arxiv: 2604.03807 · v1 · submitted 2026-04-04 · 🧮 math.OC

Recognition: 2 theorem links

· Lean Theorem

Computing Rare Probabilities of Voltage Collapse

Tongtong Jin , Anirudh Subramanyam , D. Adrian Maldonado

Authors on Pith no claims yet

Pith reviewed 2026-05-13 17:19 UTC · model grok-4.3

classification 🧮 math.OC

keywords voltage collapselarge deviation theoryinstantonpower system stabilityrare event probabilitystability boundaryprobability estimation

0 comments

The pith

Large deviation theory locates the single most likely voltage-collapse point and approximates its probability with first- and second-order expansions.

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

The paper establishes a Large Deviation Theory method for computing the tiny probabilities of voltage collapse in power grids. It reduces the problem to locating the instanton, the most probable point where random fluctuations drive the system across the stability boundary, then builds first-order and second-order probability estimates around that point. The second-order version adds the local curvature of the boundary to raise accuracy. A reader would care because these events are too infrequent for direct Monte Carlo sampling yet determine long-term grid reliability, and the approach works for both Gaussian and non-Gaussian uncertainties.

Core claim

We formulate the problem as finding the most probable failure point (the instanton) on the stability boundary and derive both first-order and second-order approximations for the collapse probability. The second-order method incorporates the local curvature of the stability boundary, yielding higher accuracy. This LDT framework generalizes methods based on Mahalanobis distance and is extensible to non-Gaussian uncertainties, with estimates shown to converge to Monte Carlo results in the rare-event regime.

What carries the argument

The instanton, the point on the stability boundary that minimizes the large-deviation rate function of the uncertainty distribution and thereby governs the exponential decay of the collapse probability.

If this is right

The second-order approximation improves accuracy over first-order by capturing boundary curvature.
The estimates converge to Monte Carlo results precisely where direct sampling becomes computationally impossible.
The framework applies to non-Gaussian uncertainty distributions.
It recovers and generalizes existing Mahalanobis-distance methods as a special case.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Operators could identify the most vulnerable operating points simply by inspecting the instanton location.
The method could be embedded in real-time security-assessment tools to flag rising collapse risk without exhaustive scenario sampling.
Similar instanton searches could be tested on dynamic or time-varying stability boundaries in larger networks.

Load-bearing premise

The rare-event probability is dominated by a single most-probable crossing point on the stability boundary, and a local first- or second-order approximation around that point suffices.

What would settle it

Monte Carlo sampling performed on the same power-system model for successively rarer collapse thresholds would cease to approach the LDT estimates instead of converging to them.

Figures

Figures reproduced from arXiv: 2604.03807 by Anirudh Subramanyam, D. Adrian Maldonado, Tongtong Jin.

**Figure 2.** Figure 2: Two-bus probability sweep for Gaussian uncertainty. The reference [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Local geometry at the GMM instanton in the 2-bus system. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

**Figure 4.** Figure 4: Two-bus probability sweep for the Gaussian-mixture extension. The [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: Five-bus probability sweep for Gaussian uncertainty. The reference [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

**Figure 6.** Figure 6: Five-bus probability sweep for the Gaussian-mixture extension. The [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗

read the original abstract

This paper introduces a framework based on Large Deviation Theory (LDT) to accurately and efficiently compute the rare probabilities of voltage collapse. We formulate the problem as finding the most probable failure point (the instanton) on the stability boundary and derive both first-order and second-order approximations for the collapse probability. The second-order method incorporates the local curvature of the stability boundary, yielding higher accuracy. This LDT framework generalizes methods based on Mahalanobis distance and is extensible to non-Gaussian uncertainties. We validate our approach on test systems, demonstrating that the LDT estimates converge to Monte Carlo results in the rare-event regime where direct sampling becomes computationally prohibitive.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a usable LDT instanton method with a curvature correction for estimating tiny voltage-collapse probabilities, generalizing Mahalanobis approaches, but the Monte Carlo convergence claim in the truly prohibitive regime rests on indirect checks.

read the letter

The main thing here is a concrete computational recipe: locate the most probable point on the stability boundary (the instanton), then use first-order and second-order Large Deviation Theory approximations that fold in local curvature to get the probability. It explicitly extends earlier distance-based shortcuts and notes it can handle non-Gaussian noise. That is the actual advance over prior work in this corner of power-system reliability analysis. On the positive side, the authors run the method on standard test systems and report that the estimates line up with Monte Carlo once the events get rare enough that plain sampling slows down. The math looks standard LDT applied to a well-posed optimization problem, with no obvious circularity or invented quantities. The second-order term is a straightforward Hessian correction, which is the sort of incremental improvement that can matter for tail accuracy. The soft spot is exactly the one the stress-test flags. The abstract says the LDT numbers converge to Monte Carlo “in the rare-event regime where direct sampling becomes computationally prohibitive.” That phrasing creates a verification gap: if you cannot run enough Monte Carlo samples at the operating point of interest, any comparison must be done at milder rarity levels or against some other reference whose own error is unproven. The paper would be stronger with an explicit statement of how far into the tail the direct comparison actually reaches and what fallback reference is used beyond that point. Minor implementation details such as how the instanton is found numerically and how sensitive the curvature term is to modeling choices are also worth checking in the full text, but they are not load-bearing flaws. This is the kind of paper that power-system reliability groups and stochastic optimization people would want to see. It is not a foundational theoretical result, but it supplies a practical tool that sits on established theory and shows numerical behavior on realistic test cases. I would bring it to a reading group for the method details and the validation discussion. It deserves a serious referee rather than a desk reject; the core idea is clear, the application is important, and the remaining questions are the normal ones that review can tighten.

Referee Report

2 major / 2 minor

Summary. The paper introduces a Large Deviation Theory (LDT) framework to compute rare probabilities of voltage collapse in power systems. It formulates the task as locating the most probable failure point (instanton) on the stability boundary and derives first-order and second-order approximations for the collapse probability, with the second-order version incorporating local curvature of the boundary. The approach is claimed to generalize Mahalanobis-distance methods and extend to non-Gaussian uncertainties, with validation on test systems showing that the LDT estimates converge to Monte Carlo results in the rare-event regime.

Significance. If the asymptotic approximations are accurate, the method would supply an efficient, scalable alternative to direct sampling for estimating extremely small voltage-collapse probabilities under uncertainty, which is valuable for probabilistic reliability assessment in power systems. The generalization beyond Gaussian assumptions and the explicit curvature correction represent concrete technical advances over existing distance-based heuristics.

major comments (2)

[Abstract] Abstract: the statement that 'LDT estimates converge to Monte Carlo results in the rare-event regime where direct sampling becomes computationally prohibitive' cannot be directly supported by Monte Carlo comparisons, because Monte Carlo sampling is infeasible precisely in that regime. Any numerical validation must therefore occur at moderate rarity levels where the large-deviation asymptotics have not yet become dominant, leaving the central convergence claim without empirical grounding at the operating point advertised.
[Validation section] Validation section: the manuscript should report the precise rarity levels (e.g., probability thresholds or noise-variance values) at which Monte Carlo reference values were obtained, together with an explicit error analysis or alternative reference method (importance sampling, multilevel splitting, etc.) whose accuracy in the tail has been independently verified. Without this, the claim that the instanton-plus-curvature approximation reliably captures the probability remains untested in the asymptotic regime.

minor comments (2)

Clarify whether the stability boundary is assumed to be smooth and whether the Hessian of the constraint is evaluated analytically or numerically; this affects the practical implementation of the second-order correction.
Provide a brief statement of the precise large-deviation rate function used and the scaling of the noise variance that drives the rare-event limit.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review of our manuscript. We address the major comments point by point below and have made revisions to the manuscript where necessary to strengthen the presentation.

read point-by-point responses

Referee: [Abstract] Abstract: the statement that 'LDT estimates converge to Monte Carlo results in the rare-event regime where direct sampling becomes computationally prohibitive' cannot be directly supported by Monte Carlo comparisons, because Monte Carlo sampling is infeasible precisely in that regime. Any numerical validation must therefore occur at moderate rarity levels where the large-deviation asymptotics have not yet become dominant, leaving the central convergence claim without empirical grounding at the operating point advertised.

Authors: We acknowledge the validity of this observation. Direct Monte Carlo sampling is indeed infeasible in the extremely rare-event regime. In our numerical experiments, we compare the LDT approximations with Monte Carlo at increasing levels of rarity (decreasing probability thresholds) within the feasible computational range, observing that the approximations become increasingly accurate as the events become rarer. This provides empirical support for the asymptotic regime. To address the concern, we will revise the abstract to state that the LDT estimates are shown to approach Monte Carlo results as the event probability decreases to levels where sampling becomes challenging, and that the method targets the prohibitive rare-event regime. revision: partial
Referee: [Validation section] Validation section: the manuscript should report the precise rarity levels (e.g., probability thresholds or noise-variance values) at which Monte Carlo reference values were obtained, together with an explicit error analysis or alternative reference method (importance sampling, multilevel splitting, etc.) whose accuracy in the tail has been independently verified. Without this, the claim that the instanton-plus-curvature approximation reliably captures the probability remains untested in the asymptotic regime.

Authors: We agree that more precise documentation is needed. In the revised manuscript, we will explicitly report the noise-variance values and corresponding Monte Carlo probability estimates used in the validation. We will also include a discussion of the statistical error in the Monte Carlo estimates and reference importance sampling as a complementary method for rare-event probability estimation, noting its prior validation in the literature for tail probabilities. revision: yes

Circularity Check

0 steps flagged

No circularity: LDT application to instanton on stability boundary is independent of fitted inputs or self-referential definitions

full rationale

The derivation begins from standard Large Deviation Theory to locate the most probable failure point (instanton) on the stability boundary and then constructs first-order and second-order local approximations that incorporate curvature. These steps rely on established LDT results rather than redefining quantities in terms of the target probability or fitting parameters to the very rare-event data being estimated. Validation against Monte Carlo is presented as an external benchmark performed on test systems, not as an internal consistency check that forces the result by construction. No self-citation chain is invoked to justify uniqueness or to smuggle an ansatz; the framework is therefore self-contained against external mathematical and numerical references.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper relies on standard Large Deviation Theory without introducing explicit free parameters or new entities in the abstract; the central claim rests on the applicability of LDT to the voltage stability problem.

axioms (1)

domain assumption Large Deviation Theory principles apply to the probability of voltage collapse in power systems
Invoked as the basis for formulating the problem and deriving the instanton approximations.

pith-pipeline@v0.9.0 · 5403 in / 1213 out tokens · 38240 ms · 2026-05-13T17:19:49.466141+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We formulate the problem as finding the most probable failure point (the instanton) on the stability boundary and derive both first-order and second-order approximations for the collapse probability. The second-order method incorporates the local curvature of the stability boundary
IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean absolute_floor_iff_bare_distinguishability unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

lim n→∞ −1/n log P(λ_n ∈ A) = I(λ*)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

[1]

New methods for computing a closest saddle node bifurcation and worst case load power margin for voltage collapse,

I. Dobson and L. Lu, “New methods for computing a closest saddle node bifurcation and worst case load power margin for voltage collapse,” IEEE Transactions on Power Systems, vol. 8, no. 3, pp. 905–913, 1993

work page 1993
[2]

Quantifying transmission reliability margin,

J. Zhang, I. Dobson, and F. Alvarado, “Quantifying transmission reliability margin,”International Journal of Electrical Power & Energy Systems, vol. 26, no. 9, p. 697–702, Nov. 2004. [Online]. Available: http://dx.doi.org/10.1016/S0142-0615(04)00071-7

work page doi:10.1016/s0142-0615(04)00071-7 2004
[3]

An iterative method for determining the most probable bifurcation in large scale power systems,

M. Mittelstaedt, S. Bauer, and A. Schnettler, “An iterative method for determining the most probable bifurcation in large scale power systems,”IEEE Transactions on Power Systems, vol. 32, no. 4, p. 2966–2973, Jul. 2017. [Online]. Available: http://dx.doi.org/10.1109/TPWRS.2016.2622742

work page doi:10.1109/tpwrs.2016.2622742 2017
[4]

Predicting failures in power grids: The case of static overloads,

M. Chertkov, F. Pan, and M. G. Stepanov, “Predicting failures in power grids: The case of static overloads,”IEEE Transactions on Smart Grid, vol. 2, no. 1, p. 162–172, Mar. 2011. [Online]. Available: http://dx.doi.org/10.1109/TSG.2010.2090912

work page doi:10.1109/tsg.2010.2090912 2011
[5]

Temperature overloads in power grids under uncertainty: A large deviations approach,

T. Nesti, J. Nair, and B. Zwart, “Temperature overloads in power grids under uncertainty: A large deviations approach,”IEEE Transactions on Control of Network Systems, vol. 6, no. 3, pp. 1161–1173, 2019

work page 2019
[6]

A kinetic monte carlo approach for simulating cascading transmission line failure,

J. Roth, D. A. Barajas-Solano, P. Stinis, J. Weare, and M. Anitescu, “A kinetic monte carlo approach for simulating cascading transmission line failure,”Multiscale Modeling & Simulation, vol. 19, no. 1, pp. 208–241, 2021

work page 2021
[7]

Failure proba- bility constrained ac optimal power flow,

A. Subramanyam, J. Roth, A. Lam, and M. Anitescu, “Failure proba- bility constrained ac optimal power flow,”IEEE Transactions on Power Systems, vol. 37, no. 6, pp. 4683–4695, 2022

work page 2022
[8]

Network cascade vul- nerability using constrained bayesian optimization,

A. Lam, M. Anitescu, and A. Subramanyam, “Network cascade vul- nerability using constrained bayesian optimization,”arXiv preprint arXiv:2304.14420, 2023

work page arXiv 2023
[9]

Electricity market- clearing with extreme events,

T. Tapia, Z. Liang, C. Konstantinou, and Y . Dvorkin, “Electricity market- clearing with extreme events,”IEEE Transactions on Energy Markets, Policy and Regulation, 2024

work page 2024
[10]

Optimization under rare chance constraints,

S. Tong, A. Subramanyam, and V . Rao, “Optimization under rare chance constraints,”SIAM Journal on Optimization, vol. 32, no. 2, pp. 930–958, 2022

work page 2022
[11]

The irrelevance of electric power system dynamics for the loading margin to voltage collapse and its sensitivities,

I. Dobson, “The irrelevance of electric power system dynamics for the loading margin to voltage collapse and its sensitivities,”Nonlinear Theory and Its Applications, IEICE, vol. 2, no. 3, pp. 263–280, 2011

work page 2011
[12]

Y . A. Kuznetsov,Elements of Applied Bifurcation Theory, 4th ed., ser. Applied Mathematical Sciences. Cham: Springer, 2023, vol. 112

work page 2023
[13]

Dembo and O

A. Dembo and O. Zeitouni,Large Deviations Techniques and Applica- tions. Springer Science & Business Media, 2009, vol. 38

work page 2009
[14]

Bucklew,Introduction to rare event simulation

J. Bucklew,Introduction to rare event simulation. Springer Science & Business Media, 2013

work page 2013
[15]

An introduction to large deviations with applications in physics,

I. N. Burenev, D. W. Cloete, V . Kharbanda, and H. Touchette, “An introduction to large deviations with applications in physics,”arXiv preprint arXiv:2503.16015, 2025

work page arXiv 2025
[16]

A direct method for computing a closest saddle node bifurcation in the load power parameter space of an electric power system,

I. Dobson, L. Lu, and Y . Hu, “A direct method for computing a closest saddle node bifurcation in the load power parameter space of an electric power system,” in1991 IEEE International Symposium on Circuits and Systems (ISCAS), 1991, pp. 3019–3022 vol.5

work page 1991
[17]

Computing a closest bifurcation instability in multidimen- sional parameter space,

I. Dobson, “Computing a closest bifurcation instability in multidimen- sional parameter space,”Journal of nonlinear science, vol. 3, no. 1, pp. 307–327, 1993

work page 1993
[18]

Extreme event probability estimation using pde-constrained optimization and large deviation theory, with application to tsunamis,

S. Tong, E. Vanden-Eijnden, and G. Stadler, “Extreme event probability estimation using pde-constrained optimization and large deviation theory, with application to tsunamis,” Communications in Applied Mathematics and Computational Science, vol. 16, no. 2, p. 181–225, Nov. 2021. [Online]. Available: http://dx.doi.org/10.2140/camcos.2021.16.181

work page doi:10.2140/camcos.2021.16.181 2021
[19]

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 2018 Power Systems Computation Conference (PSCC), June 2018, pp. 1–8

work page 2018
[20]

A most probable point-based method for efficient uncertainty analysis,

X. Du and W. Chen, “A most probable point-based method for efficient uncertainty analysis,”Journal of Design and Manufacturing Automation, vol. 4, no. 1, pp. 47–66, 2001. [Online]. Available: https://doi.org/10.1080/15320370108500218

work page doi:10.1080/15320370108500218 2001
[21]

The continuation power flow: a tool for steady state voltage stability analysis,

V . Ajjarapu and C. Christy, “The continuation power flow: a tool for steady state voltage stability analysis,”IEEE Transactions on Power Systems, vol. 7, no. 1, pp. 416–423, 1992

work page 1992
[22]

An adaptive multi-step levenberg-marquardt continuation power flow method for voltage stability assessment in the ill-conditioned power systems,

R. Pourbagher, S. Y . Derakhshandeh, and M. E. Hamedani Golshan, “An adaptive multi-step levenberg-marquardt continuation power flow method for voltage stability assessment in the ill-conditioned power systems,”International Journal of Electrical Power & Energy Systems, vol. 134, p. 107425, 2022. [Online]. Available: https://www.sciencedirect.com/science/...

work page 2022
[23]

Non- gaussian power grid frequency fluctuations characterized by l ´evy-stable laws and superstatistics,

B. Sch ¨afer, C. Beck, K. Aihara, D. Witthaut, and M. Timme, “Non- gaussian power grid frequency fluctuations characterized by l ´evy-stable laws and superstatistics,”Nature Energy, vol. 3, no. 2, p. 119–126, Jan

work page
[24]

Available: https://doi.org/10.1038/s41560-017-0058-z

[Online]. Available: http://dx.doi.org/10.1038/s41560-017-0058-z 24th Power Systems Computation Conference PSCC 2026 Limassol, Cyprus — June 8-12, 2026

work page doi:10.1038/s41560-017-0058-z 2026