Supplementary code and data for: Multipole-enhanced machine-learning dipole moment predictions in non-equilibrium PAHs

Supplementary code and data for the article:Multipole-enhanced machine-learning dipole moment predictions in non-equilibrium polycyclic aromatic hydrocarbonsby Xing Xu, et al.Department of Physics, Guangxi University, Nanning 530004, China I. Contents: Code for the four models: 1_charges (includes training and prediction modules) 2_charge+dipole (includes training and prediction modules) 3_quadrupole (includes training and prediction modules) 4_dipole+quadrupole (includes training and prediction modules) DatasetContains both training and test sets. Dependencies (lib)Includes: EPNN-NNFF-main learned_optimization II. Usage Instructions I. Installation using Anaconda Create and activate a virtual environment:conda create -n MLDM_env python=3.10.13conda activate MLDM_env Install required packages:a) Navigate to lib/learned_optimization and install the optimizer:pip install -e .b) Install NNFF:cd lib/EPNN-NNFF-main/bessel-nn-potentials-velopip install -e .c) Install EPNN:cd lib/EPNN-NNFF-main/epnn-mainpip install -e . Install additional libraries:pip install flax==0.7.4pip install optax==0.1.7pip install orbax-checkpoint==0.4.1pip install numpy==1.26.1pip install scipy==1.11.3pip install chex==0.1.84pip install oryx==0.2.7Note: Version compatibility warnings may appear but can be safely ignored. Install JAX and JAXLib:pip install jax==0.4.19 -f https://storage.googleapis.com/jax-releases/jax_releases.htmlpip install jaxlib==0.4.19 -f https://storage.googleapis.com/jax-releases/jax_releases.html II. Running the Code (Linux Terminal) Navigate to the model folder, e.g.:cd ../1_charges Execute the script:./1.sh III. Output Results Upon completion, the following will be generated: Model files, e.g.:550_model_22775_Charges_601_n1.pkl Prediction results, e.g.: pre_dipole22775_550_model_22775_Charges_601_n1.pkl.jpg Dipole22775_550_model_22775_Charges_601_n1.pkl.json pre_dipole22775_550_model_22775_Charges_601_n1.pkl.txt