cms21 / als_at_alcf_tomopy_compute

Demonstration of running Tomopy for ALS data at ALCF with Globus tools

Tomopy Reconstruction Demo for ALS at ALCF

This tutorial demonstrates how to perform remote tomography reconstructions using Tomopy for ALS data on Polaris at ALCF using Globus Flows. In this example, Globus flows will launch the application on Polaris and then transfer results from the Eagle filesystem to the Home filesystem at ALCF or a user substituted transfer endpoint at their institution.

This tutorial can be run from anywhere, it only requires a local installation of Globus software (described below) and access to a Globus Compute Endpoint setup by the user on Polaris that has access to tomopy (described in the notebook).

Tomopy on Polaris

Tomopy has been installed in a conda environment on Polaris at this path which is accessible to members of the IRIBeta allocation: /eagle/IRIBeta/als/env/tomopy.

Local Setup

This tutorial can be run from anywhere. The only requirement is a local environment, such as a conda environment, that has python 3.11 installed along with the globus packages globus_compute_sdk and globus_cli. If you have a local installation of conda you can set up an environment that can run the demo notebook with these steps:

conda create -n globus_env python==3.11
conda activate globus_env
pip install globus_compute_sdk globus_cli

Note that the tomopy environment on Polaris contains python 3.11. It is therefore necessary for this environment on your local machine to have a python version close to this version.

Tutorial

Proceed with the tutorial in the notebook.