Department of Mathematics and Statistics

Colloquium for Fall 2018

If you are interested in giving a talk, please email caherrera@csuchico.edu

Colloquia are typically held on Friday in Holt 175.  Refreshments are served at 3:00pm and the talk begins at 3:30pm.  Everyone is invited to attend. 

Friday, Sept. 7 -- Holt 175, 4:00pm

Isaiah Lankham (University of California Office of the President)
Propensity Score Matching: A Cautionary Tale

Abstract: Imagine you've been hired as a data analyst for a pre-college outreach program called 2&Thru, and you're asked to show how the program is effective.  In other words, are high-school graduates who participate more likely to enroll in, persist at, and graduate from college than if they hadn't participated?

From a statistical standpoint, this sounds like an ideal setup for randomized controlled trials (RCTs): Randomly assign students to either a treatment group (those who participate in 2&Thru) or a control group (those who don't), and compare the two groups' outcomes. However, random assignment would be unethical since the outcomes of 2&Thru participation are potentially life-altering. We can't intentionally deny program benefits, so we'll need a better way of estimating the counterfactual, meaning the results 2&Thru participants would have had if they hadn't participated.

Fortunately, there's a clever workaround called propensity score matching (PSM), which allows us to mimic the effects of randomized group assignments. By identifying the factors (aka covariates) most important in determining a student's propensity to self-select for program participation, we can match 2&Thru participants to comparable non-participants using logistic regression and build a quasi-control group. Because the entire technique hinges on selection of the "right" set of covariates, the "wrong" set will quickly lead to invalid conclusions, as will be demonstrated.

Friday, Sept. 14 -- Holt 175, 3:30pm

John Lind (California State University, Chico)
The algebra of spheres

Abstract: A thread can wind around a peg any number of times—and to a topologist this counting of the winding number is what the integers are! In this talk I will explore a generalization of this idea to higher dimensions. We can ask in a similar way how many times a sphere of a given dimension can wrap around another sphere. Contemporary methods that attempt to answer this question are complex and abstract, so I will use pictures to harness our geometric intuition. By the end, I will try to convince you that the patterns we detect in the spheres are shadows of a fundamental object of algebra.

Friday, Sept. 21 -- Holt 175, 3:30pm

Jonathan Sands (University of Vermont)
A potential game-changer for speedy factoring: Shor's algorithm in quantum computing

Abstract: In 1994, Peter Shor showed that factoring could be done in polynomial time if significant quantum computing becomes a reality. This would imply that standard cryptosystems such as RSA are no longer secure. We present Shor's algorithm for a general mathematical audience, focusing on the number theory and requiring no previous knowledge of quantum mechanics.

Past Colloquium Schedules