Amortized and Rigorous Optimality Certificates for AC Optimal Power Flow via Load-Affine Dual Libraries

Proving an alternating-current optimal power flow solution globally ε-optimal usually takes a convex-relaxation solve per instance. Most of these solves can be reused. For a fixed multiplier, the box-constrained Lagrangian bound of a conic relaxation is affine in the load, so a multiplier computed once gives a lower bound at every load by an inner product. Where the relaxation's prices vary smoothly in few load directions, packing numbers in their number bound the size of a library from above and, under a curvature assumption, from below, and solving the relaxation only where the library falls short stays within the upper bound; where prices are constant within operating regimes, one multiplier per regime suffices. On eight benchmark systems of 14 to 500 buses, one multiplier per regime certifies nearly as many loads as a thousand, and a library used first makes an interior-point-plus-relaxation pipeline 3.0 to 15.2 times faster at ε = 0.5%, depending on the baseline configuration, and 8.9 times while proving both bounds. Grown this way over a year of hourly loads of a modified 2000-bus Texas grid, a library certified 98.2% of the hours with 252 solves.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-05
DOI
https://doi.org/10.5281/zenodo.23987312
Primary Topic
Optimal Power Flow Distribution
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Amortized and Rigorous Optimality Certificates for AC Optimal Power Flow via Load-Affine Dual Libraries

Yuan Shen
Zenodo (CERN European Organization for Nuclear Research)
Optimal Power Flow Distribution
preprint

Amortized and Rigorous Optimality Certificates for AC Optimal Power Flow via Load-Affine Dual Libraries

Yuan Shen
preprint en

Abstract

Proving an alternating-current optimal power flow solution globally ε-optimal usually takes a convex-relaxation solve per instance. Most of these solves can be reused. For a fixed multiplier, the box-constrained Lagrangian bound of a conic relaxation is affine in the load, so a multiplier computed once gives a lower bound at every load by an inner product. Where the relaxation's prices vary smoothly in few load directions, packing numbers in their number bound the size of a library from above and, under a curvature assumption, from below, and solving the relaxation only where the library falls short stays within the upper bound; where prices are constant within operating regimes, one multiplier per regime suffices. On eight benchmark systems of 14 to 500 buses, one multiplier per regime certifies nearly as many loads as a thousand, and a library used first makes an interior-point-plus-relaxation pipeline 3.0 to 15.2 times faster at ε = 0.5%, depending on the baseline configuration, and 8.9 times while proving both bounds. Grown this way over a year of hourly loads of a modified 2000-bus Texas grid, a library certified 98.2% of the hours with 252 solves.

Zenodo (CERN European Organization for Nuclear Research)
University of British Columbia, Okanagan Campus (CA)
Optimal Power Flow Distribution
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon