Have a UROP opening you would like to submit?
Please fill out the form.
Surrogate Models and Code Optimization for Heat-Resilient Neighborhood Design
11: Urban Studies and Planning
May 6th for direct funding; June 5 for credit
David Birge, email@example.com
Extreme heat exposure causes more deaths per year on average in the U.S. and globally than any other natural hazard, including hurricanes and flooding. One of the most pressing issues that climate-change confronts humanity with, therefore, is how to design heat-resilient cities that protect citizens from exposure while allowing critical activities to continue. Building Technology Professor Les Norford, in collaboration with the Leventhal Center for Advanced Urbanism research staff, is developing novel heat-resilient assessment tools for early phase urban planning and architectural design. These tools integrate a number of state-of-the-art energy simulation engines and network analysis tools to calculate energy use for grid resilience, interior and exterior conditions for human safety, and exposure based accessibility to key amenities. We are seeking a summer UROP to work on helping reduce the computational overhead for both running simulations and processing the resulting data outputs. The basic models and metrics are completed, but optimizations are needed to help increase usability. This is the focus of the summer position and students will be given considerable freedom (and responsibility) to explore multiple techniques. This UROP position will ideally be for summer credit. For very strong candidates we will consider UROP for pay at the normal rate. If you are interested, please email David Birge or Les Norford to set up a brief video conference to discuss more.
The ideal candidate will be a creative problem solver, have a general interest in spatial computation, and have a strong background in the following: - Developing surrogate models and other estimators (e.g. multivariate linear regression models) - Python based data science libraries including Pandas - Python based multiprocessing tools (e.g. Dask) - Experience with Jupyter's dashboard functionality would be helpful but not necessary