Installation & Quick Start

Clone the repository

git clone https://github.com/TommyLinkl/DeePseudopot.git
cd DeePseudopot

Install dependencies
```
pip install -r requirements.txt
```

Assemble an input bundle

At minimum, include the following input files for a deep-learning pseudopotential training calculation. Please see Input Data Description for details on input keywords.

Inputs	Purpose
`NN_config.par`	Global training and optimization settings.
`system_X.par`	Defines the periodic system.
`kpoints_X.par`	\(\mathbf{k}\)-point paths in Brillouin zone for the band structures.
`bandWeights_X.par`	Per-band weights in the training loss definition.
`expBandStruct_X.par`	Reference band structures used for training.
`input_X.par`	Convergence controls, plot toggles, and miscellaneous simulation knobs.
`init_<atom>Params.par`	Analytic pseudopotential parameters per element, used for model initialization.

Launch a training run

python main.py /path/to/inputs/ /path/to/results/

Code Repository Layout (for developers)

main.py : Entry point for training pseudopotentials from an input bundle.

eval_fullBand.py : Mirrors main.py but streamlines evaluation of full band structures with inference-friendly parallelism.

docs/ : User-facing documentation, including the Input, Output, Workflow, and Troubleshooting guides published on this site.

utils/ : Core implementation modules such as file readers, Hamiltonian builders, neural-network models, training loops, Fourier transforms, and visualization utilities.

test_ham/, test_parallel/, test_memory/ : Regression and stress-test suites covering band-structure accuracy, multiprocessing behavior, eigensolvers, and memory utilization.