Integrator

An amortized variational inference model to integrate diffraction data.

Prerequisites

The installation requires a python environment manager. I use micromamba, and the following instructions assume it. Another popular manager is conda, but I do not use it nor have experience with it.

Installation

The project depends on DIALS and laue-DIALS alongside PyTorch. DIALS (and dxtbx/cctbx) and PyTorch are installed from conda-forge. Pick the environment file that matches your machine:

File	Env	For
environment.yml	integrator	CPU runtime (local, non-CUDA)
environment-cuda.yml	integrator-cuda	CUDA / GPU runtime (cluster, linux-64)
environment-dev.yml	integrator-dev	CPU + test/lint/logging tooling (development)
environment-cuda-dev.yml	integrator-cuda-dev	CUDA / GPU + dev tooling (development on GPU)

# CPU runtime
micromamba env create -f environment.yml
micromamba activate integrator

# CUDA runtime (GPU machine)
micromamba env create -f environment-cuda.yml
micromamba activate integrator-cuda

# Development (tests, ruff, mypy)
micromamba env create -f environment-dev.yml
micromamba activate integrator-dev

# Development on a GPU machine
micromamba env create -f environment-cuda-dev.yml
micromamba activate integrator-cuda-dev

Each installs DIALS, laue-DIALS, PyTorch, the integrator package (editable), and the upstream reciprocalspaceship build required to read DIALS .refl files.