Populating agent-based models with agents who give rise to dynamics and scenarios of interest

Code for paper "Gradient-assisted calibration for financial agent-based models"

Code for NeurIPS 2024 paper on "Interventionally consistent surrogates for complex simulation models".

Black-box Bayesian inference for agent-based models

Investigating properties of a random attachment model of network growth, e.g. expected largest degree

Approximate Bayesian Computation with Path Signatures

Amortised likelihood-free inference for expensive time-series simulators with signatured ratio estimation