Skip to content

MUTUEL/MCNetwork

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MCNetwork

This code uses a Kinetic Monte Carlo simulation to estimate the behaviour of variable range hopping in Dopant Networks. More information can be found here.

Installation

This installation guide has been written using Ubuntu on WSL2 (Windows Subsystem for Linux), but should be usable on a range of linux operating systems.

Install the dependencies through apt (for Ubuntu/Debian):

sudo apt-get install cmake libboost-all-dev libhdf5-serial-dev

Run the following commands to build the project :

git clone https://github.com/MUTUEL/MCNetwork.git
git clone https://github.com/mfem/mfem.git
mkdir mfem-build && cd mfem-build
cmake ../mfem
make
mkdir ../MCNetwork/build && cd ../MCNetwork/build
env MFEM_DIR=../../mfem-build cmake ..
make

Set up your conda environment (assuming you have anaconda installed already):

conda create --name mcnetwork numpy scipy matplotlib h5py
conda activate mcnetwork

Usage

Define the device parameters you want to use in data/in.txt. To make a new device with Monte Carlo optimization and random voltages go to the data folder in your shell and enter the following command:

../build/MCnetwork --mnd --optMC --rSV

Allowed options:

option description
--continue continues last optimization. optimization mode will be detected, no further options needed
--mnd make new device
--optMC optimize control voltages using Monte Carlo search
--optGen optimize control voltages using genetic algorithm
--optBasinHop optimize control voltages using basin hopping
--run just run control voltages defined in in.txt
--rSV random start voltages (only in combination with opt)
--dir arg define working dir. has to contain 'in.txt'
--verbose additionally outputs occupation, swapps and time to 'additionalData.hdf5'
--help produce help message

The created device can be analyzed using the python scripts in the scripts folder.

#TODO

Technical:

  • set "-D SWAPTRACKER" in cmake file
  • improve argument parsing
  • add log file
  • support INT data format in datafile
  • split up optimizer in child classes
  • make one central hdf5 file to combine SWAPTRACKER and TiMETRACKER

Optional

  • count only output electrode current
  • improve hash algorithm for storing mode
  • in optimzer: store only control voltages
  • in genetic algorithm: do not run unchanged genomes again */

Contributors

License

APACHE LICENSE, VERSION 2.0

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages