cdmft - Cluster Dynamical Mean-Field Theory

An implementation of the cluster dynamical mean-field theory in Python using TRIQS.

Installation

• Get Docker

• Get the cdmft Docker image by either docker pull mharland/cdmft && docker image tag mharland/cdmft cdmft or git clone https://github.com/MHarland/cdmft.git && cd cdmft && docker build -t cdmft --build-arg email=YOUREMAIL .

• Run tests docker run --rm cdmft

Run

• Change into the directory in which you have the script to run and also in which the output shall be, e.g. cd example.

• Run docker run --rm -v ${PWD}:/cdmft/run cdmft python ex_cdmft.py or with MPI docker run --rm -v ${PWD}:/cdmft/run cdmft mpirun --mca btl_vader_single_copy_mechanism none -np 8 python ex_cdmft.py

The flag --mca btl_vader_single_copy_mechanism none is unfortunately required to prevent a shared memory issue of MPI.

Develop the cdmft library

I suggest using a Docker container with a bind-mount including the code. In the same directory in which you ran the git clone command, run docker run -it -v ${PWD}:/cdmft cdmft bash. Then you share your code directories with the container and you can also run experiments in any subdirectories.