NASC-SQUIDS Joint Seminar: Parameter estimation for weakly interacting hypoelliptic diffusions
| Dates: | 13 November 2026 |
|---|
| Times: | 12:00 - 13:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Department of Mathematics |
|---|
| Who is it for: | University staff, Current University students |
|---|
| Speaker: | Yuga Iguchi |
|---|
|
|
Speaker: Yuga Iguchi (Lancaster)
Abstract: I will discuss mean-field-type stochastic differential equations (SDEs) and the estimation of their unknown parameters from discretely observed trajectories. These models describe systems of N weakly interacting particles, each governed by an SDE, with a possibly degenerate diffusion matrix. A central challenge is that the transition density, and hence the likelihood, is generally unavailable in closed form. The key idea of this work is to use a carefully designed time discretisation of the interacting particle system to construct a computationally tractable surrogate likelihood, while accounting for the possible degeneracy of the diffusion matrix. I will then present an asymptotic analysis showing that the resulting estimators are consistent and asymptotically normal, providing theoretical guarantees for using the proposed likelihood in practice. If time allows, I will conclude with ongoing work and future directions towards scalable computational approaches for systems with large N interacting particles.
The talk is based on the preprint https://arxiv.org/pdf/2508.04287
This event is sponsored by ProbAI_Hub.
Speaker
Yuga Iguchi
Organisation: University of Lancaster, ProbAI Hub
Travel and Contact Information
Find event
Frank Adams Room 1
Alan Turing Building
Manchester
