Back to search

Cornell University

Parameter estimation and inference for nonlinear dynamical systems

Abstract

dc:description.abstract

Dynamical systems, expressed as differential equations, serve as fundamental tools across various scientific disciplines for describing the temporal or spatial dynamics of real-world processes. Despite the increasing abundance of data from such processes, deriving the underlying physical laws analytically has grown more challenging. Consequently, statisticians and applied mathematicians have turned to developing techniques to estimate the differential equations governing temporally or spatially evolving systems from time series data. This dissertation contributes to this field by presenting novel advancements in both parameter estimation for differential equations of a known functional form and equation discovery for systems of an unknown form. The first method introduced is a Bayesian regression algorithm tailored for fitting a specific system of differential equations to time-course gene expression data. Subsequently, the resulting model is utilized to cluster genes based on the similarity of their temporal behavior. Following this, a regularized regression-based approach is presented for recovering systems of ordinary differential equations from time series data. Notably, this method offers classical uncertainty estimates in the form of confidence intervals and hypothesis tests for each term in the reconstructed equation. This addresses a pertinent issue in the literature on data-driven dynamical system recovery, namely how to quantify uncertainty in a computationally efficient and interpretable manner. Finally, an extension of this methodology to partial differential equations recovered from spatiotemporal data is proposed and contextualized within the broader field of scientific machine learning. Detailed discussions on future extensions of the proposed methods are provided, and software packages implementing these methods are made publicly accessible.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Statistics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Statistics
Grantor
Cornell University
Year dc:date.issued
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Venkatraman, Sara
Chair dc:contributor.chair
  • Wells, Martin
Committee members dc:contributor.committeemember
  • Basu, Sumanta
  • Hooker, Giles

Subjects

dc:subject × 3

Rights

dc:rights
Statement dc:rights
  • Attribution 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
16575403
Dc Identifier Other
ProQuest Submission ID: 14270
ProQuest Publication ID: 31242634

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Venkatraman, Sara. Parameter estimation and inference for nonlinear dynamical systems. Doctor of Philosophy thesis, Cornell University, 2024. https://hdl.handle.net/1813/116020