ExaAdmm.jl

ExaAdmm.jl implements the two-level alternating direction method of multipliers for solving the component-based decomposition of alternating current optimal power flow problems on GPUs.

The package can be installed in the Julia REPL with the command below:

] add ExaAdmm

Running the algorithms on the GPU requires either NVIDIA GPUs with CUDA.jl or KernelAbstractions.jl (KA) with the respective device support (e.g., AMDGPU.jl and ROCKernels.jl). Currently, only the ACOPF problem is supported using KA.

Currently, ExaAdmm.jl supports electrical grid files in the MATLAB format. You can download them from here. Below shows an example of solving case1354pegase.m using ExaAdmm.jl on an NVIDIA GPU

using ExaAdmm

env, mod = solve_acopf(
    "case1354pegase.m";
    rho_pq=1e1,
    rho_va=1e3,
    outer_iterlim=20,
    inner_iterlim=20,
    scale=1e-4,
    tight_factor=0.99,
    use_gpu=true,
    verbose=1
);

and the same example on an AMD GPU:

using ExaAdmm
using AMDGPU

env, mod = solve_acopf(
    "case1354pegase.m";
    rho_pq=1e1,
    rho_va=1e3,
    outer_iterlim=20,
    inner_iterlim=20,
    scale=1e-4,
    tight_factor=0.99,
    use_gpu=true,
    ka_device = ROCBackend(),
    verbose=1
)

The following table shows parameter values we used for solving pegase and ACTIVSg data.

We have used the same tight_factor=0.99, outer_iterlim=20, and inner_iterlim=1000 for all of the above data.

• Youngdae Kim and Kibaek Kim. "Accelerated Computation and Tracking of AC Optimal Power Flow Solutions using GPUs" arXiv preprint arXiv:2110.06879, 2021
• Youngdae Kim, François Pacaud, Kibaek Kim, and Mihai Anitescu. "Leveraging GPU batching for scalable nonlinear programming through massive lagrangian decomposition" arXiv preprint arXiv:2106.14995, 2021

This research was supported by the Exascale ComputingProject (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration. This material is based upon work supported by the U.S. Department of Energy, Office of Science, under contract number DE-AC02-06CH11357.