This seminar will be presented in hybrid mode.  The speaker will deliver his talk in person.

Title: Sequential estimation with control and discretionary stopping

Abstract: We show that “full-bang” control is optimal in a problem that combines features of (i) sequential least-squares estimation with Bayesian updating, for a random quantity observed in white noise; (ii) bounded control of the rate at which observations are received, with a superquadratic cost per unit time; and (iii) “fast” discretionary stopping. We describe also the optimal filtering and stopping rules in this context.

Joint work with Erik Ekstrom and Juozas Vaicenavicius.

The talk will be followed by refreshments in the Huxley Common Room at 5pm.