Master's thesis presentation. Asima is advised by Felix Dietrich.
Previous talks at the SCCS Colloquium
Asima Azmat: Identification of Stochastic Differential Equations with Artificial Neural Network
SCCS Colloquium |
We will identify stochastic differential equations (SDEs) for observables of fine-grained particle simulations in order to coarse-grain their dynamics. Simulations data will be obtained from Vadere for at least two simulations. We will approximate the drift and diffusivity functions in the SDE using a neural network. Our approach works on scattered snapshot data and is designed to handle different time steps per snapshot.