Chief Investigator:

Dr. YEUNG Chi Ho

Project Year:

2015-2016

Description:

Optimization problems correspond to the tasks to optimize a set of variables to extremize an objective function. They are at the center of a wide range of applications ranging from timetable scheduling and delivery of goods to sophisticated optimization processes in industry and commerce. Computer scientists and mathematicians have long been deriving optimization algorithms to identify the optimal solution for specific problems, but their conventional methodologies do not allow them to understand why optimization problems become difficult to solve in some parameter regimes. They leave the task to physicists, who developed a fundamental understanding on optimization problems by drawing analogy with physical systems. They identified parameter regimes where solutions exist but are difficult to find, and these understandings lead to new optimization algorithms which work beyond the limit of conventional algorithms. Nevertheless, these findings are not fully recognized by computer scientists and mathematicians as there are still missing connections between the theory and the practical optimization problems. In the proposed research, we will apply statistical physics to improve our understanding on optimization problems, and to apply these findings to derive innovative optimization algorithms readily applicable to a wide range of applications. Both real and state space, small and large systems will be studied.