Carnegie Mellon University

Computational Materials Science

Macroscale and Mesoscale
Amit Acharya Kaushik DayalMaarten De Boer Elizabeth Holm | Anthony Ingraffea | Noa Marom | Alan McGaughey | Burak Ozdoganlar  | Anthony Rollett | Paul SalvadorRobert Sekerka | Venkat Viswanathan Bryan Webler | Paul Wynblatt


Atomic Scale and Quantum Mechanics
John Kitchin Elizabeth Holm | Noa Marom | Kaushik Dayal | Alan McGaughey | Michael Widom | Anthony Ingraffea Di Xiao


Interfaces
Kaushik Dayal | David Kinderlehrer | Greg Rohrer | Anthony RollettRobert Suter | Paul SalvadorShlomo Ta’asan | Paul Wynblatt 


Data Science and Machine Learning
Marc De Graef | Elizabeth Holm | Noa Marom | Anthony Rollett | Robert Suter


A succession of national initiatives in materials science & engineering has increasingly emphasized the importance of computational methods for materials discovery and for acceleration of materials development. The complexity of materials means that theory and simulation must be developed at several different length and time scales, spanning electronic structure to component in size, and femtoseconds to years in time. For any given equation set there are many different solutions and therefore codes, which are supported by everything from fully commercial to informal open source. The computational techniques cover all material types and applications, which means that computational materials science is intertwined with essentially all areas of materials research within the department and the university in general. Notable efforts include electronic structure calculations (aka first-principles), molecular dynamics, phase field, Monte Carlo, photonics, and micromechanics using finite element or spectral methods.

Materials science is becoming increasingly data intensive as experiments and simulations generate terabytes of data at a fast rate. Thus, a currently developing thrust is to apply machine learning to and data science techniques to materials science. For example, MSE has always made heavy use of images (i.e., micrographs) but generally in a qualitative manner. Although quantities such as grain size are useful to extract, most images are vastly more data-rich than our field has appreciated. Computer vision (CV) can automate and accelerate the analysis of microscope images and reveal structure-property correlations.  Partnerships with Physics, Mathematics, Chemistry, and all the other Engineering departments are significant.