Big Data and Nanotechnology
Amanda S Barnard
Office of the Chief Executive (OCE) Science Leader
Head, Virtual Nanoscience Laboratory
Commonwealth Scientific and Industrial Research Organisation
343 Royal Parade, Parkville, VIC 3052, Australia
Scientists and engineers have been generating big data for decades. In most cases, however, this has been as part of a targeted search for one specific piece of information; one particular data point. Much of the remaining data, although a necessary part of the search, remains underutilised. An example of this is thousands of configurations that are tested during random structure searching algorithms used to rapidly test hypothetical materials; stored and then forgotten when the lowest energy “stable” configurations are found. This data is a valuable resource just waiting to be mined using an array of established mathematical techniques. In this presentation we will explore the use of some simple statistical methods and multivariate data analytics to predict the impact of distributions and mixtures in ensembles of nanostructures. We will see how to predict structure/property relationships for entire samples of structures, and how we can simulate the impact of different manufacturing processes that restrict the polydispersivity before they enter production. Polydispersivity is not necessarily detrimental to performance, and not all attempts to achieve monodispersivity will yield results.
