Rapid growth in computing power combined with striking advances in algorithm development over the past few decades have ushered in a new, long-anticipated era in which scientists can computationally model realistic materials, assembled from their atoms up, with predictive accuracy. Indeed, being able to predict the properties of materials before ever setting foot in a lab is now a common luxury that many scientists can enjoy with just the click of a mouse. However, the development of many of the software packages that have enabled this new era of desktop science are far from one-man or one-woman efforts: they require a community of theorists, programmers, and experimentalists all working together in a collaborative fashion towards a single end. In this talk, I will describe my experiences as a theorist/modeler contributing to two such large-scale software efforts respectively directed at developing the many-body simulations package, QMCPACK, and the atomistic machine learning package, AMP. In addition to outlining the new science these efforts have revealed, I will reflect on the advantages and disadvantages of such large-scale collaborations, the key roles that theorists, programmers, and experimentalists assume, and the personal and team qualities that ultimately make such endeavors successful.

Building predictive simulations of materials through collaboration

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.