Kamphof, Samuel (2020) Reservoir-based prediction of convective motions. Bachelor's Thesis, Mathematics.
|
Text
bMATH_2020_KamphofSW.pdf Download (4MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (97kB) |
Abstract
This thesis explores the possibility of solving Burgers' equation with a model-free reservoir predictor: an echo state network. Two types of numerical experiments are performed. One simply trains the network on spatial training data obtained through means of the finite volume method. The other trains a network on the Fourier coefficients by applying the fast Fourier transform on the spatial data. The underlying idea behind this is that due to the diffusion the higher frequency components are less relevant and could be removed from the training data, meaning one would see a decrease in computation time. The training of the Fourier coefficients performs significantly worse than the training of the spatial data when the readout matrix is reservoir focused. But both methods see an increase in performance when the readout matrix is input-focused.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Wubs, F.W. and Grzegorczyk, M.A. |
| Degree programme: | Mathematics |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 13 Jul 2020 13:38 |
| Last Modified: | 13 Jul 2020 13:38 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/22612 |
Actions (login required)
 |
View Item |