Skip to content

Fokker Planck based Data Assimilation method using Fourier Neural Operators as integrator

Notifications You must be signed in to change notification settings

AdamJamesSheppard/A-step-towards-non-parametic-data-assimilation-FP-DA-

Repository files navigation

Please cite the pdf if used in scientific/industrial applications - this is a preliminary investigation and notes. This is a step towards Fokker-Planck based data assimilation, it is a slow process towards non-parametic data assimilation for real time data fusion for dynamical systems exhibiting stochastic-chaotic behavouir

These codes generate alot of data and require around 15-20gb of RAM if used with current settings, RAM usage varies alot when using MPI - data generated from these algorithms is unbounded. Datasets generated for the purpose of documentation reached 58.6gb

Mathematics required to read pdf - stochastic processes/Ito Calculus/Group Theory/Partial Differential Equations/Familiarity with Markov Processes and Feller-Groups (C_0 semigroups of operators generated by continuous time markov chains)

Ongoing Investigation - Experimental