Roch Guerin
Washington University in St Louis, MO·Computer Science
Understanding climate change depends on computer simulations of how the chemistry of Earth's atmosphere changes over time. GEOS-Chem is one of these tools, a 3D atmospheric chemistry simulator, and it is slow because it is so complex.
- Fields
Computer Science
Earth & Environmental Science
- Topics
GEOS-Chem
atmospheric chemistry modeling
atmospheric composition
- Grants
- 1 known grant, through Dec 2026
What they work on
- Design methods to assign GEOS-Chem atmospheric columns on its cubed-sphere grid to processors so loads stay even, accounting for cost peaks at sunrise and sunset.
- Solve the resulting vector scheduling problem, a multi-cost task-balancing problem, while accounting for both computation and communication between neighboring columns.
- Develop task assignment solutions that work across different computing environments, including the cloud, and release them through GitHub and the main GEOS-Chem software.
More about the research
Cloud computing offers nearly unlimited processors, and the simulation can be split into many small tasks to run in parallel. The hard part is deciding which processor gets which task so the whole run finishes as fast as possible. This collaborative project, joining computer scientists with atmospheric scientists at Washington University and MIT, tackles that scheduling problem.
Why it matters
Faster runs of GEOS-Chem would let atmospheric scientists test more scenarios of how atmospheric composition responds to change. That could speed up research on what drives climate change and on ways to reduce it. Improvements are planned for release to the GEOS-Chem community through GitHub and the main software.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Douglas Thain
University of Notre Dame, IN·Computer Science
Develop NBFlow, a toolkit that records notebook executions and converts them into workflows that schedule tasks concurrently on high performance clusters or clouds.
Xu Liang
University of Pittsburgh, PA·Computer Science
Build an "information coupling" mechanism in CyberWater2 that lets heterogeneous Earth system models exchange data two ways across platforms, disciplines and institutions without hand-written glue code or edits to each model's source.
Colleen Jones
Utah State University, UT·Earth & Environmental Science
Conduct a year-long field campaign near Salt Lake City measuring elemental and divalent mercury, related oxidants, and chemical and weather variables, using a dual-channel continuous mercury system that avoids the low bias of standard methods.
The school
Washington University in St Louis: 97 labs on PI Pages, $1.2B a year on research, 365 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.