heikoburau / ANNonGPU

Implementation of the Stochastic Reconfiguration algorithm on the GPU using C++/CUDA.

ANNonGPU

This library implements the algorithm of Carleo and Troyer for representing quantum many-body wavefunction using various artificial neural network (ANN) architectures on the GPU.

Installation

cd ANNonGPU
cmake . -DCMAKE_INSTALL_PREFIX=<e.g. $HOME/.local>
make install

(for python interface only)
cd pyANNonGPU
python setup.py install

Features

• Supported network architectures: RBM, deep CNN, vCN
• Markov-Chain Monte-Carlo sampling and exact summation
• Arbitrary quantum operators
• Generic interface for gradient descent optimization
• Stochastic reconfiguration using TDVP