Elements: Cyber-infrastructure for Interactive Computation and Display of Materials Datasets

University Of California-Santa Barbara, Santa Barbara CA

Investigators

Abstract

This project will accelerate the progress of scientific research by creating a software infrastructure named TINC (The Toolkit for Interactive Computation). TINC will allow scientists to interactively work with very complex information generated by high performance computing (HPC) clusters. It will provide software tools to assist scientists by facilitating virtual experimentation through interactive visualization, speeding up time to discovery. By tying together the scientists' workflow of computation, scientific data analysis in scripting languages, and visualization, TINC will enable new ways of sharing and disseminating results by allowing researchers to share not only their results, but also their interactive workflow as part of their publications. This research will begin its focus on an important and essential need in the materials science community, speeding up time to discovery of new materials through rapid prototyping using computation. The basic science to be generated through the application of the TINC infrastructure is the study of the electrochemical properties of electrode materials for Sodium ion batteries that will help overcome materials challenges that are preventing the commercialization of this promising technology for large-scale grid storage applications. This important proof of concept will facilitate delving deep into the science while focusing on the generalization of the tool to other disciplines as well. The ultimate goal of TINC is to create a new paradigm for high performance computing, facilitating ease of use by tying together interactive visualization with computation. This paradigm shift may facilitate a connection not only to a wider scientific community but also to an informed general public as well through TINC's focus on reproducibility and provenance tracing. TINC is a computational toolkit that expedites data discovery by improving the interaction workflow in complex data analysis. This improvement is achieved by tightly integrating interactivity, computation and visualization with complex scientific data. By managing the connection between data parameters and on the fly computation, TINC simultaneously tackles the issues of reproducibility and interactive control in the exploration of data with large parameter spaces. With the integration of scripting languages and data notebooks, scientists can study their data with the ease of interactive computation and display. Through a robust caching mechanism it will enable new ways of sharing and disseminating results by allowing researchers to share not only their results, but also their interactive workflow as part of their publications. TINC will tightly integrate interactivity, computation and visualization in the research loop, allowing scientists to more quickly and more deeply understand, compare and validate their data. Thus, TINC will facilitate the merging of complex scientific computational models with high performance interactive visualization and will enable real-time exploration of empirical and theoretical models and large experimental datasets. Provided as a set of python and C++ libraries, TINC will handle parameter space mapping to data, interactive triggering of computation on this parameter space and caching to enable scalability, performance, full reproducibility and data provenance tracking. TINC will be applied to a statistical mechanics study of ionic transport mechanisms and ionic insertion processes in layered intercalation compounds that are candidate electrode materials for Sodium ion batteries. This is of critical importance to the area of materials simulation that focuses on the study of transport mechanisms in alloy systems, where visualizing specific mechanisms experimentally is difficult. TINC will allow computational researchers to propose and verify transport mechanisms in a way not previously possible. This award by the NSF Office of Advanced Cyberinfrastructure is jointly supported by the Division of Materials Research within the NSF Directorate of Mathematical and Physical Sciences. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.