Project B07
Statistics for population models with stochastic (partial) delay differential equations

Project description

This project examines stochastic delay differential equations in population genetics and ecology, focusing on dormancy in microbial populations. It aims to develop statistical methods for estimating and testing key parameters, including delay presence, in Wright-Fisher type equations with time delay. The challenge is to analyze estimators and test statistics for these complex processes, addressing both parametric and nonparametric inference approaches with an emphasis on statistical efficiency. 

Principal Investigators

Markus Reiß (HU Berlin)

Maite Wilke Berenguer (HU Berlin)

Project members

Sascha Gaudlitz (HU Berlin)

Publications

• Pasemann, G., & Reiß, M., 2026. Information bounds for inference in stochastic evolution equations observed under noise. SIAM Journal on Mathematical Analysis, 58(2), 1485–1529. https://doi.org/10.1137/25M1761884 

• Dexheimer, N., Gaudlitz, S., & Schmidt-Hieber, J., 2026. Spike-timing-dependent Hebbian learning as noisy gradient descent. Advances in Neural Information Processing Systems, 38, 106580–106620. https://proceedings.neurips.cc/paper_files/paper/2025/hash/994a9663c3bef2439e605acf340f3bef-Abstract-Conference.html 

Preprints

• Gaudlitz, S., & Gökalp, H. M., 2026. Parameter estimation in a fully coupled partially observed Ornstein-Uhlenbeck process. https://doi.org/10.48550/ARXIV.2606.30443 

• Altmeyer, R, Gaudlitz, S., 2025. Nonparametric Bayesian Inference for Stochastic Reaction-Diffusion Equations. https://doi.org/10.48550/ARXIV.2507.06857