Latest news
New publications, emerging research, software, workshops, and other developments—kept separate from my speaking engagements and organizational activities.
News archive
From September 2026, I will serve on Utrecht University’s Mathematical Institute Advisory Committee, contributing to advice on research, education, staff policy, workload, finance, and institutional collaboration.
The paper introduces a consistent dimensionality-reduction method for stochastic filtering in reaction networks, combining reduced-variance particle estimation with low-dimensional filtering equations to improve efficiency in large systems.
A new preprint introduces exponential damping into the SWIFT framework, enabling direct frequency-domain coefficient computation, sharper truncation rules, and improved accuracy with fewer Fourier coefficients.
At MCQMC 2026 in Edinburgh, I co-organized a minisymposium on Monte Carlo methods for stochastic reaction networks and presented Fourier–RQMC work on multivariate shortfall risk. At the Bachelier World Congress in Bologna, I co-organized minisymposia on energy markets and climate finance and on transform methods, and presented work on quasi-Monte Carlo domain transformations for multi-asset option pricing.
A four-day Lorentz Center workshop bringing together mathematical finance, machine learning, sustainable finance, and FinTech through mini-courses, invited talks, industry sessions, and collaborative discussion.
A new preprint develops computational methods for intraday electricity trading, connecting stochastic optimal control with data-driven decision-making for renewable energy producers.
This work combines path signatures with multi-head attention to learn parameters in fractional Brownian motion-driven stochastic differential equations, balancing estimation accuracy with model complexity.
This preprint develops single- and multi-level randomized quasi-Monte Carlo methods for systemic risk measurement and capital allocation before aggregation.
The study investigates data-driven alternatives to expensive hydrodynamic simulations. Weighted losses, convolutional layers, and look-back windows improve the representation of rare coastal extremes.
Related pages