Ue-Li Pen

Research Summary

[ Computational Astrophysics, Cosmology ]

Pen works primarily on cosmology, numerical simulations, weak lensing, large scale structure and the cosmic microwave background. Highlights of the last year include the installation of the McKenzie cluster, Canada's fastest supercomputer and the development of fast MHD and PM N-body codes. Various smaller side projects were completed and submitted, including a new perspective on naked singularity in general relativity (with A. Frolov), and gravitational microlensing (with R. Schmidt).

Research Projects:
(September 2002 - August 2003)

Weak Lensing Modelling and Data Analysis

This period has seen the fruition of several research frontiers. Pen's group has entered a production stage for the weak lensing effort. Over this year, in collaboration with students P. Zhang, T. Lu, visitor T. Zhang, and researchers J. Dubinski, Y. Mellier, L. van Waerbeke, which resulted in the submission of 7 new papers on weak lensing during this period. The simulations relied on the previous generation of HPC hardware: the CITA alphaserver GS320 and the large memory (512 GB) Itanium cluster. This effort is now expanding into precision modelling and analysis, which is based on a parallel efficient N-body code written with H. Merz.

Hydrodynamics Simulations

The hydrodynamic effort has matured, and a new parallel MHD code (with P. Arras and S. Wong) and out-of-core cosmological hydrocode (with H. Trac) are now in production mode. Five papers were submitted on this topic over the year, covering the methodology, the Sunyaev-Zeldovich effect, and black hole accretion.

Infrastructure

Efforts on the infrastructure have been very successful, with the very efficient acquistion, installation, and operation of the McKenzie cluster. This 536 processor economical beowulf cluster was first conceived in September 2002, with vendor negotiations through November. The machine was delivered and installed in December, and by January was running production science code as the fastest computer in Canada. The low cost and fast installation was made possible through the efforts of a key innovative team of R. Humble, C. Loken, P. Martin and J. Dubinski. This success is now used to leverage an new generation machine through a CFI proposal, which could bring Canada to the top three scientifc computing nations in the world.

[ Back to CITA People | CITA Research 2003 | CITA Annual Report 2003 ]