William Kleiber
University of Colorado Boulder, CO·Mathematics & Statistics
Some things in nature and in markets do not vary smoothly; they jump. Sunlight reaching a rooftop solar panel can drop by half in seconds as a cloud passes, and a cryptocurrency price can lurch in minutes.
- Fields
Mathematics & Statistics
Earth & Environmental Science
- Topics
spatial statistics
space-time models
non-Gaussian processes
- Grants
- 1 known grant, through Aug 2027
What they work on
- Develop statistical models for multivariate non-Gaussian processes that reproduce intermittent, jump-like behavior, rather than the smooth bell-curve variability assumed by standard space-time models.
- Characterize the variability and intermittency of solar irradiance across space and time using in-situ pyranometer measurements, instruments that record incoming sunlight at a point on the ground.
- Model the joint behavior of major cryptocurrencies from historical price records, so that portfolio and risk studies can use simulated scenarios despite the short real price history.
- Produce methodology for generating realistic synthetic datasets usable in grid impact assessments for distributed rooftop photovoltaic generation and in financial risk analysis.
More about the research
Standard statistical models assume variability follows a bell-shaped Gaussian curve, which cannot reproduce that abrupt, intermittent behavior, and when several panels or several currencies move together, the models also need to capture how the jumps are linked across locations or assets. This project develops new statistical frameworks for multivariate non-Gaussian processes with jump-like behavior, so that realistic artificial datasets can be generated where real data are too scarce for risk and impact studies.
Why it matters
Utilities planning for neighborhoods full of rooftop solar need to know how often and how sharply output can swing; investors and businesses holding multiple cryptocurrencies need to know how those assets move together under stress. By producing realistic synthetic data where real records are too short or too sparse, this work could improve grid impact assessments and financial risk studies, with methods transferable to meteorology, environmental science, and ecology.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Lei Wu
Stevens Institute of Technology, NJ·Engineering
Design likelihood functions and prior distributions for Bayesian models that capture realistic measurement errors and physical characteristics of smart electricity distribution systems.
Anirudh Subramanyam
Pennsylvania State University, PA·Engineering
Develop optimization methods for AC Optimal Power Flow that ensure grid safety limits hold with high probability when wind, solar and load forecast errors are non-Gaussian, using Gaussian mixture models.
Sara Eftekharnejad
Syracuse University, NY·Engineering·Early-career
Build a scalable framework to quantify interdependent power grid uncertainties from renewable generation and component outages, combining cascading failure modeling with machine learning and statistics for near-real-time analysis.
The school
University of Colorado Boulder: 36 labs on PI Pages, $718M a year on research, 447 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.