CasNet

A repository used to perform experiments for the work on CasNet.

1. Installation & Setup

type in command line:

• git clone git@github.com:rhjvanworkum/CasNet.git
• git submodule init
• git submodule update
• conda env create -f environment.yml
• conda activate casnet
• cd pyscf
• pip install -e . (this might take a while)
• source env.sh (after setting variables in env.sh file)

This should set up the whole environment (Watch out: We're our own fork a of PySCF that returns macro, micro & inner iterations of the CASSCF object)

2. Performing CASSCF calculations in PySCF

2.1 Run calculations

python data/casscf/pyscf/run_casscf_calculations.py --geometry_folder geometries/geom_scan_200/ --output_folder pyscf/geom_scan_200_sto_6g/ --basis sto_6g

2.2 Analyze results

python data/check_casscf_calculations.py --output_folder pyscf/geom_scan_200_sto_6g/

2.3 Save results in ASE DB

python data/db/save_casscf_calculations_to_db.py --geometry_folder geometries/geom_scan_200/ --output_folder pyscf/geom_scan_200_sto_6g/

2.4 Generate split

check data/splits/ to see example of split generating scripts

3. Training models

3.1 Train ML-MO model

python train_model.py --db_name geom_scan_200_sto_6g.db --split_name geom_scan_200.npz --property mo_coeffs_adjusted --model_name gs200_sto_6g_F

3.2 Train ML-F model

python train_model.py --db_name geom_scan_200_sto_6g.db --split_name geom_scan_200.npz --property F --model_name gs200_sto_6g_F

3.3 Train PhiSNet model

Add a configuration in phisnet_fork/configurations/ folder (e.g. named 'test_conf.txt') and then type:
python phisnet_fork/train.py @phisnet_fork/configurations/test_conf.txt

4. Evaluating models

4.1 Evaluate model loss on train/val/test set

python evaluation/compute_model_loss.py --db_name fulvene_geom_scan_250.db --split_name fulvene_gs_250_inter.npz --mo_model geom_scan_250/fulvene_gs250_MO --F_model geom_scan_250/fulvene_gs250_F --phisnet_model geom_scan_250/fulvene_gs250_phisnet

4.2 Evaluate orbitals on validation/test set

python evaluation/compute_SCF_iterations.py --geometry_folder geometries/geom_scan_200/ --split_name geom_scan_200.npz --mode ML-MO --model gs200_sto_6g_F --basis sto_6g

python evaluation/compute_CASCI_energy_errors.py --geometry_folder geometries/geom_scan_200/ --split_name geom_scan_200.npz --mode ML-MO --model gs200_sto_6g_F --basis sto_6g

4.3 Look at predicted orbitals for a particular geometry

python evaluation/paper_figures/plot_orbitals.py
(check this file for setting geometry & orbital model & MO indices)