Self-Adaptive Simulations
Computer simulations of physical systems modeled by partial differential equations and particles process quantities of interest on spatial meshes by updating the system state at discrete time intervals. Traditional simulations of many important systems have reached a barrier as existing computational techniques prove to be ill equipped to deal with the complex, multi-physics, multi-scale nature of such systems. To address emerging modeling requirements of today's world, we are developing high-performance, self-adaptive simulation techniques based on state-of-the-art advances in computational mathematics and computer science. Our novel technology (cast into the broad classes of
event-driven and
time-driven simulations) has already enabled remarkable breakthroughs in propulsion and space physics. We are currently extending it to computational fluid dynamics and biology applications.
Data Mining
Advances in technology has led to an unprecedented ability to collect and store data. Data come in a variety of forms and formats ranging from textual content of web pages on the Internet to radio waves from distant galaxies. As a result, we now live in a data-centric world and the ability to extract useful information from raw data (i.e., identify patterns and establish relationships) has never been more critical to our needs. We are working on new technologies for image/data processing, which includes development of intelligent agents and use of discrete event methodology for deriving data dependencies and evolutionary rules.
We have seen the impact that data mining and predictive modeling tools have had on accelerating the discovery and knowledge creation in a variety of fields and areas, including biological, medical and physical sciences as well as the other fields. Any data-rich environment can benefit from these powerful methods and techniques.
Multi-Resolution Full Particle Simulations
Electromagnetic Particle-in-Cell (EM-PIC) simulations are pervasive in a wide variety of fields ranging from design of High-Power Microwave (HPM) sources for radar and communications applications to studies of magnetic reconnection and collisionless shocks in plasma physics. Presently, there are no EM-PIC codes that can adequately resolve multi-scale plasma systems with complex boundaries. The most promising approach is Structured Adaptive Mesh refinement (SAMR). However, there exist a number of well-known difficulties with extending SAMR to EM-PIC simulations. We have recently developed techniques to resolve these problems and are currently developing a multi-resolution, 3D, parallel, objected-oriented, electromagnetic PIC code, EMPOWER (Electro-Magnetic Particle Operation With Extended Resolution). This code will enable first-ever, adaptive, high-resolution simulations of internal and boundary structures of complex electronic devices and plasma configurations.
Technologies
