Project description

See the project report.

Repository structure

All python scripts can be found in the scripts folder:

• DoubleWell_model.py sets up the model, calculates the pullback and implements the integration scheme.
• It is called by AMS.py and MC.py which run the AMS-algorithm and straight-up Monte-Carlo simulations, respectively.
• The available score functions for AMS are specified in Score_functions.py. The options are PB-score and simple(~x) score.

Outputs from the scripts are saved as following:

• Monte-Carlo and AMS results are stored in the simulations folder as txt-files. They contain all necessary information, such as model parameters, runtimes and estimated probabilities.
• Figures are saved to the plots folder as png-files.

Collaborate

git clone git@github.com:fabian-lnsm/Pullback-score-AMS.git

References

[1] Jacques-Dumas, V., van Westen, R. M., Bouchet, F., and Dijkstra, H. A.: Data-driven methods to estimate the committor function in conceptual ocean models, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2022-1362, 2022.
[2] Val´erian Jacques-Dumas, Ren´e M van Westen, and Henk A Dijkstra. Estimation of amoc transition probabilities using a machine learning–based rare- event algorithm. Artificial Intelligence for the Earth Systems, 3(4):e240002, 2024.
[3] Peter Kloeden and Meihua Yang. An introduction to nonautonomous dynamical systems and their at- tractors, volume 21. World Scientific, 2020
[4] Pascal Wang, Daniele Castellana, and Henk Dijk- stra. Improvements to the use of the trajectory- adaptive multilevel sampling algorithm for the study of rare events. Nonlinear Processes in Geophysics Discussions, 2020:1–24, 2020