Skip to Main Content

2018 Research Highlight: MEDE-Data Science Cloud (MEDE-DSC)

MEDE-Data Science Cloud (MEDE-DSC)

CMEDE Researchers
Prof. Tamás Budavári,
Johns Hopkins University
Mr. David Elbert
Johns Hopkins University
Prof. Lori Graham-Brady
Johns Hopkins University
Prof. K.T. Ramesh
Johns Hopkins University
Prof. Erica Schoenberger
Johns Hopkins University
Dr. Adam Sierakowski
Johns Hopkins University

The MEDE Data Science Cloud’s materials-specific infrastructure provides datacuration, visualization, and analysis for diverse, materials-domain problems. The MEDE-DSC is built on SciServer with data-centric computing infrastructure and collaborative integration into the materials design loop. Shared data are accessible from local, containerized, computational tools using a web based, Jupyter frontend. Version-controlled containers and notebooks bring power, consistency and transparency while moving towards reproducible, narrated computation. RESTful APIs provide integration to other MGI resources.

This year the MEDE-DSC has provided support for analysis of HIDRA (Highvoltage, In-situ, Diagnostic Radiographic Apparatus) data from the WMRD ballistics range at ARL. In collaboration with Dr. Brian Schuster, we’re working on accelerated analysis for time-resolved imaging of failure and fracture in boron-carbide ceramics. The analysis automates image registration and feature correlation across HIDRA’s eight flash X-ray images allowing capture of penetrator parameters including dwell time, velocity, rod consumption, and penetration depth. Automating repetitive data extraction expands the options for experimental design and scaling.

The MEDE-DSC continues to work with Dr. Shawn Coleman on data curation for atomistic simulations of grain-boundaries in canonical materials. For this project, we’ve prototyped hosting grain-boundary data in the NIST Materials Data Curation System (MDCS) and utilized the RESTful API access for data access. Data federation is done with OAI-PMH data harvester and provider functionality.

A central role for the MEDE-DSC is helping MEDE researchers meet Big Data challenges from advances in instrumentation and computational modeling. Towards this end, we continue to develop more effective ways to capture and import the large, diverse data commonplace in materials today. In collaboration with the PARADIM Materials Innovation Project at Johns Hopkins, our work includes developing automated data streaming from instruments and user facilities.