Dataset for MRes Thesis: Probabilistic Operator Learning for Climate Model Parameterisation

This dataset was collated for use in experiments presented in the MRes Thesis "Probabilistic Operator Learning for Climate Model Parameterisation" submitted to the University of Cambridge.All data included was generated by other researchers, this is simply a subset to allow easy reproduction of the experiments contained in the work above.

The data for the Burgers' equation (burgers_data_R10.mat) and the Darcy Flow (piececonst_r421_N1024_smooth1.mat, piececonst_r421_N1024_smooth2.mat) experiments were generated by Lu et al. (2022). Creative Commons Attribution Non Commercial Share Alike 4.0 International applies.

The data for the Helmholtz (Helmholtz_inputs.npy, Helmholtz_outputs.npy) and Navier-Stokes (NavierStokes_inputs.npy, NavierStokes_outputs.npy) experiments were generated by de Hoop et al. (2022). Creative Commons Attribution 4.0 International applies.

References:

1. Lu L, Meng X, Cai S, Mao Z, Goswami S, Zhang Z, et al. A comprehensive and fair comparison of two neural operators (with practical extensions) based on FAIR data. Computer Methods in Applied Mechanics and Engineering. 2022 Apr 1;393:114778.

2. de Hoop MV, Huang DZ, Null EQ, Stuart AM. The Cost-Accuracy Trade-Off in Operator Learning with Neural Networks. JML. 2022 Jun;1(3):299–341.