Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

Scalable Global Optimization for AC-OPF via Quadratic Convex Relaxation and Branch-and-Bound

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Branching on voltage magnitudes and angle differences turns the loose quadratic convex relaxation of AC-OPF into a near-global certificate on benchmark cases.

desk verdict Standard spatial B&B with QC-relaxation bounds; the empirical lower-bound tightening is real on small cases, but the global-optimality claim and the pseudocode do not hold up. read the letter →

arxiv 2505.18435 v1 pith:VZFVBXHC submitted 2025-05-24 math.OC

Mohammadreza Iranpour , Mohammad Rasoul Narimani This is my paper · ORCID
classification math.OC MSC 90C2690C30
keywords optimalpowerflowAC-OPFquadraticconvexrelaxationbranch-and-boundglobaloptimizationenvelopeslowerbounds
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes a hybrid scheme for the alternating-current optimal power flow problem (AC-OPF): a branch-and-bound search that splits voltage magnitudes and angle differences into smaller intervals, with a quadratic convex (QC) relaxation solved in each subregion to obtain a lower bound. Subregions whose lower bound already exceeds the cost of a feasible AC solution are pruned, and the surviving regions are split again for a fixed number of levels. The paper claims that this shrinks the optimality gap from 22.96 percent to 0.01 percent on case30_ieee and to zero or near zero on several other benchmark cases, while keeping the number of subproblems in the thousands rather than an exponential explosion. If these bounds hold, the method gives power-system operators a practical way to certify that a fast local solution is globally near-optimal.

What carries the argument

The central object is the quadratic convex (QC) relaxation of the AC power flow equations, in which squared voltage magnitudes, bilinear voltage products, and sine and cosine terms are replaced by convex envelopes. The branch-and-bound step splits the interval of one selected variable, either a voltage magnitude or an angle difference, at its midpoint into two child regions, solves the QC relaxation on each child to obtain a valid lower bound, and prunes children whose lower bound exceeds the feasible AC-OPF upper bound. This converts a single loose global relaxation into many smaller, tighter relaxations whose minimum surviving lower bound is reported as the BB-QC bound.

What would settle it

Take a benchmark with a known certified global optimum, run Algorithm 1 as written, and compare its returned QC lower bound against that global optimum: if the returned lower bound exceeds the global optimum, or if a region containing the optimum was pruned, the claim that the method preserves global optimality fails. A direct check is to enumerate the kept child boxes after the final level and verify that their union, together with the pruned boxes, covers the original voltage and angle box.

Watch

Extended reading notes

Core claim

The paper's central claim is that a branch-and-bound process built on the QC relaxation can reduce the number of subproblems explored while preserving the potential to reach the global optimum. The key reported effect is that branching on voltage-related variables makes the relaxation's lower bound much tighter: the paper's tables show the optimality gap dropping from 22.96 percent to 0.01 percent on case30_ieee, from 23.69 percent to 3.89 percent on case14_ieee_sad, and to zero on several other benchmark cases, while the number of child nodes solved stays between roughly a dozen and thirty thousand.

Load-bearing premise

The method's reported optimality gaps are valid only if, after the fixed number of branching levels, the kept child regions cover the whole feasible set except regions that were safely pruned; the paper does not specify a branching-variable rule or prove this coverage, and its pseudocode includes a pruning condition that is unreachable for a valid relaxation.

Editorial extensions

If this is right

  • The reported BB-QC gaps imply that a local AC-OPF solution can be certified as within 0.01 percent of global optimality on case30_ieee, where the plain QC relaxation leaves a 22.96 percent gap.
  • Because the tree depth is fixed and only one variable splits per level, the number of subproblems stays manageable, suggesting the approach can scale to systems where full branch-and-bound is impractical.
  • The improvement relies only on refining voltage-related bounds, which the paper argues concentrates the search on the variables that drive OPF's non-convexity.
  • The lower-bound tightening can be separated from the upper-bound search, so the method can be combined with any feasible AC-OPF solver to produce a near-global certificate.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The fixed-depth tree is better interpreted as a lower-bound improvement heuristic than as a complete global-optimality certificate: a rigorous certificate would require either full coverage of unpruned regions or an explicit bound on what was discarded, and the paper's 'potential to reach the global optimum' language leaves this open.
  • The pseudocode's prune condition that rejects a child whose QC objective is below the root QC lower bound is unreachable for a valid relaxation, since a child relaxation can only be tighter and therefore have a lower bound at least as large as the root's; this suggests the documented pruning logic may not reliably produce the reported gaps as valid global lower bounds.
  • If the observed lower-bound tightening is real, the same splitting scheme could be tested on other non-convex power-system variables, such as transformer tap ratios or line-switching decisions, where convex relaxations also suffer from loose bounds.
  • The paper's histogram diagnostics could be turned into an adaptive branching rule: choose the variable whose split produces the largest rightward shift in normalized child costs, which the paper's own analysis suggests would accelerate convergence.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes a branch-and-bound assisted Quadratic Convex (QC) relaxation framework for AC-OPF. The idea is to start from the standard QC relaxation, obtain an AC-feasible upper bound from a local solver, and then repeatedly bisect one voltage magnitude or angle-difference variable at each level, solving the QC relaxation on each child subregion and pruning children whose relaxed objective appears too high. The authors report on PGLib-OPF benchmark cases that the method reduces the optimality gap from the root QC relaxation to very small values (e.g., from 22.96% to 0.01% for case30_ieee) while exploring only a modest number of subproblems. The central claim is that the approach preserves the potential to reach the global optimum and provides a much tighter lower bound than the root QC relaxation.

Significance. If the algorithmic guarantees claimed in the abstract were correct, this would be a meaningful contribution: it would show that a QC-based spatial branch-and-bound can systematically tighten loose relaxations on public PGLib benchmarks at far lower cost than SDP-based branch-and-bound. The paper has notable strengths: it is grounded in public benchmark instances, uses the standard PowerModels.jl framework, and the QC relaxation is taken from the literature without fitting parameters to the reported gaps. However, the central lower-bounding claim is not established. The algorithm as written terminates at fixed depth without a coverage certificate, the pruning logic contains an impossible condition, the returned lower bound is never updated, and the reported child objective values are relaxed quantities, not feasible AC-OPF solutions. The empirical gaps are therefore not certified global optimality gaps; at best, the paper reports a heuristic observation about some child relaxation values.

major comments (5)
  1. [Algorithm 1 and Section IV-A, Table I] The algorithm terminates after a fixed number of levels n (Algorithm 1, line 7) without any certificate that the union of the kept child regions and the pruned regions covers the original QC-feasible set. The paper never states or proves a coverage invariant. Consequently, the BB-QC gap reported in Table I (e.g., 0.01% for case30_ieee vs. a 22.96% root QC gap) is not a certified global lower bound for the original AC-OPF problem. The abstract's claim of 'preserving the potential to reach the global optimum' is therefore unsupported by the described algorithm.
  2. [Algorithm 1, line 19] Line 19 prunes a child when fchild < QC_lower_bound. For a valid relaxation, restricting a subregion can only increase or leave unchanged the optimal value of the relaxed problem relative to the root relaxation; the relaxed objective cannot fall below the root lower bound in exact arithmetic. This condition is therefore impossible for a correctly implemented relaxation, indicating that the pseudocode does not correspond to the implemented algorithm. If the condition is implemented literally, it would discard valid regions and break the coverage property needed for any global lower bound.
  3. [Algorithm 1, lines 3, 7-30] The variable QC_lower_bound is initialized at line 3 and never updated inside the loop. The returned lower bound is therefore the root relaxation value, not the minimum lower bound over active children. Section IV-A reports BB-QC gaps far smaller than the root QC gap (e.g., 0.01% vs. 22.96% for case30_ieee), so those numbers cannot be derived from Algorithm 1 as written. Either the pseudocode omits the essential update of the lower bound to the minimum over child nodes, or the reported results come from a different computation. The empirical central claim is consequently unreproducible.
  4. [Algorithm 1, lines 23-24; Section III-B] The algorithm updates 'best_solution' whenever fchild is lower than the current best, where fchild is the optimal value of the QC relaxation over the child region. The QC relaxation is a lower-bounding problem; its optimal value is not in general attainable by any AC-feasible point, and Section III-B itself states that the subproblem is solved 'to find a local lower bound' on the objective function. Thus the returned 'Best feasible solution' (Algorithm 1, Ensure) is not shown to be feasible for the original AC-OPF. This invalidates the interpretation of child objective values as candidate feasible solutions in Figures 6-12 and Section IV-D.
  5. [Section III-B, branching strategy] The branching variable selection rule is never specified. Algorithm 1 says 'Select variable v_l in V to split' without defining the rule, and the text gives no criterion beyond 'one variable to divide.' The numerical results in Table I and Figures 7-12 depend on this choice, so the experiments are not reproducible. A precise rule, such as largest bound range, most nonconvexity, or a fixed default with tie-breaking, must be stated.
minor comments (5)
  1. [Section III-B] There are several typos and placeholder references, including 'teh problem', '[]' placeholders, and 'P BA' / 's.t. l=i' formatting issues in Section II. These should be corrected.
  2. [Figures 8-10] The captions for Figures 8, 9, and 10 describe 'three branching levels' even though the horizontal axes show 14, 30, and 57 levels, respectively. The captions should match the actual number of levels plotted.
  3. [Equation (8)] The definition of the optimality gap in Equation (8) uses 'Local Solution' and 'Lower Bound' without clarifying which values are used for the 'BB-QC' gaps in Table I. Since the lower bound is not updated in Algorithm 1, the reader cannot tell what quantity the reported 'BB-QC gap' represents.
  4. [Algorithm 1, lines 11-14] The midpoint computation uses 'lower_bound' and 'upper_bound' without an index, which is ambiguous when multiple variables have different bounds. Clarify that these refer to the bounds of the selected variable v_l.
  5. [Section IV, implementation details] The implementation states 'Gurobi 8.0' as the solver; if a newer version was used, that should be corrected. The paper should also report solver tolerances and any feasibility tolerances used when solving the AC-OPF upper-bound subproblem.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the reported bounds are standard QC relaxations over PGLib subregions, and the author self-citations are background or illustrative only.

full rationale

I walked the paper's derivation chain and found no step in which a claimed result is equivalent to an input by construction. The central computation is: AC-OPF (Eq. 1) is relaxed to the standard QC formulation (Eq. 7), then the B&B procedure bisects voltage-magnitude or angle-difference intervals and solves the same QC relaxation on each subregion. The lower bounds reported in Table I are therefore obtained by solving an independent, externally defined relaxation on public PGLib benchmark cases; no parameter is fitted to the reported gaps, and no equation is defined in terms of the BB-QC gap. The author self-citations are not load-bearing: reference [16] is cited only as background for QC relaxation performance, and reference [24] supplies the three-bus example used for visualization. The empty citation 'in []' near the three-bus discussion is a reference typo, not a circular argument. The paper does contain serious correctness concerns, such as fixed-depth termination without an explicit coverage certificate, a pruning condition in Algorithm 1 line 19 ('if fchild < QC_lower_bound') that is impossible for a valid relaxation, and the use of QC child objective values as if they were feasible AC solutions when updating 'best_solution'. These are validity and reproducibility issues, not circularity: they do not make the output equal to the input by definition, and they do not involve fitting a parameter to the target result.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central computations depend on standard QC relaxation theory and on algorithmic choices that are not fully specified. The main free parameters are the fixed branching depth, the unspecified branching rule, and the externally computed AC-OPF upper bound. No new physical or mathematical entities are introduced.

free parameters (3)
  • number of branching levels n = 6 to 71 depending on the case
    Algorithm 1 terminates at a fixed depth n chosen per test case. This choice directly determines the reported BB-QC gap and runtime, and there is no stopping rule based on the optimality gap.
  • branching variable selection rule = not specified
    Algorithm 1 line 10 says 'Select variable v_l in V to split' without defining how the variable is chosen. The experiments mention voltage-based decisions but not which bus or angle is selected, and this choice materially affects the pruning efficiency and the final lower bound.
  • AC-OPF upper bound from an unspecified local solver = depends on the solver and tolerance
    The algorithm uses a feasible AC-OPF solution as the pruning upper bound, but the paper does not state which local solver, convergence tolerance, or feasibility checks were used. A non-global or slightly infeasible upper bound would change the pruning decisions and the reported gaps.
assumptions (3)
  • domain assumption The QC relaxation provides a valid lower bound for AC-OPF over any bound-restricted subregion.
    The branch-and-bound pruning relies on the QC relaxation being a valid convex relaxation of the AC power flow equations, as established in the cited literature [19].
  • domain assumption The AC-OPF solution used as the upper bound is feasible and valid.
    The algorithm prunes any region whose QC lower bound exceeds this AC-OPF value. If the upper bound is not feasible or is not a true upper bound, the pruning could remove regions containing the global optimum. The paper does not specify the solver or feasibility verification.
  • domain assumption The union of kept child regions after n levels covers all regions not yet pruned.
    The minimum child lower bound is reported as the BB-QC lower bound, which is valid only if the kept children cover the remaining feasible space. The algorithm provides no proof of coverage, especially since the branching variable selection rule is unspecified and some children are discarded without storing their bounds.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Scalable Global Optimization for AC-OPF via Quadratic Convex Relaxation and Branch-and-Bound." pith.science (2026). https://pith.science/paper/VZFVBXHC

@misc{pith2026250518435,
  author       = {Pith},
  title        = {Pith review of: Scalable Global Optimization for AC-OPF via Quadratic Convex Relaxation and Branch-and-Bound},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VZFVBXHC}},
  note         = {Machine review of arXiv:2505.18435}
}
read the original abstract

The Optimal Power Flow (OPF) problem is central to the reliable and efficient operation of power systems, yet its non-convex nature poses significant challenges for finding globally optimal solutions. While convex relaxation techniques such as Quadratic Convex (QC) relaxation have shown promise in providing tight lower bounds, they typically do not guarantee global optimality. Conversely, global optimization methods like the Branch and Bound (B\&B) algorithm can ensure optimality but often suffer from high computational costs due to the large search space involved. This paper proposes a novel B\&B-assisted QC relaxation framework for solving the AC-OPF problem that leverages the strengths of both approaches. The method systematically partitions the domains of key OPF variables, specifically, voltage magnitudes and voltage angle differences, into two equal subintervals at each iteration. The QC relaxation is then applied to each subregion to compute a valid lower bound. These bounds are compared against an upper bound obtained from a feasible AC-OPF solution identified at the outset. Subregions that yield lower bounds exceeding the upper bound are pruned from the search, eliminating non-promising portions of the feasible space. By integrating the efficiency of the QC relaxation with the global search structure of the B\&B algorithm, the proposed method significantly reduces the number of subproblems explored while preserving the potential to reach the global optimum. The algorithm is implemented using the PowerModels.jl package and evaluated on a range of PGLib-OPF benchmark cases. Results demonstrate that this hybrid strategy improves computational tractability and solution quality, particularly for large OPF instances.

Figures

Figures reproduced from arXiv: 2505.18435 by the authors.

Figure 1
Figure 1. Overview of the BB-QC-OPF framework. The flowchart illustrates how the Branch-and-Bound algorithm iteratively splits a single variable at each level, solving the QC-OPF problem within updated bounds for each child node. The figure highlights how variables’ bounds are inherited from parent to child nodes throughout the branching process. proposed method are outlined in Algorithm 1, and its overall process is illustra… view at source ↗
Figure 2
Figure 2. Feasible space of the cyclic three bus system from [24], with 0.9 ă V1,V2,V3 ă 1.1 and ´2π ă ∆θ1,∆θ2, and ∆θ3 ă 2π. are useful for testing because they are known to be challenging, there is often a big difference between the best-known solutions and the lower bounds from convex relaxation approaches. We implemented our method using the Julia v1.11.5, JuMP v1.25.0 [26], the PowerModels package [27], and the Gurobi 8.… view at source ↗
Figure 5
Figure 5. Feasible space of cyclic three bus system from [24], with 0.9 ă V1 ă 1, 0.9 ă V2 ă 1, 1 ă V3 ă 1.1, and ´2π ă ∆θ1,∆θ2, and ∆θ3 ă 2π. Algorithm 1 Branch-and-Bound for Optimal Power Flow with QC Relaxation Require: Network data N, QC-OPF bounds lower_bound, upper_bound, number of levels n, set of variables V Ensure: Best feasible solution best_solution, QC lower bound QC_lower_bound, AC upper bound AC_upper_bound 1: I… view at source ↗
Figures from the paper (5 more)
Figure 6
Figure 6. Figure 6: Distribution of valid and invalid child solutions across different PGLib-OPF cases. The chart highlights the proportion of valid versus invalid children generated during the Branch-and-Bound process. of dimensionality, particularly in larger test cases. However, by app…
Figure 8
Figure 8. Figure 8: Candle chart illustrating the objective value distribution across three branching levels for “Case14_ieee”. The chart includes the floor (best child objective), roof (worst valid child objective), upper shadow (gap to QC-OPF), and average value (mean of child objective…
Figure 10
Figure 10. Figure 10: Candle chart illustrating the objective value distribution across three branching levels for “Case57_ieee”. The chart includes the floor (best child objective), roof (worst valid child objective), upper shadow (gap to QC-OPF), and average value (mean of child objectiv…
Figure 11
Figure 11. Figure 11: Histograms showing the distribution of normalized objective values for child solutions across different test cases. Each chart illustrates how the proposed method improves solution quality relative to the QC and AC objective values, highlighting the progression toward…
Figure 12
Figure 12. Figure 12: Optimality gap distributions for selected PGLib-OPF cases using the proposed QC-assisted Branch-and-Bound method. Candlesticks show solution spread and relaxation strength relative to the AC-OPF benchmark. performance; rather, each branching step incrementally tighten…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Solving Three-phase AC Infeasibility Analysis to Near-zero Optimality Gap

    eess.SY 2025-08 conditional novelty 5.0 of 10

    A presolved bilinear reformulation of three-phase infeasibility analysis achieves certified near-zero optimality gaps on large distribution feeders.

Reference graph

Works this paper leans on

28 extracted references · 26 canonical work pages · cited by 1 Pith paper

  1. [14]

    Global optimization of optimal power flow using a quadratic convex relaxation,

    M. Barati et al. , “Global optimization of optimal power flow using a quadratic convex relaxation,” IEEE Transactions on Power Systems , vol. 35, no. 5, pp. 3924–3936, 2020

    work page 2020

  2. [17]

    Global optimization of optimal power flow using a branch & bound algorithm,

    A. Gopalakrishnan, A. U. Raghunathan, D. Nikovski, and L. T. Biegler, “Global optimization of optimal power flow using a branch & bound algorithm,” in 2012 50th Annual Allerton Conference on Communication, Control, and Computing (Allerton) . IEEE, 2012, pp. 609–616

    work page 2012

  3. [1]

    A survey of relaxations and approximations of the power flow equations,

    D. K. Molzahn, I. A. Hiskens et al. , “A survey of relaxations and approximations of the power flow equations,” F oundations and Trends® in Electric Energy Systems , vol. 4, no. 1-2, pp. 1–221, 2019

    work page 2019

  4. [2]

    Contribution to the economic dispatch problem,

    J. Carpentier, “Contribution to the economic dispatch problem,” Bulletin de la Societe Francoise des Electriciens , vol. 3, no. 8, pp. 431–447, 1962

    work page 1962

  5. [3]

    Local solutions of the optimal power flow problem,

    W. A. Bukhsh, A. Grothey, K. I. McKinnon, and P. A. Trodden, “Local solutions of the optimal power flow problem,” IEEE Transactions on Power Systems, vol. 28, no. 4, pp. 4780–4788, 2013

    work page 2013

  6. [4]

    Strong np-hardness of ac power flows feasibility,

    D. Bienstock and A. Verma, “Strong np-hardness of ac power flows feasibility,” Operations Research Letters , vol. 47, no. 6, pp. 494–501, 2019

    work page 2019

  7. [5]

    A Review of Selected Optimal Power Flow Literature to 1993. Parts I and II,

    J. Momoh, R. Adapa, and M. El-Hawary, “A Review of Selected Optimal Power Flow Literature to 1993. Parts I and II,” IEEE Trans. Power Syst. , vol. 14, no. 1, pp. 96–111, Feb. 1999

    work page 1993

  8. [6]

    Survey of Approaches to Solving the ACOPF (OPF Paper 4),

    A. Castillo and R. O’Neill, “Survey of Approaches to Solving the ACOPF (OPF Paper 4),” FERC, Tech. Rep., Mar. 2013

    work page 2013

Show all 28 references
  1. [7]

    Strengthening the sdp relaxation of ac power flows with convex envelopes, bound tightening, and valid inequalities,

    C. Coffrin, H. L. Hijazi, and P. Van Hentenryck, “Strengthening the sdp relaxation of ac power flows with convex envelopes, bound tightening, and valid inequalities,” IEEE Transactions on Power Systems , vol. 32, no. 5, pp. 3549–3558, 2016

    2016

  2. [8]

    Convex quadratic relaxations for mixed-integer nonlinear programs in power systems,

    H. Hijazi, C. Coffrin, and P. V . Hentenryck, “Convex quadratic relaxations for mixed-integer nonlinear programs in power systems,” Mathematical Programming Computation, vol. 9, pp. 321–367, 2017

    2017

  3. [9]

    A branching framework for global optimization,

    L. Liberti et al. , “A branching framework for global optimization,” Mathematical Programming, vol. 131, no. 1, pp. 73–97, 2012

    2012

  4. [10]

    Non-convex mixed-integer nonlinear pro- gramming: A survey,

    S. Burer and A. N. Letchford, “Non-convex mixed-integer nonlinear pro- gramming: A survey,” Surveys in Operations Research and Management Science, vol. 17, no. 2, pp. 97–106, 2012

    2012

  5. [11]

    Verifying global optimality of candidate solutions to polynomial optimization problems using a determinant relaxation hierarchy,

    S. Xu, R. Ma, D. K. Molzahn, H. Hijazi, and C. Josz, “Verifying global optimality of candidate solutions to polynomial optimization problems using a determinant relaxation hierarchy,” in 2021 60th IEEE Conference on Decision and Control (CDC) . IEEE, 2021, pp. 3143–3148

    2021

  6. [12]

    Zero duality gap in optimal power flow problem,

    J. Lavaei and S. H. Low, “Zero duality gap in optimal power flow problem,” IEEE Transactions on Power systems , vol. 27, no. 1, pp. 92– 107, 2011

    2011

  7. [13]

    Matrix minor reformulation and socp-based spatial branch-and-cut method for the ac optimal power flow problem,

    B. Kocuk, S. S. Dey, and X. A. Sun, “Matrix minor reformulation and socp-based spatial branch-and-cut method for the ac optimal power flow problem,” Mathematical Programming Computation , vol. 10, no. 4, pp. 557–596, 2018

    2018

  8. [15]

    Correcting optimal transmis- sion switching for ac power flows,

    C. Barrows, S. Blumsack, and P. Hines, “Correcting optimal transmis- sion switching for ac power flows,” in 2014 47th Hawaii International Conference on System Sciences . IEEE, 2014, pp. 2374–2379

    2014

  9. [16]

    Tightening qc relaxations of ac optimal power flow through improved linear convex envelopes,

    M. R. Narimani, D. K. Molzahn, K. R. Davis, and M. L. Crow, “Tightening qc relaxations of ac optimal power flow through improved linear convex envelopes,” IEEE Transactions on Power Systems , 2024

    2024

  10. [18]

    A global solution algorithm for ac optimal power flow through linear constrained quadratic programming,

    M. Barati, “A global solution algorithm for ac optimal power flow through linear constrained quadratic programming,” arXiv preprint arXiv:2406.11899, 2024

    2024 arXiv

  11. [19]

    The qc relaxation: A theoretical and computational study on optimal power flow,

    C. Coffrin, H. L. Hijazi, and P. Van Hentenryck, “The qc relaxation: A theoretical and computational study on optimal power flow,” IEEE Transactions on Power Systems , vol. 31, no. 4, pp. 3008–3018, 2015

    2015

  12. [20]

    Computability of Global Solutions to Factorable Non- convex Programs: Part I–Convex Underestimating Problems,

    G. McCormick, “Computability of Global Solutions to Factorable Non- convex Programs: Part I–Convex Underestimating Problems,” Math. Prog., vol. 10, no. 1, pp. 147–175, 1976

    1976

  13. [21]

    Bound tightening for the alternating current optimal power flow problem,

    C. Chen, A. Atamtürk, and S. S. Oren, “Bound tightening for the alternating current optimal power flow problem,” IEEE Transactions on Power Systems, vol. 31, no. 5, pp. 3729–3736, 2015

    2015

  14. [22]

    A semidefinite optimization-based branch-and-bound algorithm for several reactive opti- mal power flow problems,

    J. Sliwak, M. Anjos, L. Létocart, and E. Traversi, “A semidefinite optimization-based branch-and-bound algorithm for several reactive opti- mal power flow problems,” arXiv preprint arXiv:2103.13648 , 2021

    2021 arXiv

  15. [23]

    Modern optimization models and techniques for electric power systems operation,

    A. Sun and D. T. Phan, “Modern optimization models and techniques for electric power systems operation,” Electr . Power Syst. Oper , pp. 1–22, 2017

    2017

  16. [24]

    Empirical investigation of non-convexities in optimal power flow problems,

    M. R. Narimani, D. K. Molzahn, D. Wu, and M. L. Crow, “Empirical investigation of non-convexities in optimal power flow problems,” in2018 Annual American Control Conference (ACC) . IEEE, 2018, pp. 3847– 3854

    2018

  17. [25]

    The power grid library for benchmarking ac optimal power flow algorithms,

    S. Babaeinejadsarookolaee, A. Birchfield, R. D. Christie, C. Coffrin, C. DeMarco, R. Diao, M. Ferris, S. Fliscounakis, S. Greene, R. Huang et al. , “The power grid library for benchmarking ac optimal power flow algorithms,” arXiv preprint arXiv:1908.02788 , 2019

    1908 arXiv

  18. [26]

    JuMP: A Modeling Language for Mathematical Optimization,

    I. Dunning, J. Huchette, and M. Lubin, “JuMP: A Modeling Language for Mathematical Optimization,” SIAM Rev., vol. 59, no. 2, pp. 259–320, June 2017

    2017

  19. [27]

    PowerModels.jl: An Open-Source Framework for Exploring Power Flow Formulations,

    C. Coffrin, R. Bent, K. Sundar, Y . Ng, and M. Lubin, “PowerModels.jl: An Open-Source Framework for Exploring Power Flow Formulations,” in Power Syst. Comput. Conf. (PSCC) , June 2018

    2018

  20. [28]

    Non-Convex Mixed-Integer Nonlinear Programming: A Survey,

    S. Burer and A. N. Letchford, “Non-Convex Mixed-Integer Nonlinear Programming: A Survey,” Surveys Oper . Res. Manage. Sci., vol. 17, no. 2, pp. 97–106, 2012

    2012

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.