Information for Prospective Graduate Students
The goal of this page is to explain in simple words my research interests. In case you consider joining forces with me I hope this material will help you better understand where I'm coming from and answer some of your questions. You'll notice that the underlying theme is that of using computers to understand the dynamics of complex (multibody) systems. As such, it relies on knowledge in Mechanical Engineering, Applied Math, and Computer Science. Sometimes people call this mix of things Computational Science. With this in mind, my research interests are as follows:
a) Finding out how systems change in time, and more precisely, developing computer methods to simulate their time evolution. Lately, I became interested in finding out how uncertainty (as in the nature of the force acting on a mechanical system, for instance) reflects into the time evolution (behavior) of a system. In general we only have a rough idea about the nature of the inputs that act on the models we use in simulation, and estimating system behavior under these circumstances is challenging. In the past I've also worked on developing numerical methods for real-time simulation of mechanical systems. With focus on vehicle simulation, the idea was to be able to predict the position and velocity of a car model fast enough for a human to be immersed in a virtual reality environment in which the driver and the vehicle could interact with the surrounding environment/traffic. Finally, I am interested in the simulation of large problems with frictional contact. I have worked with two colleagues on rigid body frictional contact, but I plan to branch into deformable body frictional contact as well. Some animations are available on my research web-page.
b) System level integration for high fidelity vehicle simulation. Over the next four to five years I would like to integrate high accuracy models of vehicle, tire, and terrain to perform large scale simulation of off-road vehicle ride. I hope this work will eventually lead to a simulation capability supporting a level of fidelity that would one to predict/determine tire wear, tire tread design for improved traction, how the suspension should be tuned for maximum speed when running off-road (on sand, for instance), suspension control for improved stability, etc. The underlying theme here is that of cross-discipline simulation.
c) Trying to find out how I can improve the tedious process of modeling. In the past, I worked in industry for a company focused on enabling the concept of "virtual prototyping", a process that drew on a simulation package called ADAMS. The idea here is to replace hardware prototypes of a car, for instance, with a virtual prototype of it, that is, a computer model of the car system. Once a model is available you can quickly and effectively simulate its behavior. You can also start changing all kinds of design parameters to hopefully improve the performance of the system, run simulations that are costly or dangerous to perform (roll-over analysis, for instance), etc. Doing all these things requires a model of the system though. As I said, I'm looking into ways in which producing a system model can be simplified and formalized at a level where you can ideally have a machine do it for you.
d) Learning about the electronic density around nuclei in a nanostructure
as well as the extent to which implicit numerical integration is
feasible in Molecular Dynamics simulation. This is a very rich
and tough problem and I'm only looking at a computational aspect of it in an
effort to reduce the amount of time required to simulate what is
happening at atomic and/or subatomic level. To put things in perspective, a 60-atom
simulation of a nanostructure used to take weeks to complete
(Kohn-Sham DFT). Going to hundreds of thousand of atoms is beyond our
possibilities today. This is a joint project with researchers at Argonne National
Lab.