The Uncertainty Quantification and Optimization (UQ&O) Group develops methods and software for optimizing, controlling, and assessing uncertainty in highly complex systems. We develop and deploy state-of-the-art HPC algorithms for statistical inference, numerical optimization, machine learning, and more. Group members often work closely with subject matter experts from diverse disciplines both inside and outside LLNL.
UQ&O researchers often contribute to multi-institution collaborations tackling the most challenging problems. Current projects include research into the optimal control of quantum computers, optimization of power and gas networks, uncertainty quantification in materials science and advanced manufacturing as well as foundational work on computer algorithms and software. Group members contribute to several open-source software tools—for example HiOP and Pyomo for numerical optimization; Quandary and juqbox.jl for quantum optimal control; and PSUADE for uncertainty quantification.
Jim Gaffney: uncertainty quantification, experimental calibration of simulation codes, physics-constrained machine learning
Xiao Chen: uncertainty quantification, machine learning, inverse problems, reduced order modeling, data assimilation, PDE-constrained optimization
Nai-Yuan Chiang: numerical optimization, high performance computing, power grid simulation and optimization
Hillary Fairbanks: uncertainty quantification, multilevel/multi-fidelity algorithms for large-scale, nonlinear Bayesian inference
Stefanie Guenther: PDE-constrained optimization, optimal control for machine learning, optimal control for quantum computing, adjoints and automatic differentiation
Yu-Ting (Tim) Hsu: graph neural networks for atomistic structures and microstructures, physics-informed machine learning
Anders Petersson: numerical methods for large-scale wave propagation problems, ODE and PDE constrained optimization, optimal control of quantum systems, seismic wave propagation
Cosmin Petra: mathematical optimization, large-scale optimization, high performance computing optimization solvers, stochastic optimization
Claudio Santiago: integer programming, convex optimization, black box optimization
Charles Tong: uncertainty quantification, optimal experimental design, numerical solution of linear systems of equations, scientific software
Jingyi (Frank) Wang: non-smooth, non-convex optimization algorithms, power grid optimization, parameterized optimization based on finite element models
Jean-Paul Watson: optimization under uncertainty, scenario construction and analysis, mixed-integer programming, and critical infrastructure operations, planning, and resilience