An Introduction to Computational Complexity Via Games

Document Type

Presentation Abstract

Presentation Date

4-11-2022

Abstract

In this talk I will discuss an approach to solving the famous P=NP question using games. For those unfamiliar with computational complexity, I will describe the complexity classes P and NP, as well as a few other complexity classes, including coNP, L and NL. I will then describe the million-dollar problem that asks whether P=NP and show how one can use a classic two-person combinatorial game, known as an Ehrenfeucht-Fraisse game (along with its relatives), to try to separate complexity classes. I will give some simple examples of how these games are played and then describe a newly rediscovered game that my colleagues and I at IBM are exploring that are potentially more powerful than these classical games.

About the Speaker: Jon is a member of the research staff at the IBM T.J. Watson Research Center in New York. Jon has been with IBM for the last 25 years. Along with several colleagues, he developed the strategy component of the IBM Watson Jeopardy-playing system that in 2011 defeated the two most successful human Jeopardy players on live television. He has built two commercial robots and worked with the Toronto Raptors of the National Basketball Association on a system to help with trades and draft picks. From 2016-2018 Jon was the chief scientist of IBM’s two African research labs, one in Nairobi, Kenya, and the other in Johannesburg, South Africa. Since returning from Africa, Jon’s work has focused on applications of mathematical logic to theoretical questions in computer science, like the P=NP question.

This is an in-person talk also available via Zoom.

Additional Details

April 11, 2022 at 3:00 p.m. Math 103 & Zoom

This document is currently not available here.

Share

COinS