Solving Backward Stochastic Differential Equations via Neural Network
Dnr:
NAISS 2025/22-794
Type:
NAISS Small Compute
Principal Investigator:
Giulia Pucci
Affiliation:
Kungliga Tekniska högskolan
Start Date:
2025-05-22
End Date:
2026-06-01
Primary Classification:
10106: Probability Theory and Statistics (Statistics with medical aspects at 30118 and with social aspects at 50907)
Webpage:
Allocation
- Dardel-GPU at PDC: 200 GPU-h/month
- Dardel at PDC: 10 x 1000 core-h/month
Abstract
This project aims to develop and implement machine learning algorithms for solving backward stochastic differential equations (BSDEs) with jumps via feedforward neural networks. The primary applications are in contract pricing and decision-making in energy markets, as well as in the planning and deployment of renewable energy infrastructure under uncertainty and market shocks.