Spinor-GPE

An open-source Python package for GPU-accelerated solutions to the two-component nonlinear Schrödinger (Gross-Pitaevskii) equation, with applications to spin-orbit coupled Bose-Einstein condensates.

Benjamin D. Smith, Logan W. Cooke, & Lindsay J. LeBlanc · Python (PyTorch)

Cite
@misc{smith2021spinor,
  title = {Spinor-GPE},
  author = {Benjamin D. Smith and Logan W. Cooke and Lindsay J. LeBlanc},
  year = {2021},
  url = {https://www.benjamindavidsmith.com/software/spinor-gpe/}
}

Spinor-GPE was developed in Professor Lindsay J. LeBlanc’s ultracold quantum gases group at the University of Alberta. It solves the two-component Gross-Pitaevskii equation on graphics processing units, which turns simulations that scale as a power law in grid size on a CPU into ones that scale linearly. The method and benchmarks are described in the companion paper in Computer Physics Communications.

The package is built on NumPy, with the computational heavy lifting done in PyTorch, which runs on either a conventional processor or a CUDA-enabled NVIDIA graphics card.