Rohini Bala Chandran is an Assistant Professor in the Department of Mechanical Engineering. Her research focuses on developing computational models integrated with experimental analyses to probe the interplay of heat and mass transfer, fluid flow, and chemical reactions that play a central role in a host of thermal, thermochemical, and electrochemical energy systems. Her group leverages high-performance computing to understand coupled transport and kinetic phenomena, and to perform multi-parameter optimization to design and operate efficient devices for energy conversion and storage. Current research activities in her group focus on solar fuels from water and carbon dioxide, thermal energy storage at high temperatures, and wastewater treatment.
Phani Motamarri is an Assistant Research Scientist in the department of Mechanical Engineering. His research interests lie in the broad scope of computational materials science with emphasis on computational nano-science leading to applications in the areas of mechanics of materials and energy. His research is strongly multidisciplinary, drawing ideas from applied mathematics, data science, quantum-mechanics, solid-mechanics, materials science and scientific computing.
The current research focus lies in developing systematically improvable real-space computational methodologies and associated mathematical techniques for conducting large-scale electronic-structure (ab-initio) calculations -via- density functional theory (DFT). Massively parallel and scalable numerical algorithms using finite-elements (DFT-FE) are developed as a part of this research effort, which enabled large-scale DFT calculations on tens of thousands of atoms for the first time using finite-element basis. These computational methods will aid fundamental studies on defects in materials, molecular and nanoscale systems which otherwise would have been difficult to study with the existing state of the art computational methods. Current areas of application include — (a) first-principles modelling of energetics of point defects and dislocations in Al, Mg and its alloys which are popular in light-weighting applications to provide useful inputs to meso-scale and continuum models, (b) providing all-electron DFT input to advanced electronic structure approaches like the GW method for accurate prediction of electronic properties in semiconductor-materials.
Brendan Kochunas is an Assistant Research Scientist in the Department of Nuclear Engineering and Radiological Science. Dr. Kochunas work focus on high performance computing methods, especially parallel algorithms for the 3D Boltmann Transport Equation. He is the lead developer and primary author of the MPACT (Michigan Parallel Characterstics based Transport) code. Currently, leading the development of MPACT and its application within CASL (www.casl.gov) constitutes his research activities.
Dr. Kochunas is the lead instructor of MICDE course Methods and Practice of Scientific Computing. He has created a novel and integrated class curriculum that immerse U-M students in many HPC tools and resources, and teaches them to effectively use these in scientific computing research.
Jesse Capecelatro is an Assistant Professor in the Department of Mechanical Engineering. His research is focused on developing large-scale simulation capabilities for prediction and design of the complex multi-physics and multiphase flows relevant to energy and the environment. To achieve this, his group develops robust and scalable numerical methods to leverage world-class supercomputing resources. Current research activities include adjoint-based sensitivity of turbulent combustion, modeling strongly-coupled particle-laden flows, and multiphase aeroacoustics.
Don Siegel is an Associate Professor affiliated with the Mechanical Engineering Department and the Department of Material Science and Engineering. His research targets the discovery, characterization, and understanding of novel materials for energy-related applications. These efforts primarily employ atomic scale modeling to predict thermodynamic properties and kinetics. These data provide the necessary ingredients for identifying performance limiting mechanisms and for the “virtual screening” of candidate compounds having desired properties. Prof. Siegel is currently exploring several varieties of energy storage materials, lightweight structural alloys, and materials suitable for use in carbon capture applications.
Eric Michielssen is a Professor of Electrical Engineering and Computer Science – Electrical and Computer Engineering Division and Associate Vice President for Advanced Research Computing.
His research interests include all aspects of theoretical, applied, and computational electromagnetics, with emphasis on the development of fast (primarily) integral-equation-based techniques for analyzing electromagnetic phenomena. His group studies fast multipole methods for analyzing static and high frequency electronic and optical devices, fast direct solvers for scattering analysis, and butterfly algorithms for compressing matrices that arise in the integral equation solution of large-scale electromagnetic problems.
Furthermore, the group works on plane-wave-time-domain algorithms that extend fast multipole concepts to the time domain, and develop time-domain versions of pre-corrected FFT/adaptive integral methods. Collectively, these algorithms allow the integral equation analysis of time-harmonic and transient electromagnetic phenomena in large-scale linear and nonlinear surface scatterers, antennas, and circuits.
Recently, the group developed powerful Calderon multiplicative preconditioners for accelerating time domain integral equation solvers applied to the analysis of multiscale phenomena, and used the above analysis techniques to develop new closed-loop and multi-objective optimization tools for synthesizing electromagnetic devices, as well as to assist in uncertainty quantification studies relating to electromagnetic compatibility and bioelectromagnetic problems.
His research makes use of rich information contained in the spectrally resolve observations (chiefly from space) to probe the climate system and gauge the performance of climate models. Topics of his ongoing projects include formulation and design of climate monitoring system based on accurate in-flight calibration system, spectrally resolved radiation budget and radiative feedbacks, detecting spectral signals of climate changes, and model evaluations using spectral data set. In the course of such studies, huge amount of data sets from observations or climate model simulations are fed into radiative transfer model to general spectral radiances at thousands of channels for each grid on the globe and for each time interval. To accurately and efficiently carry out such calculation is only possible with massive high performance computing and, as of today, such task is still computationally challenging.
David Sept is a Professor in the Department of Biomedical Engineering, and he is affiliated with the Center for Computational Medicine and Bioinformatics. The Sept lab works in the area of computational biology and we use a wide array of computational techniques to study protein, drug and cellular systems. In addition to “standard” simulation techniques like molecular dynamics, we are developing new simulation and analysis methods for application in more complex systems.
Most of his research and teaching involves parallel computing of some form: design of scalable algorithms and data structures; applications to numerous scientific problems such as a large multidisciplinary team modeling space weather or a small interdisciplinary group doing imputation on datasets of social preferences; and performance analysis, both experimental and analytical. These projects have used a variety of computer architectures, ranging from tens to hundreds of thousands of cores. He also works on algorithms for abstract fine-grain parallel computer models motivated by concerns such as time/number-of-processors/peak-power tradeoffs and the constraints imposed by the fact that computation is done in 2- or 3-dimensional space. Further, he develops serial algorithms for optimizing adaptive sampling problems such as adaptive clinical trials, algorithms for isotonic regression, and various other computer science problems.
High power laser plasma interactions are interesting for applications such as the generation of energetic, directional electron, photon, ion and neutron beams or inertial fusion energy. Because of the strong electric and magnetic fields that lead to extremely far from equilibrium distributions, describing realistic high power laser interactions with plasma typically requires codes using a fully kinetic description. Professor Thomas’ research involves collisional plasma simulation using Vlasov-Fokker-Planck codes, including implicit methods using Krylov solvers for heat transport problems relating to inertial fusion energy. He is also interested in plasma simulation using particle-in-cell methods, including coupling the plasma code to very energetic photons using a Monte-Carlo method, for ultra intense short pulse laser interactions in radiation dominated regimes.