This project is an implementation of a small part of neural-network quantum states in Tensorflow to speed-up the process with graphics processing units (GPU). The implementation is based on the NetKet library [1] and Science paper by Carleo and Troyer [2]. We also propose several transfer learning protocol for the scalability of the neural-network quantum states based on our paper here [3].
This project is based on Python programming language. We suggest to use Python 2 instead of Python 3. These are the library requirements for the project:
tensorflow==1.15ortensorflow-gpu==1.15numpyscipymatplotlib
It is also available as requirements.txt in the project and do
pip install -r requirements.txt
to install the necessary libraries.
We have provided some scripts named script-[model].py as an example to run the program with the given parameters. The description of the scripts are the following:
script-ising.py: run one-dimensional Ising model from cold-start.script-ising-transfer.py: run one-dimensional Ising model with transfer.script-heisenberg.py: run one-dimensional Heisenberg model from cold-start.script-heisenberg-2d.py: run two-dimensional Heisenberg model from cold-start.script-heisenberg-transfer.py: run one-dimensional Heisenberg model with transfer.script-heisenberg-2d-transfer.py: run two-dimensional Heisenberg model with transfer.
The parameters for the script are explained in script-ising.py. To run, simply use the command python script-ising.py. The script will create a folder called results to store all the results (this can be changed in the script).
[1] G. Carleo, K. Choo, D. Hofmann, J. E. T. Smith,T. Westerhout, F. Alet, E. J. Davis, S. Efthymiou,I. Glasser, S.-H. Lin, M. Mauri, G. Mazzola, C. B. Mendl,E. van Nieuwenburg, O. O’Reilly, H. Th ́eveniaut, G. Tor-lai, F. Vicentini, and A. Wietek, SoftwareX , 100311(2019).
[2] G. Carleo and M. Troyer, Science 355, 602 (2017)
[3] R. Zen, L. My, R. Tan, F. Hebert, M. Gattobigio,C. Miniatura, D. Poletti, and S. Bressan, arXiv:1908.09883 (2019).