NNEVE is a collection of neural network based solutions to physics based problems. As for now only network for quantum oscillator approximation is fully implemented. Hopefully soon will arrive neural network for solving Navier-Stokes equation based on limited number of measurement points..

Installation

This project is uploaded to PyPI as nneve, therefore can be installed with pip install (Python 3.7 or newer is required)

pip install nneve Collecting nneve Downloading nneve-22.6.28-py3-none-any.whl (20 kB) Successfully installed nneve-22.6.28

Quick example

To view quantum oscillator approximation for states 1 to 7 you can load precalculated weights and acquire model object with following snippet:

from matplotlib import pyplot as plt

from nneve.quantum_oscillator.examples import default_qo_network

# acquire network object with precalculated weights
# for quantum oscillator state 1 (base)
network = default_qo_network(state=1)
network.plot_solution()

plt.plot()

To manually run learning cycle check out "How to run QONetwork learning cycle" in Quantum Oscillator section of docs.

Documentation

Online documentation is available at argmaster.github.io/NNEVE/

Builing docs locally is possible and well automated using tox virtual environments. To be able to build documentation you have to acquire Python==3.8 and tox>=3.24, then you will be able to build docs

tox -e docs docs create: ~/repos/nneve/.tox/docs docs installdeps: -rrequirements-docs.txt ... docs run-test: commands[0] | mkdocs build INFO - [macros] - Macros arguments:... INFO - Documentation built in 0.49 seconds docs: commands succeeded congratulations :)