW-RS, which sample possible reaction pair (lasp.in file)
    • input_split: Parameters for DESW, which identify transition state of possible reaction pair (lasp.in file)
    • CuZnCHO.pot: G-NN potential file
    • custom_para.py: Parameter for MMLPS
    • pathsample.py: Main program
    • start.arc: Starting structure for this branch. It should contain a slab and several molecules.

    To run the simulation, run the following command：

    python pathsample.py

    or you can submit it to the job queue with jobs.sh.

    An analyze directory should be built by pathsample.py, which contain all reaction pair sampled.

  3. Ananlyze a single branch of MMLPS:

    The analyzation is carried out by pathanalyze.py in analyze directory

    To obtain the best reaction pathway, run:

    pathanalyze.py -surface -readallmin 2 -LGibbs 500 -ly ReacNProd.arc