Chang Wang

We study the regulation of algorithmic (non-)collusion amongst sellers by auditing their data as introduced by Hartline et al. [2024]. We develop an auditing method that tests whether a seller’s pessimistic calibrated regret is low. This method relaxes the previous requirement that a pricing algorithm must use fully-supported price distributions to be auditable. Additionally, we strengthen the justification for using vanishing calibrated regret, versus vanishing best-in-hindsight regret, as the non-collusion definition, by showing that even without side information, the pricing algorithms that only satisfy weaker vanishing best-in-hindsight regret allow an opponent to manipulate them into posting supra-competitive prices.

We provide a beyond-worst-case study on the framework of a dynamic decision-making scenario with resource constraints. We first show a condition which, if violated, implies that the worst-case regret is unavoidable. We then provide an algorithm that achieves logarithmic regret rates under the condition via the re-solving heuristic. Our new algorithm still achieves near-optimal regret in the worst-case.

We study multiple budget-control mechanisms under the scenario where the buyers can report their value distributions to the seller. We prove that well-studied mechanisms are strategically equivalent under mild conditions. En route, we apply the bid-discount method to first-price auctions and establish desired properties of the resulting new mechanism.

We establish lower bounds and upper bounds on the the performance of online throttling in repeated second-price auctions for stochastic/adversarial buyer values and full/partial information structures. We also compare throttling with pacing and establish that throttling is the asymptotically optimal dynamic bidding strategy when the buyer’s values are adversarial.

I wrote this book with my high school chemistry teacher in the summer after I graduated from high school. This is a textbook (in Chinese) that reviews important concepts and teaches a lot of problem-solving techniques for Chinese Chemistry Olympiad.

A textbook (in Chinese) designed for sophomore’s honors discrete mathematics course that concentrates on useful techniques in computer science research and features in conciseness and readability. By-product of TAing the discrete math course at PKU.