missinglpf / cu-powermarket-admm

CUDA implementation of the alternating direction method of multipliers (ADMM) for power market simulation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

cu-powermarket-admm

CUDA implementation of the alternating direction method of multipliers (ADMM) for power market simulation

Background

For details, please see

  • Kraning et al. (2014). Dynamic Network Energy Management via Proximal Message Passing. Foundations and Trends in Optimization, 1(2):70-122.
  • Rintamäki, Tuomas (2015). Demand and wind power scenarios for predicting power prices in a transmission-constrained system

Compilation

The code utilizes CUDA's dynamic parallelism. Run the following commands

nvcc -arch=sm_50 -dc main.cu -o main.o  
nvcc -arch=sm_50 main.o -lcudadevrt -o main

Examples

The input directory includes data files from a real world power system. The power system is modelled with 12 nodes, 18 transmission lines, 12 loads, and 12 generators with piecewise linear cost curves (each curve consists of 19 pieces in total). The example simulates 624 timesteps.

About

CUDA implementation of the alternating direction method of multipliers (ADMM) for power market simulation


Languages

Language:Cuda 100.0%