The OPTImat Advanced Material Science Simulation Sensitivity, Uncertainty Quantification, and Optimization System

As modeling and simulation become more prevalent and important in industry, government, and academic research, it becomes critical to enable tools for ensuring that the results of those simulations are accurate with quantified confidence. Sensitivity analysis, uncertainty quantification, and optimization techniques should be applied to simulation work more often than they currently are. The issues standing in the way of wider use are tied to accessibility of tools, understanding of the techniques, and the need for their application. An existing scientific cloud computing site is available to assist users in addressing accessibility. Access to simulation tools needs to be “democratized” by addressing user workflows with streamlined input generation and output analysis to enable accelerated material discovery. This project provides these tools through a graphical web-browser-based cloud computing site that brings these techniques to a wider audience, initially focusing on atomistic and molecular scale analysis of materials. The base Phase II successfully built a cloud-based system for performing atomic scale simulations with a streamlined graphical user interface. No installation by the user is required and it seamlessly operates from a desktop browser. Compute clusters are assembled in software at a major cloud computing provider. An atomic-scale simulation tool is currently supported through the interface and automated parameter studies can be performed with the system. Features for tracking dataset and result provenance on the system have been included. The Phase IIA program will extend the system built during the base Phase II to complete a full, commercially deployable, software-as-a-service cloud computing site. Additional scientific simulation tools will be added, an interface for optimization calculations designed and built, tools for post-processing and analyzing simulation results added, and a full commercial site feature set integrated from an existing cloud computing site. A series of public releases will be held during the project to ensure transition to commercial operation. Sensitivity, uncertainty quantification, and optimization calculations are all analyses that most simulation professionals know should be done, but in many cases are not. This is often due to cost and difficulty. Allowing the production of these advanced analyses in a widely available commercial product will be useful for advancing the state-of-the-art in simulation across many industries and scientific areas and will also cut costs and improve results.