Research Interests

Numerical optimization, especially problems involving real (as opposed to integer or discrete) variables. I'm interested in the theory, algorithms, and implementations, and in applications of all types.

Positions

• George B. Dantzig Professor of Computer Sciences.
• Sheldon B. Lubar Chair of Computer Sciences.
• Amar and Balinder Sohi Professor of Computer Sciences.
• Discovery Fellow in the Wisconsin Institute for Discovery at UW-Madison.
• Past Chair of the Mathematical Optimization Society .
• SIAM Fellow and former member of the Board of Trustees of SIAM (2005-2014).
• Director of the Institute for Foundations of Data Science.

Research Projects

• Design and analysis of algorithms for continuous optimization
• Using optimization to solve problems in signal and image processing, machine learning, computational statistics, process control, and other areas.

Some Past Research Projects

• Optimization software: PCx (linear programming), OOQP (convex quadratic programming).
• Compressed sensing software: GPSR, SpaRSA. Also GPU codes for signal and image processing, and TV denoising software, and the LPS code for regularized logistic regression.

Teaching

Office Hours:
Fall 2020: Monday 4-5, Thursday 4-5
Please email to arrange meeting on Zoom

Current:

Next:
CS524 (Spring 2021)

Previously Taught:
CS726: Nonlinear Optimization I (UW, Fall 2019)
CS525: Linear Programming (UW, Fall 2015)
CS730: Nonlinear Optimization II (Spring 2020)
CS524: Introduction to Optimization (Fall 2018)

Publications

Links

• See some of my talks on Videolectures.
• MW and the earlier metaNEOS project: Grid Computing and Optimization
• NEOS Server, and NEOS Guide / Wiki, now hosted at the Wisconsin Institutes for Discovery after many successful years at Argonne National Lab.
• Optimization Online, the latest and greatest eprints on optimization (be sure to post your finest work!)
• Interior-Point Methods Online, an older archive of interior-point papers and stuff
• Mathematical Optimization Society and SIAM Activity Group on Optimization
• GPSR: Gradient Projection for Sparse Reconstruction and SpaRSA: Sparse Reconstruction by Separable Approximation. We also have a GPU implementation of SpaRSA.
• TV Denoising Software
• Sampling methods for Stochastic Programming
  
• The Onion and Doonesbury