Background: Henning Mortveit is an Associate Professor in the Network Dynamics and Simulation Science Laboratory at the Biocomplexity Institute of Virginia Tech as well as an Associate Professor in the Department of Mathematics. He received his doctorate in mathematics from the Norwegian University of Science and Technology in 2000, and spent five years at the Los Alamos National Laboratory before joining Virginia Tech in 2005. He leads the NDSSL software team and the synthetic information work.
Research: How can we make a city and its physical infrastructures more resilient? How can existing infrastructures be modified to make a city more robust with respect to disasters? How do we start to reason about these questions?
Modern cities are examples of highly complex systems that can be captured as massively interacting systems whose network components co-evolve with time. To capture these systems in a precise mathematical manner is a challenge that must address:
His research involves all of these aspects. The work is centered around the framework of Graph Dynamical Systems (GDS). This class of dynamical systems was introduced as a natural mathematical framework that permits precise modeling of massively interacting systems. They address all of the points above permitting mathematical analysis, modeling, efficient implementions and rigorous verification & validation. Examples of such systems captured by GDS include socio-technical systems (interconnected physical networks and infrastructures, their inter-dependencies, as well as the interplay with human behavior), biological systems, and general distributed systems. Specific research topics include:
GDS fundamental theory: structure-to-function analysis, in particular stability analysis as it pertains to sensitivity analysis, uncertainty quantification and validation. Brute force computations are usually intractable as far as computations go, so this work is about gaining insight about behavior using mathematical theory. Examples of questions include:
- If vertex functions (the way system entities operate) are perturbed, will system behavior change significantly?
- If the network has certain properties (e.g. those characteristic of a power grid), can we guarantee that all dynamics terminate at a steady state?
Critical infrastructure and societal resilience: How does one plan for disaster handling? How does one identify the best or most cost-efficient ways to protect infrastructures and their operation? This work focuses on modeling of infrastructures, their interconnections, human behavioral effects, and on mapping these into validated, scalable simulation models capable of informing policy makers.
Software and system design for scalable, scientific computing: ensuring scientifically reproducible computation for large scale simulation models is a big challenge. Tracking all data sources, their provenance, their transformations to fit required input formats, the tracking of the tools and algorithms used to transform the data, as well as the simulation models used give rise to a host of challenges. This work addresses this problem as well as efficient ways to add and combine simulation models for innovative and rapid modeling and analysis with integrated validation and data quality assessments.
Visualization: scientific visualization of spatial phenomena, in particular for illustrations of dynamics of large, interaction-based systems involving synthetic information.
GDS models on specialized hardware (dormant): acceleration of simulation models by implementing computation-intensive componets on hardware like field programmable gate arrays (FPGAs).