Introduction

DeepErwin

DeepErwin is python 3.8+ package that implements and optimizes wave function models for numerical solutions to the multi-electron Schrödinger equation. DeepErwin is built on JAX, allowing to define wavefunctions using the familiar numpy-syntax and compiling to highly-performant GPU-models.

In particular DeepErwin supports:

• Optimizing a wavefunction for a single nuclear geometry

• Optimizing wavefunctions for multiple nuclear geometries in parallel, while sharing neural network weights across these wavefunctions to speed-up optimization

• Using pre-trained weights of a network to speed-up optimization for entirely new wavefunctions

• Using second-order optimizers such as KFAC or BFGS

A detailed description of our method and the corresponding results can be found in our publications:

• Gold-standard solutions to the Schrödinger equation using deep learning: How much physics do we need?

• Solving the electronic Schrödinger equation for multiple nuclear geometries with weight-sharing deep neural networks

Please cite the corresponding papers, whenever you use any parts of DeepErwin.

Getting Started

The quickest way to get started with DeepErwin is to have a look at our DeepErwin Tutorial. It covers installation, usage of core functionality and major configuration options.

Afterwards take a look at the comprehensive documentation of the source code and APIs: Full documentation for developers

About

DeepErwin is a collaborative effort of Michael Scherbela, and Leon Gerard, Rafael Reisenhofer, Philipp Grohs, and Philipp Marquetand (all University of Vienna). For questions regarding this code, freel free to reach out via e-mail