The usage of neural networks to derive physical laws without any kind of pre-existing bias is a promising but relatively new field with the long-term goal to construct an artificial intelligence physicist that is able to derive physical laws from experimental data. In this project, a step is taken in the direction of solving complex problems by tackling the double pendulum, the simplest chaotic system. To do so, a neural network architecture, adapted from previous work, is used to find the relevant parameters of the system in multiple configurations of the pendulum. Afterwards, the possibility of a neural network derived general solution of the problem is discussed through the relevant aspects that increase its complexity.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-501808 |
| Date | January 2023 |
| Creators | Fusté Costa, Max |
| Publisher | Uppsala universitet, Högenergifysik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | FYSAST ; FYSPROJ1306 |
Page generated in 0.002 seconds