Ph.D. work

My current research focuses on studying and solving NP-complete problems. To better understand NP-complete problems and algorithms for solving them, I am trying to identify instance characteristics and find relations between instance characteristics and algorithm performance. My tasks include identifying instance characteristics that correlate with algorithm performance, evaluating the importance of those characteristics in respect to algorithm behavior, building and analyzing algorithm performance predictors. To better solving NP-complete problems, I build algorithm portfolios to solve algorithm selection problems automatically. Such portfolios pick one or more algorithms (with different shares of CPU time) from a set of candidate algorithms using empirical hardness models as algorithm-performance predictors. Currently, I am working on the propositional satisfiability problem (SAT) and the traveling salesman problem (TSP), as exemplars for NP-complete decision and optimization problems. For detailed information, please visit our pages of Empirical Hardness Models, SATzilla and Hydra.

Part of my research is supported by MITACS .

Previous work in computer science

I finished my M.S. degree at University of Nebraska-Lincoln, and my supervisor was Prof. Berthe Y. Choueiry. I worked in the Constraint Systems Laboratory (ConSystLab), and the title of my M.S. thesis was "Consistency Methods for Temporal Reasoning". My work included the design and empirical evaluation of several new algorithms for pre-procechssing and solving temporal CSP. This research has yielded five published papers.

Previous work in chemistry

I got my M.S. degree in chemistry at Nanjing University, China. My research in chemistry focused on preparation and characterization of Oxide Catalyzer. I also participated in many other research projects in the Key Laboratory of Mesoscopic Chemistry of MOE.