• My research interests have been primarily in numerical modeling of mantle dy namics which include a wide variety of problems, such as how plate tectonics work, how plates and the underlying mantle interact with each other, what kinds of surface features are produced by the dynamic interaction between the mantle and plates, how the stress field produced by mantle convection is related to that deduced from observations of seismicity, crustal isostasy and so on (see a list of my papers). In addition to these large-scale dynamics problems, I also have interests in problems like viscoelast icity, hydrothermal circulation, melting in magma chambers, and crack dynamics.
• I have strong interests in high performance computing and finite element analysis. I have incorporated tectonic faults into 3D models with a finite element code CITCOM. As for all large-scale computations these days, parallel computing is a must. I have ported and optimized the CITCOM onto different parallel comput ers including shared memory and distributed memory machines with MPI. The parallelized CITCOM performs well on these parallel computers, and its efficiency on the 512 nodes Intel Paragon at Caltech is better than 90% for a medium size of problems.
• Recently, I have modified the CITCOM from 3D Cartesian to 3D spherical geometries (the new code is called CitcomS), so we can work on real data for rea l planets like Earth and Mars. CitcomS uses a unique design for numerical grids that fits to parallel computing (see an example of how I grid a planet and cut it to fit to parallel computers). Citco mS also uses a full multigrid solver (Yes, parallelized full multigr id solver). Here is a product of CitcomS -- an animation of mantle convection with surface plates.

go back