Faculty Spotlight: Aaron Sidford
MS&E welcomes new faculty member, Aaron Sidford! We are excited to have Professor Sidford join the Management Science & Engineering faculty starting Fall 2016.
"I am excited to push the theory of optimization and algorithm design to new heights!"
What is your research area within MS&E?
My research interests lie primarily in the areas of optimization and algorithm design.
The goal of my research is to build new provably efficient methods for extracting information from large data-sets. I work both on classic pervasive problems in this area, as well as newer problems motivated by recent, practical constraints that arise when attempting to efficiently process at data scale.
I am particularly interested in the intersection of continuous optimization, graph theory, numerical linear algebra, and data structures.
If you have a lot of data, know what you want to compute, and want to try to speed it up, I am happy to try to help.
What made you decide to be a professor? Why Stanford?
I want to be a professor to teach optimization and algorithm design and push these areas to new heights! These are exciting times to be studying optimization and algorithm design!
The last decade has seen numerous breakthroughs in long-standing open problems related to processing large data sets. Many of these breakthroughs stem from carefully combining techniques across a broad range of disciplines.
I believe that Stanford is a wonderful environment for continuing this tradition, as well as a great opportunity for working with a broad, algorithmic community.
How did you choose your field of research?
I have always been drawn to open mathematical problems, and I believe that my field is an exciting domain where new mathematical insights could influence practice.
I started my research by attempting to make progress on a few fundamental problems in graph optimization. This work opened my eyes to the much larger landscape of research I currently pursue.
Who has influenced your work? How?
There is a long list of researchers whom I have been fortunate to meet and work with over the years.
My adviser, Jonathan Kelner, his advisor, Daniel Spielman, and a collaborator of his, Shang-Hua Teng, are some particularly strong influences on this list.
Jon's enthusiastic and bold approach to research catapulted me forward and provided me with an academic framework I aspire to. Dan and Shang-Hua wrote some of the papers that have had, perhaps, the biggest influence on my work, providing cornerstones for tying together continuous optimization with combinatorial structures.
What project are you currently working on?
With a group of researchers at MIT and Georgia Tech, we are currently working on a project to obtain provably faster simulations of random walks. We have a very recent result showing that, for a broad class of random processes, we can compute the limiting behavior faster than the previous best-known bounds.
The result leverages techniques from a range of disciplines, and adds new approaches in a way that we hope will serve as a springboard for more work in this area.
What advice do you have for new MS&E students?
I think optimization is an exciting area undergoing rapid changes with abundant future potential, especially when considering rapidly growing data sets and the increasing popularity of machine learning and new mathematical tools.
I would suggest you be bold and have fun pushing the boundaries further.
What courses will you be teaching?
I will be teaching a range of courses in optimization and efficient algorithm design.
In Winter Quarter, I will be teaching MS&E 313: Almost Linear Time Graph Algorithms: a course on the theory of optimization, covering the mathematical foundation of provably algorithm design for continuous optimization problems.
In Spring Quarter, I will be teaching MS&E 213: Introduction to Optimization Theory: a course on how to solve a broad range of graph optimization problems in almost linear time.
Can you share any recent publications or papers with us?
Yes, here are four recently published papers:
- Subquadratic Submodular Function Minimization
- Approximating Cycles in Directed Graphs: Fast Algorithms for Girth and Roundtrip Spanners
- Almost-Linear-Time Algorithms for Markov Chains and New Spectral Primitives for Directed Graphs
- Accelerated Methods for Non-Convex Optimization
I could tell a long story about each one, but maybe we'll save that for another time!