The ability to predict quantitatively how natural and engineered systems will behave is a critical element to the growth and advancement of technology. Such predictions are developed through the application of computers to mathematical models of the system of interest. All such systems are made of materials that are ultimately composed of countless molecules, but often it is not necessary to acknowledge this fact to make useful quantitative predictions. However, with the continued growth and sophistication of nanotechnology, the development of advanced materials, and the growing interest in extremes of application, recognition of the underlying role of molecules in producing physical behavior is indispensable for reliable and accurate predictions. The principles required to achieve this are well established in the laws of quantum mechanics, but the means to convert this understanding into predictions about macroscopic material behaviors is quite limited. Recent advances and trends in computer science and engineering present opportunities to remedy this situation. Two key developments are the advent of multicore processors and distributed computing in general and, apart from this, a focus on data-driven knowledge. Computationally-intensive methods must be reconsidered from the ground up to make real use of these advances in computer science. The promise of new computing technologies is not likely to be met without a more fundamental and radical reformulation of the basic computational approach. This project aims to contribute to these transformations. Future developments open up new avenues for experimental validation of first-principles computational chemistry methods (i.e., requiring no input from experiment), improving them and thereby technologies where they can be applied.
By supplying a route to fluid properties from first principles, this research provides an enabling technology across much of science and engineering while translating fundamental chemistry into applications via new cyber-approaches. Advances made in this project are expected to impact most directly chemical engineering, computational chemistry, and computer science, with a potentially broad array of secondary impacts in areas such as nanotechnology, materials science and engineering, geology, energy, atmospheric science and any other of the myriad fields that can benefit from the capability to predict and model material properties. The development of data-analysis schemes as part of this project can be extended and applied in unforeseen ways to other systems of discrete objects, and the computer-programming tools developed here aim to be sufficiently general to allow application to a diverse set of problems well beyond those that motivate this work. Additionally, education and outreach are promoted via a workshop for high-school students, and via distribution of new easy-to-use open-source software enabling others to apply the methods developed here in their own applications.
This is a Cyber-Enabled Discovery and Innovation Program award and is co-funded by the Division of Chemistry, the Office of Multidisciplinary Activities, the Directorate of Computer & Information Science and the Division of Chemical, Bioengineering, Environmental, and Transport Systems.
