Noemi Petra

Assistant Professor
University of California, Merced


PhD (2010), Applied Mathematics, University of Maryland
ICES postdoctoral fellow (2010-2011), and Research
Associate (2011-2014), The University of Texas at Austin

DiaMonD Research

Large-scale inverse problems, optimal experimental design, dimensionality reduction

DiaMonD Collaborations

Alexanderian (NC State), Cui (Monash), Ghattas (UT Austin), Marzouk (MIT), Peherstorfer (Wisconsin-Madison), Stadler (NYU, Courant), Willcox (MIT)


Noemi Petra is Assistant Professor of Applied Mathematics in the School of Natural Sciences at the University of California, Merced. Petra’s research is in large-scale inverse problems governed by differential equation, uncertainty quantification in inference and prediction, and optimal experimental design. Her current research focuses on structure exploiting scalable methods for efficient posterior exploration in high dimensions in the context of ice sheet inverse problems. Additionally, Petra works on (state and parameter) dimension reduction techniques that exploit local sensitivity of the data to (unknown/uncertain) parameters that have promise to mitigate the computational challenges in inference stemming from the high-dimensional state and uncertain parameters.

Impact of DiaMonD

Petra was a Postdoctoral Fellow and later Research Associate with Ghattas at UT Austin, partially supported by DiaMonD. Her research contributions in the area of uncertainty quantification in inference and prediction and optimization for control and design are the result of close collaborations with the DiaMonD PIs and their groups. Collaborations with Ghattas (UT Austin) led to scalable and efficient algorithms for the propagation of uncertainty from data through inference to prediction for large-scale Antarctic ice sheet flow problems. Collaborations with Alexanderian (former UT Austin, now NC State), Ghattas (UT Austin), and Stadler (former UT Austin, now NYU, Courant) led to a quadratic Taylor series approximation-based method for optimal control of systems governed by partial differential equations with uncertain parameter fields. DiaMonD’s vibrant researchers and broad cross-cutting themes offered a unique opportunity for Petra to establish long-term research collaborations. Petra is currently working on model reduction for Bayesian inference with Cui (Monash University), Ghattas (UT Austin), Marzouk (MIT), Peherstorfer (Wisconsin-Madison), and Willcox (MIT).