Physics- & Data-Informed Computational Modeling
About
Computational science and engineering (CSE) harnesses the ever-growing power of computers to push the limits of our understanding of nature, our ability to predict and control physical phenomena, and create new technological advances. CSE complements physical experiments and theoretical developments by means of numerical simulation and analysis tools.
CSE efforts in the AME department at USC span several disciplines, including continuum mechanics (fluids, solids and their interactions), combustion, control and dynamical systems, applied mathematics and statistics, biology and medicine.
Browsing through the websites of our faculty and research labs, you will find a wide range of applications where CSE is enabling scientific discovery and technological breakthroughs, such as: biolocomotion (modeling of swimming and flying bodies), high-speed flight and propulsion (scramjet engines), novel aircraft design and propulsion integration (boundary layer ingestion), oncology (metastasis modeling), geophysical flows (stratification and stability), environment (stochastic modeling of carbon sequestration), energy generation (inertial confinement fusion), astrophysics (supernova explosions), and more.
Research in CSE at USC’s AME department encompasses developments in physics-based modeling, numerical methods, uncertainty quantification (UQ), and data-driven feature extraction, learning and visualization. Parallel computing permeates many of these developments, enabling the use of supercomputers to greatly increase the size of the problems that we can solve and the speed at which we can solve them. High performance computing (HPC) is used, for example, for the most demanding high-fidelity multi-physics simulations in computational fluid dynamics (CFD), as well as to orchestrate ensemble simulations with reduced-order models (ROM) in robust optimization problems for design, model validation and quantification of uncertainties. Big-data analytics allows us to probe and explore the vast numerical datasets that result from simulations and experiments. Both HPC and big-data analysis converge to answer our scientific inquiries.
Rapid evolution of current trends in computer architectures enables unprecedented processing power by exploiting increasingly complex levels of parallelism and architecture heterogeneity (CPUs, coprocessors/accelerators, interconnects). Through our CSE-related courses and research labs in AME, our students learn the specialized programming techniques required to develop and efficiently run their software in parallel on the supercomputers at USC and elsewhere.