(1c) Managing Data Spaces, Performing MD, and Analyzing Trajectories with Signac, HOOMD-Blue, and Freud 

Glotzer, S. C., University of Michigan
Anderson, J. A., University of Michigan
Adorf, C. S., University of Michigan
Harper, E. S., University of Michigan
We demonstrate how to efficiently develop a complete, integrated, and reproducible molecular-dynamics workflow covering everything from the setup of a parameter space to the execution of simulations and the post-processing of output data for analysis and visualization. Participants are introduced to the basics of managing data spaces with signac [1]. That includes the management and searching of data and metadata. We show how to setup and execute restartable molecular-dynamics simulations with the GPU-accelerated particle simulation toolkit HOOMD-blue [2] with emphasis on the definition of initial configurations, boundary conditions, and simulation protocols. We present the analysis of trajectory data using the parallelized high-performance analysis code freud [3]. The individual steps will be integrated with signac-flow, which allows us to chain and execute data space operations and easily submit workflows to high-performance clusters.

All tutorial material is presented in the form of published Python Jupyter Notebooks. This enables participants to easily follow all examples, and follow-up on the presented workflows.

[1] https://glotzerlab.engin.umich.edu/signac
[2] https://glotzerlab.engin.umich.edu/hoomd-blue/
[3] https://glotzerlab.engin.umich.edu/freud/