In this repository we provide a simple PyTorch implementation to solve the 1D quantum harmonic oscillator (HO) with the stochastic reconfiguration method (SR) and two different types of complex-valued neural networks (NNs) employed as avefunction ansätze, or neural quantum states (NQS).
The first network consists of a single neuron, or perceptron, with a Gaussian activation function (GASP). The second one, is a common multilayer perceptron or NQS. In both cases, we use SR to obtain the ground state and perform subsequent dynamics upon a displacement of the trapping potential.
├── assets
├── LaTeX # LaTex files
│ ├── GASP.pdf > Report
├── Mathematica # SR calculations for GASP
│ ├── GASP_2.nb > Two complex parameters
│ └── GASP.nb > One complex parameter
├── Python # Pytorch code
│ ├── analysis.py > data processing methods
│ ├── integrators.py > RK4 and Euler integrators
│ ├── main.py > main code
│ ├── models.py > GASP and NQS classes
│ ├── parameters.py > list of parameters
│ ├── plots.py > plotting methods
│ ├── stochastic_reconfiguration.py > SR methods
│ └── utilities.py > useful methods
├── .gitignore
├── environment_simple.yml # environment dependencies simplified
├── environment.yml # environment dependencies
├── LICENSE
├── notes.txt
└── README.md
- Clone the repository:
Open Command Palette:
ctrl + P
After selectig the repository, a pop-up window will ask where to save it.
- Create the environment:
Open terminal:
ctrl + `
conda env create -f environment.yml
You can choose a specific name using
-n
conda env create -n new_environment_name -f environment.yml
Use
environment_simple.ymlfor minimal installation.
- Activate environment.
Using the terminal:
conda activate name_of_environment
Using Command Pallete:
You should see your environment in the list
