Winter Quarter Update

I have greatly expanded the functionality of DeferredAcceptance, my school-choice code library. New features include the nonatomic form of the DA algorithm, several new tiebreaking mechanisms, and further replications of recent experimental results from the literature on stable assignments. A couple of interesting research results of the past ten years include a tiebreaking mechanism that lets students name a “target school” where they will receive enhanced admissions priority, and a two-round dynamic reassignment mechanism that uses the spaced freed up by students who leave the market (e.g. to attend private school) to compensate for inequality in the primary assignment round.

I also coded a practice implementation of an interior-point method for solving linear programs. During winter break, our lab has been working through Jorge Nocedal and Stephen Wright’s Numerical Optimization textbook, and the section I was responsible for included the predictor-corrector algorithm given above.

During winter quarter, I completed a course in revenue management and pricing. For my final presentation in that class, I reviewed a paper that concerns which products should be presented to a customer in order to maximize revenue (assortment optimization):

One rewarding aspect of this class was the exposure to a wide range of models in choice theory, which allows us to express mathematically the processes consumers use when deciding which product to buy. So far, in my research in school choice, I have focused on the assignment aspect of the problem, which takes place after students’ preferences are revealed. But robust simulation of school-choice markets requires effective modeling of uncertainty and information asymmetry on the demand side.