Neural Network Matrix Product Operator: A Multi-Dimensionally Integrable Machine Learning Potential

Manuscript

- https://doi.org/10.1103/PhysRevResearch.7.023217

For Pompon-main.zip

See README.md at first for installation

Manuscript data for train/validate/test are in docs/data/*npy. They can be loaded by numpy.load.

The reference geometry information of H2CO molecule is docs/data/bagel_h2co_dft.s0.harmonic.json

training script is docs/notebook/_h2co_opt.py. It is easily executable by uv run _h2co_opt.py if you have installed uv. 

The trained weight is in docs/data/nnmpo_final_rmse_8.365e-04.h5. (HDF5 format)

NN-MPO to MPO is in docs/notebook/create-random-mpo.ipynb and docs/notebook/nnmpo_to_itensor_mpo.ipynb (Need ITensors.jl version 0.6.x, For version 0.7.x or later, ITensorMPS.jl would be needed.)

DMRG calculation by ITensors.jl is docs/notebook/itensor_vDMRG.ipynb.

If you have any questions, please post an issue on GitHub.

Discvar-main.zip is an implementation of discrete variable representation (DVR).

dnn-torch.py: MLP benchmark for the same dataset. Requirements: pytorch, tqdm, numpy, matplotlib

exact-ham.h5: Exact MPO of H2CO. (Bond dimension M=731)

nnmpo-ham.h5: NNMPO of H2CO.(Bond dimension M=16)

exact-ham-dmrg.jl: Julia script for phonon DMRG of Exact MPO

nnmpo-ham-dmrg.jl: Julia script for phonon DMRG of NN-MPO

bagel_h2co_dft.mcmc_inverse_pes.db: SQL database of MCMC sampled data

bagel_h2co_dft.s0.harmonic.json: Information about equilibrium geometry