This repo contains all of the code required to reproduce the results from our paper https://doi.org/10.1063/5.0141145 .
In addition to standard python libraries, pyscf is required, as are HyQD's coupled cluster module and Quantum Systems module. Quantum systems requires BSE.
The folder coupled_cluster/cc-machinelearning contains 12 files, 3 of which have a name that begins with plot. Running the remaining 9 files produces a one file each, that can be read with pickle.
The plot*.py-files reproduce all figures (except for figure 1) and the numerical results.
Depending on the method considered, the pickle-readable files contains information about the energies using the different methods considered, as well as information about the sample cluster operator, the sample geometry, the parameters learned by the machine-learning algorithm etc.
In the files HF*.py, we have thoroughly commented what is written to file, such that the data should be easily accessible and verifiable. For other two molecules, we essentially produce the same data. The plot*.py-files contain examples how to access the data.
In addition, we made a file extract_ML_information_HF.py which exemplifies how to extract the cluster amplitudes, the orthogonalized cluster amplitudes, and the ML-coefficients