My research connects mathematical modelling, scientific machine learning and numerical verification.
Scientific machine learning
I compare reduced models and learned operators against numerical reference solutions, separating accuracy on familiar data from generalisation to new initial conditions.
Inverse problems and uncertainty
I study learned reconstruction under model mismatch and distribution shift, and ensemble filtering for chaotic dynamics.
Numerical PDEs and fluid dynamics
My work includes finite-element cylinder-flow benchmarks and constrained passive-scalar mixing. Analytical references and convergence studies help distinguish numerical effects from physical conclusions.