SciML for retrieving underlying dynamics from high frequency data
Dnr:
NAISS 2024/22-487
Type:
NAISS Small Compute
Principal Investigator:
Jon Norberg
Affiliation:
Stockholms universitet
Start Date:
2024-04-02
End Date:
2025-05-01
Primary Classification:
10611: Ecology
Webpage:
Allocation
- Centre Storage at NSC: 500 GiB
- Tetralith at NSC: 10 x 1000 core-h/month
Abstract
I have 4 years of hourly data from a in situ plankton camera with AI species detection, i.e. genera level population dynamics of plankton for 4 years. I want to apply SciML methods from the Julia ecosystem to Retreive the underlying dynamics of the system using a neural net training. This first phase is to examine the types of neural-network needed and the sensitivity to simulation parameter choices.