I am an applied mathematician working in the Optimization and Uncertainty Estimation department in the Computation, Computers, Information and Mathematics center. My academic background is in mathematical biology, but my research interests in computational mathematics are more broad, with strong ties to applications.
- Inverse problems and sensitivity analysis with engineering and biological models using simulated and/or experimental data. Model calibration and extrapolation under uncertainty.
- Algorithms combining optimization and uncertainty quantification. Non-intrusive methods for efficient uncertainty quantification.
- Modeling and control theory, especially with biological applications, including population dynamics, in-host infection dynamics and optimal treatment interruption strategies.
- Scientific computation, including simulation of models, optimization, and parallel computing.
- Close collaboration with mathematicians, statisticians, engineers, and other disciplinary scientists to model systems and analyze data.
Upcoming and Ongoing Appearances
DAKOTA is a
freely available software framework for
large-scale engineering optimization and
uncertainty analysis. I serve on the DAKOTA
development team, and developed and maintain its interface
and its scaling capabilities including logarithmic as well
as automatic and user-prescribed characteristic value
scaling. I support DAKOTA users with applications and
perform installation/debugging on various platforms.
I am developing a network-based model for
disease propagation. The model will be used in
an inverse problem context to determine the initial
location and severity of a disease outbreak, given
knowledge of early symptomatic people in hospitals. The
model is high fidelity in that it accounts for all
individuals moving about a geographic region. Our goals
for it include the ability to simulate various diseases
via modules, parallel scalability, reduced order modeling
through population sampling, and flexibility in input
I am researching new algorithms to perform
model calibration under uncertainty. The
goal of this work is to use ensemble experimental data to
characterize distributions of model input parameters in
order to perform further model simulations and
Top of page
Brian M. Adams
Senior Member of Technical Staff