NetKet 3: Machine Learning Toolbox for Many-Body Quantum Systems

TLDR

NetKet is built around neural quantum states and supplies efficient algorithms for their evaluation and optimization. We introduce version 3 of NetKet, a machine‑learning toolbox for many‑body quantum physics. NetKet 3 is built on JAX, enabling differentiable programming, JIT compilation, and automatic differentiation for arbitrary neural‑network ansätze, while adding GPU/TPU support, symmetry‑group handling, chunking for large systems, dynamics drivers, and modularity.

Abstract

We introduce version 3 of NetKet, the machine learning toolbox for many-body quantum physics. NetKet is built around neural quantum states and provides efficient algorithms for their evaluation and optimization. This new version is built on top of JAX, a differentiable programming and accelerated linear algebra framework for the Python programming language. The most significant new feature is the possibility to define arbitrary neural network ansätze in pure Python code using the concise notation of machine-learning frameworks, which allows for just-in-time compilation as well as the implicit generation of gradients thanks to automatic differentiation. NetKet 3 also comes with support for GPU and TPU accelerators, advanced support for discrete symmetry groups, chunking to scale up to thousands of degrees of freedom, drivers for quantum dynamics applications, and improved modularity, allowing users to use only parts of the toolbox as a foundation for their own code.