Current Research
Teaching Pedagogy Research
- The impact of optional groupwork on grades
- Threshold Concepts in an Introductory Numerical Methods Course
PhD Research
My research considered the development of robust and efficient algorithms for performing parameter estimation for models specified by systems of ordinary and delay differential equations. Topics of ongoing interest include:
- parallel methods for approximating model sensitivities
- application of methods to new models and real data
- handling the case of unobserved state variables
- extending the approaches to other classes of differential equations
Accounting for Numerical Uncertainty
The development of numerical methods that report the uncertainty introduced by numerical error is necessary to ensure the results of numerical simulations can be correctly assessed. Ignoring these numerical errors can lead to misleading results.
Neural nets and ODEs
Neural nets can be used to simulate ODEs and ResNet architectures are numerical discretizations of ODEs.