As recommended in the ACTS workshop guideline, I am writing this letter to outline my current work and future research plans related to computational science and engineering.
Currently, I am working in the automatic differentiation (AD) team of Argonne National laboratory as a postdoctoral researcher. We are using AD technique to examine the sensitivity of a climate model (uncoupled dynamic-thermodynamic sea-ice model) to various modeling parameters, which requires expensive computational cost with conventional finite difference methods. We are also exploring the possibility of using the AD-generated sensitivity result for tuning the model parameters to maximize the agreement between simulated results and observational data. The work will provide improved simulation results compared to observed sea ice conditions in the polar regions, which are important components in modeling Earth's climate system.
In another application, we have applied AD technique to a parallel, finite-element multi-phase flow simulator for efficient computations of Jacobian matrices. The major applications of the code are in the areas of contaminant transport of subsurface system and in oil, gas, and water flow of petroleum reservoirs. The simulation result of the code shows the detail description of explicit fracture effect on the flow behaviors in highly fractured porous media. With a broad range of numerical experiment, we conclude that AD technique provides accurate and fast derivative calculations for this large scale computational problem.
My future plans are to extend the current work. The first plan is to finish the development of a robust parameter tuning scheme for the sea-ice model, which will allow multidecadal simulations. Automatically generated by the AD derived code, the second derivative information (Hessian matrix) will be used with the Newton's methods. In the second plan, a virtual reality system will be used to visualize the simulation result of the parallel, finite-element multi-phase flow code with the control option of computational steering. The new porous media flow simulator with computational steering option will lead the users in environmental and petroleum industries to better understanding the subsurface flow scenarios in highly fractured subsurface system.
As briefly summarized above, most of my work and future plans is strongly related to the applications of the high performance computing tools such as PETSc, MPI, ADIFOR and CUMULVS. Thus, I believe that the ACTS workshop would be the best chance for me to update my experience with the most advanced scientific computational tools. I hope that you give the best consideration on my application. If you need more information, please let me know.
Sincerely yours.
Jong G. Kim, Ph.D
Mathematics and Computer Science Division
Argonne National Laboratory
9700 S Cass Avenue, Argonne, IL 60439
email:jkim@mcs.anl.gov
Publications:
1. Jong G. Kim and Milind D. Deo, "Finite Element, Discrete-Fracture Model for Multiphase Flow in Porous Media," AIChE Journal, 46, 6, 1120-1130 (2000).
2. Jong G Kim and Paul D. Hovland, "Sensitivity Analysis and Parameter Tuning of a Sea Ice Model," Preprint ANL/MCS-P819-0500, Mathematics and Computer Science Division, Argonne National Laboratory (2000).
3. Jong G Kim and Paul D. Hovland, "Sensitivity Analysis and Parameter Tuning of a Sea Ice Model," Accepted for publication, the Third International Conference on Automatic Differentiation: From Simulation to Optimization, Nice, France, 2000 (2000).
4. Jong G. Kim and Milind D. Deo, "Inexact Newton-Krylov Methods for the Solution of Implicit Oil Reseroir Simulation Problems," SPE 51908, SPE Reservoir Simulation Symposium, Houston (1999).
5. Jong G. Kim and Milind D. Deo, "Comparison of the Performance of a Discrete Fracture, Multiphase Model with Those Using Conventional Methods," SPE 51928, SPE Reservoir Simulation Symposium, Houston, TX (1999).
6. Jong G. Kim and Milind D. Deo, "High Performance Computing in Oil Reservoir Simulation, Paper 244f, AIChE Annual Meeting, Miami Beach, FL (1998).
7. Jong G. Kim and Milind D. Deo, "Multiphase Flow Through Discrete Fracture Networks, Parallel Implementation," Paper 250e, AIChE Annual Meeting, Miami Beach, FL (1998).c(th)2085 5047 y Fi(APS)h(T)-7 b(opical)29 b(Conference)g(on)g(A)n