Current Research ProjectsThe Model Coupling Toolkit
I co-lead the development of the Model Coupling Toolkit (MCT). MCT is a set of software tools for coupling message-passing parallel models to create a parallel coupled model. MCT provides the following core coupling services: a component model registry; domain decomposition descriptors; communications schedulers for parallel intercomponent data transfer and intracomponent data redistribution; a flexible and indexible (i.e., random-access) field data storage datatype; a time averaging and accumulation buffer datatype; a general spatial grid representation capaple of supporting unstructured grids; parallel tools for intergrid interpolation implemented as matrix-vector multiplication; spatial integration and averaging tools (including paired integrals to support conservative interpolation); and tools for merging data from multiple components for use by another component. You can learn more about MCT from our Web Site.
Design and Development of a Terascale Community Climate System Model
This is a joint DOE-NSF project focused on the software and scientific development of CCSM. DOE funding comes from the Scientific Discovery through Advanced Computing (SciDAC) initiative. The three major goals are 1) increasing model performance (in particular throughput) to enable longer simulations, 2) improvement of the overall software quality of the model to make it easier to maintain and extend, and 3) addition of new processes (e.g., atmospheric chemistry and biogeochemistry). Argonne's major contribution to this effort has been the design and development of a modular, distributed-memory, high-performance flux coupler. This new coupler is built on top of a software package called the Model Coupling Toolkit.
Common Component Architecture (CCA) Forum
DOE laboratories have long led the charge to create open-source software to support rapid development of scientific simulation applications. In the past, these efforts have produced numerical libraries and toolkits (for example the Portable Extensible Toolkit for Scientific computation (PETSc)). DOE has also suported the development of software frameworks, which are language-independent software environments for building component-based applications. The Common Component Architecture Forum seeks to define a universal software framework for simulation software. The main manifestation of CCA is the DOE SciDAC-funded Center for Component Technology for Terascale Simulation Software. My main role in CCA is climate applications lead.
Earth System Modeling Framework (ESMF) Project
NASA's Earth System Technology
Office-Computational Technology (ESTO-CT, formerly HPCC) is devoting substantial
funding towards the development of a software framework for use by the
climate modeling and weather forecasting communities. This software
framework has been dubbed the Earth System Modeloing Framework (ESMF),
and I am a co-investigator on the proposal for the framework core.
More information about ESMF can be found here.
Past ProjectsPhysical-space Statistical Analysis System (PSAS)
From 1996-1999 I worked for NASA's Data Assimilation Office (DAO) and the Meteorology Department / Earth System Science Interdisciplinary Center (ESSIC) at the University of Maryland. During this time I worked on various parts of their GEOS Data Assimilation System. Most notably, I helped build their on-line data quality control (QC) system, and participated in the ongoing development of the Physical-space Statistical Analysis System (PSAS), and its message-passing parallel version MPPSAS. Just so you know, PSAS is pronounced pizzaz.
Documentation of the Physical-Space Statistical Analysis System (PSAS) Part II: The Factored-Operator Formulation of Error Covariances; J. Guo, J. W. Larson, G. Gaspari, A. da Silva, and P. M. Lyster, 1998: DAO Office Note 98-04 (download a postscript file).
Documentation of the Physical-Space Statistical Analysis System (PSAS) Part III: The Software Implementation of the PSAS; J. W. Larson, J. Guo, G. Gaspari, A. da Silva, and P. M. Lyster, 1998: DAO Office Note 98-05 (download a postscript file).
Description of the Software Implementation of the PSAS, Slides from my presentation at the HPCC PSAS Workshop, October 26, 1998.
A glossary of terms found in the above slides.
My Less Recent PastFrom 1992-1996 I was a postdoctoral fellow at Macquarie University, and then at the Australian National University. During my time at Macquarie, I was involved in the Model Evaluation Consortium for Climate Assessment (MECCA--this was the forerunner to ACACIA). At ANU, worked in the Centre for Resource and Environmental Studies on regional and global climate issues. I also worked with computer scientists in the Department of Computer Science and the ANU Supercomputer Facility on porting of parallel regional and global weather models to Fujitsu supercomputers.
My Student DaysI started my career as a theoretical physicist. During my student days I did research in soliton theory and in integrability of dynamical systems--basically lots of pen-and-paper stuff. My work in soliton theory demonstrated the applicability of the focusing nonlinear Schroedinger (NLS) equation as a complete envelope theory for the sine-Gordon equation. My dissertation work concerned the integrability of charged particle dynamics in the Earth's magnetotail region (and in particular dynamics in neutral sheet and neutral-line magnetic/electric field combinations). I was able to prove that charged particle dynamics in the neutral sheet configuration were integrable in terms of the second Painleve transcendent. While working on my dissertation, I did a lot of symbolic computation using Mathematica, and also wrote my own particle-pusher code to perform numerical studies to confirm my analytical results.
Professional Societies of which I am a Member
About the Painting at the Top of This PageIt is entitled "Campus Lunch at W & M," and was painted by my uncle Robert Kilbride.