# Seminars - Applied Mathematics

Upcoming Seminars:

### Contact: Jeremy Kozdon (jekozdon@nps.edu) for zoom information

#### Thursday, June 03, 2021 (via Zoom at 1500)

Quantifying parameter uncertainty for convection within a climate model

Dr. Oliver R. A. Dunbar , postdoctoral scholar, Environmental Sciences and Engineering, California Institute of Technology.

Abstract:

Current state-of-the-art climate models produce uncertain predictions, as evidenced by the variability in competing models, but they are typically ill-equipped to quantify this uncertainty. The models necessarily contain simplified physical schemes used to represent small-scale dynamics or poorly understood physics. The schemes depend upon parameters that are calibrated (often by hand) to fit data, though there may be a range of parameters that feasibly produce a given piece of data. In climate models, the uncertainty of the parameters used for convection schemes is the dominate uncertainties of resulting decadal predictions; it is therefore essential to quantify it to gain meaningful. Unfortunately, this task is far more computationally intensive than parameter calibration, and historically has been out of reach of climate models. However, we formulate a suitable Bayesian inverse problem for time-averaged statistical data, and successfully make uncertainty quantification possible by applying the new Calibrate-Emulate-Sample (CES) methodology. CES is based on three steps: a first Calibration step, which takes the climate model as a black box input, and is well adapted to high performance computing architectures; a second Emulation step automates, smooths, and speeds up calculation of the black box climate model by several orders of magnitude, by making use of Gaussian processes (a machine learning tool); a final Sampling step may then be applied using standard methods from computational statistics to quantify the uncertainty in the calibration.

In this talk, we consider an idealized aquaplanet general circulation model (GCM). We use CES to perform uncertainty quantification on the closure parameters for convection.

Biography:

My current interests are in mathematical and statistical modeling for physical systems, and in the corresponding inverse and data assimilation problems to learn from data.

I have experience with mathematical methods such as optimization and variational methods for partial differential equations, modeling free boundary and shape optimization problems, regularization for deterministic inverse problems, and fluid and solid mechanics. I also have experience in statistical methods such as Bayesian inverse problems, uncertainty quantification, data assimilation, Bayesian experimental design. Most recently I have been working on machine learning and model emulation, graph-based learning, and partial differential equations on graphs.

#### Thursday, June 10, 2021 (via Zoom at 1500)

Why Does Deep Learning Work for High Dimensional Problems?

Prof. Wei Kang , Department of Applied Mathematics, Naval Postgraduate School

Abstract:

Deep learning has had many impressive empirical successes in science and industries. On the other hand, the lack of theoretical understanding of the field has been a large barrier to the adoption of the technology. In this talk, we will discuss some compositional features of high dimensional problems and their mathematical properties that shed light on the question why deep learning works for high dimensional problems. It is widely observed in science and engineering that complicated and high dimensional information input-output relations can be represented as compositions of functions with low input dimensions. Their compositional structures can be effectively represented using layered directed acyclic graphs (DAGs). Based on the layered DAG formulation, an algebraic framework and approximation theory are developed for compositional functions including neural networks. The theory leads to the proof of several complexity/approximation bounds of deep neural networks for problems of regression and dynamical systems.

#### Past Seminars

- February 11, 2021 | Zoom | 1500

Topics at the Intersection of Deep Learning and Control Theory

Prof. Wei Kang, Department of Applied Mathematics, Naval Postgraduate School - February 18, 2021 | Zoom | 1500

Optimal Boundary Control of a Nonlinear Reaction Diffusion Equation via Completing the Square and Al'brekht's Method

Prof. Arthur J. Krener, Department of Applied Mathematics, Naval Postgraduate School - March 4, 2021 | Zoom | 1500

A domain decomposition Rayleigh-Ritz algorithm for symmetric generalized eigenvalue problems

Dr. Vassilis Kalantzis, Research Staff Member, IBM Research USA, Thomas J. Watson Research Center - March 11, 2021 | Zoom | 1500

A Split-Form, Stable, Hybrid Continuous/Discontinuous Galerkin Spectral Element Method for Wave Propagation

Prof. David Kopriva, Department of Mathematics, Florida State University and Computational Science Research Center, San Diego State University - April 9, 2021 | Zoom | 1500

An approximation theory perspective on deep learning

Prof. Alex Townsend , Department of Mathematics, Cornell University - April 22, 2021 | Zoom | 1500

Polynomial-free, Variable High-order Methods using Gaussian Process Modeling for Computational Astrophysics

Prof. Dongwook Lee , Applied Mathematics, University of California, Santa Cruz - May 06, 2021 | Zoom | 1500

Hidden Physics Models

Prof. Maziar Raissi , Applied Mathematics, University of Colorado Boulder - May 13, 2021 | Zoom | 1500

The Coming of Game Theory

Prof. Guillermo Owen, Department of Applied Mathematics, Naval Postgraduate School - June 03, 2021 | Zoom | 1500

Quantifying parameter uncertainty for convection within a climate model

Dr. Oliver R. A. Dunbar , postdoctoral scholar, Environmental Sciences and Engineering, California Institute of Technology. - June 10, 2021 | Zoom | 1500

Why Does Deep Learning Work for High Dimensional Problems?

Prof. Wei Kang, Department of Applied Mathematics, Naval Postgraduate School

- February 25, 2020 | Spanagel 257 | 1500

Wasserstein Gradient Flow for Stochastic Prediction, Filtering and Control: Theory and Algorithms

Prof. Abhishek Halder , Department of Applied Mathematics, University of California, Santa Cruz

- January 10, 2019 | Spanagel 257 | 1500

Control through canalization in modeling the innate immune response to ischemic injury

Prof. Elena S. Dimitrova, School of Mathematical and Statistical Sciences, Clemson University - January 14, 2019 | Spanagel 257 | 1500

Computational physics at extreme scales: efficient solvers for discontinuous Galerkin methods

Dr. Will Pazner , Center for Applied Scientific Computing, Lawrence Livermore National Laboratory - January 28, 2019 | Spanagel 257 | 1500

Rapid mixing bounds for Hamiltonian Monte Carlo under strong log-concavity

Dr. Oren Mangoubi, Computer Science, Ecole polytechnique fédérale de Lausanne (EPFL) - Thursday, February 21, 2019 | Spanagel 257 | 1500

title: 50 Years History of the Cross Correlation between m-Sequences

Prof. Tor Helleseth , Department of Informatics, University of Bergen - Monday, March 04, 2019 | Location: Watkins 146 | 1500

title: Regulation-Triggered Batch Learning: A New Hope for Adaptive Aircraft Control

Prof. Miroslav Krstic, Department of Mechanical and Aerospace Engineering, University of California, San Diego - April 17, 2019 | Spanagel 257 | 1500

title: The Power of Interpolation: From Linear Algebra and Approximation Theory to Exascale and Beyond

Dr. Anthony P. Austin, Department of Mathematics, Virginia Tech - July 25, 2019 | Spanagel 257 | 1500

title: Chebfun: Numerical Computing with Functions

Dr. Anthony P. Austin, Department of Mathematics, Virginia Tech - November 04, 2019 | Spanagel 257 | 1500

title: Chebfun: Numerical Computing with Functions

Dr. Boumediene Hamzi, ment of Mathematics Imperial College London