Resonant HPC: Open Source Pre- and Post-Processing in the HPC environment using Jupyter and ParaView

Abstract:

The environmental scientific software community is facing challenges to utilize advancements in computing and web technologies to enable end-to-end modeling workflows that scale with their data and simulation codes. These challenges present a need for pre-processing tools that enable scientists to quickly prepare data, meshes, and model simulation inputs for their workflows with post-processing visualization and analysis tools in an HPC-capable turnkey platform. ResonantHPC is an integrated, open-source platform addressing current challenges around the pre- and post-processing needs of scalable end-to-end workflows. Our work demonstrates success for Advanced Terrestrial Simulator (ATS) with an end-to-end workflow for defining model input parameters using Computational Model Builder (CMB), running simulations in an HPC environment, and post-processing 3D visualization using ParaView and Jupyter. This system enables researchers to prepare, execute, and analyze the results of remote exascale-level simulations through user-oriented data-entry forms, automatic generation of simulation input decks, HPC simulation job submission, and server-side 3D visualization interfaced to in Jupyter. This platform is designed to be extensible to integrate with the broader ecosystem of simulation software in environmental science research and is available at https://kitware.github.io/resonanthpc/