Tutorials

This page hosts the materials for the tutorial "GPU-accelerated optimization for power systems with MadNLP and ExaModels" presented at PowerTech 2025 by Alexis Montoison, François Pacaud and Sungho Shin.

Description

The tutorials present how to solve the AC power flow using GPU-accelerated nonlinear programming. They proceed by order of difficulty, starting by the solution of the power flow equations and end with the solution of large-scale optimal power flow instances.

Data

You can download all the instances we used in the tutorials on this webpage.

Table of contents

The table of contents is:

• Getting started
• • GPU programming using CUDA.jl
  • Modeling with ExaModels.jl
• Tutorial 1: solving the power-flow equations with ExaModels
• • Implementing the power flow equations with ExaModels
  • Solving the power flow equations using the Newton algorithm
• Tutorial 2: solving the power-flow equations in batch on the GPU
• • Batched power flow with ExaModels
  • Solving the power flow equations in batch on the GPU
• Tutorial 3: solving constrained power flow with MadNLP
• • Constrained power flow
  • Solution with the interior-point solver MadNLP
  • Deporting the solution on the GPU
• Tutorial 4: solving optimal power flow with MadNLP
• • Optimal power flow model
  • Solving optimal power flow on the GPU
• Tutorial 5: solving the optimal power flow with ExaModelsPower.jl
• • What is ExaModelsPower.jl?
  • Getting started
  • Static optimal power flow
• Tutorials
• • Description
  • Data
  • Table of contents