Approximate Bayesian Inference for Scaling and Merging at Advanced Lightsources

Scaling and merging for large diffraction datasets using stochastic variational inference and deep learning.

This project is under development.

First create a conda env with dials,

conda create -yn abismal -c conda-forge dials
conda activate abismal

Next install abismal. For the CPU version, run

pip install --upgrade pip
pip install abismal

For NVIDIA CUDA support, we recommend you use the anaconda python distribution. The following will create a new conda environment and install abismal:

pip install --upgrade pip
pip install abismal[cuda]

You can now use abismal with GPU acceleration by running conda activate abismal. You can test GPU support by typing abismal --list-devices.