The repository implements the deep RL TRPO with LSTM on Kidney Exchange based on Tensorflow. The Kidney Exchange simulator based on the OpenAI Gym can be found here. A simplified dynamic set packing model can be found here.
- Python3
- OpenAI Gym
- Kidney Exchange environment
- OpenAI baselines
- OpenMPI
- Mpi4y
- Spams
- BLAS
- Gurobi
- Tensorflow
- NumPy
- SciPy
- NetworkX
If you have a issue with the Mpi4y,you may have to set the environment variable TMPDIR
as /tmp
by:
export TMPDIR=/tmp
If you use multithreading program and OpenBLAS at the same time, you may have to change the environment variable of OpenBLAS multihreading since it may controdict with your original multithreading setting.
export OPENBLAS_NUM_THREADS=1
in the environment variables. Or
Call openblas_set_num_threads(1)
in the application on runtime. Or
Build OpenBLAS single thread version, e.g. make USE_THREAD=0