KerM

Kernel Mixing Model for Turbulent Combustion and Transport PDF method.

This repo is for 0-D simple validation, for PaSR validation please refer to PaSR

1. Usage

• Fortran version:

  • src_fortran/mixing_test.f90: main program for testing.
  • src_fortran/mixing_model_kerm.f90: source code of KerM with differential diffusion supported.
  • src_fortran/mixing_model_iem.f90: source code of IEM, used for comparison.
  • src_fortran/mixing_model_mcurl.f90: source code of MC, used for comparison.
  • src_fortran/emst.f and src_fortran/emst_subs.f: source code of EMST adopted from ¹, used for comparison.

  # build executable file (single precision float)
  make fortran
  # run simulation, faster than cpp version
  ./mix
  # show results
  python performance.py fortran

  • Comparison results of 10k particles for EMST and 50k particles for other models (KerM sigma_k=0.25)

  • Performance of mixing models (right side is achieved by further optimization and acceleration, but leads to around 2% timescale error)

    

• C++ Version:

  • src_cpp/MixingModels.hpp: implementation of mixing models
  • src_cpp/main.cpp: run simulation
  • performance.py: show results

  # build executable file
  make
  # run simulation, taking around 10 seconds for N=1000
  ./mix
  # show results
  python performance.py

  Comparison results of 4k particles for EMST and 100k particles for other models (KerM sigma_k=0.25)

  

  Performance of mixing models (EMST-D do not account aging properties and no IEM assisted, for original EMST implementation, please refer to the Fortran version)

  

• Python Version:

  • src_python/models.py: implementation of mixing models
  • src_python/mixing_py.py: run simulation

  # run simulation, EMST taking around 10 minute for N=1000
  python src_python/mixing_py.py

  Comparison results of 1k particles for EMST and 10k particles for other models (uniform weighted samples, KerM sigma_k=0.25)

  

2. Implementations

Please refer to TheoryGuide.pdf

Footnotes

1. Original EMST implementation https://tcg.mae.cornell.edu/emst/ ↩