Dr Maximilian Gelbrecht

Postdoctoral Researcher

Contact

Potsdam Institute for Climate Impact Research (PIK)
Gelbrecht[at]pik-potsdam.de
P.O. Box 60 12 03
14412 Potsdam

PostDoc at PIK and TU Munich in the Group on Artificial Intelligence in the Anthropocene

Former PhD Student in RD IV / HU Berlin IRTG 1740 

 


I'm Max, a scientist passioniate about exploring machine learning for nonlinear dynamics and Earth System Sciences.
Currently, I am mainly working on making Earth system models differentiable and GPU-compatible to enable seamless integration with ML.

Checkout my two main projects at the moment:

  • ☁️ SpeedyWeather.jl: A global, spectral atmospheric model with an everything flexible attitude
  • 🌱 Terrarium.jl: Modelling land and ecosystems across temporal and spatial scales

If you want to learn something about ML for dynamical systems, checkout my lecture notes:
🤖 Modelling and machine learning for dynamical systems in Julia

To stay up to date follow me on GitHub

  • Moses, Cheng, Churavy, Gelbrecht et. al, DJ4Earth: Differentiable, and Performance-portable Earth System Modeling via Program Transformations, (under review at JAMES)
  • Gelbrecht, M., Klöwer, M., & Boers, N. (2025). PseudospectralNet: Towards hybrid atmospheric models for climate simulations (JAMES)

  • Klöwer, M., Gelbrecht, M., Hotta, D., Willmert, J., Silvestri, S., Wagner, G. L., ... & Hill, C. (2024). SpeedyWeather. jl: Reinventing atmospheric general circulation models towards interactivity and extensibility. Journal of Open Source Software, 9(98), 6323.

  • Gelbrecht, M., White, A., Bathiany, S., & Boers, N. (2023). Differentiable programming for Earth system modeling. Geoscientific Model Development, 16(11), 3123-3135.

  • White, A., Kilbertus, N., Gelbrecht, M., & Boers, N. (2023). Stabilized neural differential equations for learning dynamics with explicit constraints. Advances in Neural Information Processing Systems, 36, 12929-12950.

  • Kraemer, K. H., Gelbrecht, M., Pavithran, I., Sujith, R. I., & Marwan, N. (2022). Optimal state space reconstruction via Monte Carlo decision tree search. Nonlinear Dynamics, 108(2), 1525-1545.

  • Maximilian Gelbrecht, Niklas Boers, Jürgen Kurths: "Neural Partial Differential Equations for Chaotic Systems",  In: New Journal of Physics (2021) https://doi.org/10.1088/1367-2630/abeb90 (Open Access)
  • Maximilian Gelbrecht, Valerio Lucarini, Niklas Boers, and Jürgen Kurths. “Analysis of a bistable climate toy model with physics-based machine learning methods”.  arXiv:2011.12227
  • Nico Wunderling, Maximilian Gelbrecht, Ricarda Winkelmann, Jürgen Kurths, and Jonathon F Donges. “Basin stability and limit cycles in a conceptual model for climate tipping cascades”. In: New Journal of Physics (2020) (Open Access)
  • Maximilian Gelbrecht, Jürgen Kurths, and Frank Hellmann. “Monte Carlo basin bifurcation analysis”. In: New Journal of Physics 22.3 (Mar. 2020), p. 033032 (https://github.com/maximilian-gelbrecht/MCBB.jl) (Open Access)
  • Maximilian Gelbrecht, Niklas Boers, and Jürgen Kurths. “Phase coherence between precipitation in South America and Rossby waves”. In: Science Advances 4.12 (2018). (Open Access)
  • Maximilian Gelbrecht, Niklas Boers, and Jürgen Kurths. “A complex network representation of wind flows”. In: Chaos: An Interdisciplinary Journal of Nonlinear Science 27.3 (2017), p. 035808