arxiv: 2605.00750 · v1 · submitted 2026-05-01 · 📊 stat.OT · math.PR· math.ST· nlin.AO· stat.TH

Recognition: unknown

Quenched Amplification and Tail Shaping in Networked Systems with Memory and Regime Switching

Mauricio Herrera-Mar\'in

Authors on Pith no claims yet

Pith reviewed 2026-05-09 14:56 UTC · model grok-4.3

classification 📊 stat.OT math.PRmath.STnlin.AOstat.TH

keywords memoryamplificationliftedregimesystemsextremenetworkedoperator

0 comments

The pith

Quenched amplification arises generically in linear regime-switching networks with Volterra memory, producing power-law burst tails whose exponent depends on dwell-time rates and lifted-operator growth, with a dynamic intervention to truncate tail risk.

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

Many real systems like traffic networks, power grids, or biological signaling switch between good and bad operating modes and remember what happened earlier. Even when the average behavior stays bounded, occasional huge spikes can appear. The authors embed the memory into a larger but finite set of equations and show that the geometry of how the system grows in bad modes creates these spikes. The size of the spikes follows a power law whose steepness is set by how long the bad modes last versus how fast the system amplifies during them. They give a simple online calculation, using the logarithmic norm of the lifted equations, that flags when amplification is likely. They then describe a control strategy that steps in only along the dangerous channels to force contraction and cut off the extreme events while leaving everyday operation untouched.

Core claim

Despite finite-horizon annealed boundedness, we show that quenched amplification emerges generically from the interaction of regime persistence, memory accumulation, and non-normal lifted operator geometry. A lower bound on burst-size distributions reveals power-law tails whose exponent is determined by the ratio between unfavorable dwell-time rates and an operator-defined instantaneous growth parameter.

Load-bearing premise

The system admits a finite-dimensional lifted ODE embedding that preserves the essential non-normal geometry and regime statistics, and that the proposed intervention can enforce contraction along amplification channels without altering exogenous regime dwell times or typical behavior.

Figures

Figures reproduced from arXiv: 2605.00750 by Mauricio Herrera-Mar\'in.

**Figure 1.** Figure 1: Illustrative directed sensor network used in the numerical experiments. Nodes view at source ↗

**Figure 2.** Figure 2: Memory ON (no DDDAS): quenched amplification and tail risk under regime switching. This figure summarizes an ensemble of Monte Carlo trajectories of the periodically forced networked system with SOE memory activated in the unfavorable regime. (Top-left) Quenched sample paths. Thin curves show E(t) = ∥x(t)∥2 for multiple realizations (same parameters, different regime paths), displayed on a logarithmic s… view at source ↗

**Figure 3.** Figure 3: Memory ON + DDDAS: tail-risk mitigation via a two-indicator hysteretic policy. Same simulation setting as view at source ↗

**Figure 4.** Figure 4: Memory OFF (control): regime switching alone does not generate heavy tails. Same network, forcing, and regime switching as in Figs. 2–3, but with memory gain set to zero (no SOE memory contribution). (Top-left) Quenched sample paths. All realizations of E(t) = ∥x(t)∥2 remain tightly clustered around a bounded oscillatory response on the log scale, and the maximum-burst trajectory is not qualitatively dis… view at source ↗

**Figure 5.** Figure 5: SAFE-in-U baseline, where the unfavorable regime is permanently replaced view at source ↗

**Figure 6.** Figure 6: Comparison of burst-size tail distributions across control strategies. Shown view at source ↗

read the original abstract

Networked systems operating under intermittent adverse conditions and long memory can remain stable on average while exhibiting rare but extreme trajectory-level excursions. We study linear regime-switching network dynamics with Volterra-type memory, formulated through a finite-dimensional lifted ordinary differential equation embedding. Despite finite-horizon annealed boundedness, we show that quenched amplification emerges generically from the interaction of regime persistence, memory accumulation, and non-normal lifted operator geometry. A lower bound on burst-size distributions reveals power-law tails whose exponent is determined by the ratio between unfavorable dwell-time rates and an operator-defined instantaneous growth parameter. This parameter is computable online via the Euclidean logarithmic norm of the lifted operator, yielding a practical early-warning indicator. Building on this structure, we introduce a dynamic data-driven intervention strategy that enforces contraction on demand along rare amplification channels, thereby shaping or truncating tail risk without altering exogenous regime statistics or typical system behavior. The results provide a geometrically grounded and operationally actionable framework for understanding and mitigating extreme events in memory-driven regime-switching systems.

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 links memory and regime switching in linear networks to quenched amplification and power-law burst tails through a lifted operator, with an online log-norm indicator and a targeted intervention, though the central derivations need direct checking.

read the letter

The main thing to know is that this work shows how finite-horizon average stability can coexist with rare but large trajectory bursts that follow power-law tails, driven by the combination of regime persistence, memory accumulation, and non-normal geometry in a lifted finite-dimensional ODE model. It also supplies an online early-warning signal from the Euclidean logarithmic norm and a data-driven control that contracts along the bad channels without touching normal operation or exogenous dwell times.

Referee Report

1 major / 0 minor

Summary. The manuscript studies linear regime-switching network dynamics with Volterra-type memory, formulated through a finite-dimensional lifted ODE embedding. It claims that despite finite-horizon annealed boundedness, quenched amplification emerges generically from the interaction of regime persistence, memory accumulation, and non-normal lifted operator geometry. A lower bound on burst-size distributions is asserted to reveal power-law tails whose exponent is determined by the ratio between unfavorable dwell-time rates and an operator-defined instantaneous growth parameter (computable online via the Euclidean logarithmic norm of the lifted operator). A dynamic data-driven intervention is proposed to enforce contraction along rare amplification channels, shaping or truncating tail risk without altering exogenous regime statistics or typical behavior.

Significance. If the derivations hold, the work supplies a geometrically grounded framework for understanding extreme events in memory-driven regime-switching systems together with practical early-warning indicators and an intervention strategy that preserves typical behavior and exogenous dwell times. The online computability of the growth parameter via the logarithmic norm is a clear operational strength.

major comments (1)

Abstract: the claim that a lower bound on burst-size distributions exists and that the power-law exponent is determined solely by the ratio of unfavorable dwell-time rates to the logarithmic norm is stated without derivation steps, error estimates, or explicit verification that the exponent is independent of fitting choices. The central claims rest on the lifted-operator construction, whose details are not shown; this is load-bearing for the quenched-amplification and tail-shaping results.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback. We appreciate the recognition of the framework's potential and have revised the manuscript to improve the visibility of the central derivations and the lifted-operator construction while preserving the original results.

read point-by-point responses

Referee: Abstract: the claim that a lower bound on burst-size distributions exists and that the power-law exponent is determined solely by the ratio of unfavorable dwell-time rates to the logarithmic norm is stated without derivation steps, error estimates, or explicit verification that the exponent is independent of fitting choices. The central claims rest on the lifted-operator construction, whose details are not shown; this is load-bearing for the quenched-amplification and tail-shaping results.

Authors: We agree that the abstract, by design, states results concisely without full derivations. The explicit construction of the finite-dimensional lifted ODE embedding the Volterra memory, the analytic lower bound on burst-size distributions, and the derivation of the power-law exponent (as the ratio of the unfavorable dwell-time rate to the Euclidean logarithmic norm of the lifted operator) are given in Sections 2.2 and 3.1–3.3. Section 2.2 details the regime-switching lifted operator and its non-normal geometry. Section 3 derives the tail exponent in closed form, proving independence from fitting choices via the exact expression involving the logarithmic norm; we have now added explicit error estimates and a supplementary numerical check confirming robustness across parameter regimes. To address visibility, we have revised the abstract to reference these sections and inserted a concise summary of the lifted-operator construction at the end of the introduction. These changes make the load-bearing elements more prominent without altering the claims or results. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation proceeds from the finite-dimensional lifted ODE embedding (an explicit modeling choice that preserves non-normal geometry and regime statistics) to the interaction of regime persistence, memory accumulation, and operator geometry. The instantaneous growth parameter is defined directly as the Euclidean logarithmic norm of the lifted operator, a computable quantity rather than a fitted or self-referential input. The power-law exponent on burst-size tails is then obtained as the ratio of unfavorable dwell-time rates to this norm; this is a derived relation from the quenched dynamics, not a renaming or redefinition of the inputs. The intervention strategy builds on the same geometric structure without altering exogenous statistics. No self-citation load-bearing steps, fitted inputs called predictions, or ansatz smuggling appear in the provided chain. The argument remains self-contained against the stated assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard linear dynamical-systems assumptions plus the specific construction of a finite-dimensional lift that encodes Volterra memory; no new physical entities are postulated.

axioms (2)

domain assumption Dynamics are linear regime-switching with Volterra-type memory and admit a finite-dimensional lifted ODE embedding
Explicitly stated as the modeling framework in the abstract.
domain assumption The lifted operator is non-normal and its Euclidean logarithmic norm supplies the instantaneous growth parameter
Used to define the growth parameter that enters the power-law exponent.

pith-pipeline@v0.9.0 · 5481 in / 1431 out tokens · 63557 ms · 2026-05-09T14:56:38.360248+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

16 extracted references

[1]

On approximation of functions by exponential sums.Applied and Computational Harmonic Analysis, 19(1):17–48, 2005

Gregory Beylkin and Lucas Monzón. On approximation of functions by exponential sums.Applied and Computational Harmonic Analysis, 19(1):17–48, 2005

2005
[2]

Springer, 2016

Dariusz Buraczewski, Ewa Damek, and Thomas Mikosch.Stochastic Models with Power-Law Tails: The EquationX=AX+B. Springer, 2016

2016
[3]

Dynamic data driven applications systems.Proceedings of the IEEE, 93(3):486–502, 2004

Frederica Darema. Dynamic data driven applications systems.Proceedings of the IEEE, 93(3):486–502, 2004

2004
[4]

Characterizing dynamic data driven applications systems (dddas)

Frederica Darema. Characterizing dynamic data driven applications systems (dddas). InComputational Science – ICCS 2009, pages 472–481. Springer, 2009

2009
[5]

Tails of a markov-modulated linear system.Comptes Rendus Mathematique, 338(11):865–868, 2004

Benoît de Saporta. Tails of a markov-modulated linear system.Comptes Rendus Mathematique, 338(11):865–868, 2004

2004
[6]

Tail of a linear diffusion with markov switching.The Annals of Applied Probability, 15(2):992–1018, 2005

Benoît de Saporta and Jean-François Yao. Tail of a linear diffusion with markov switching.The Annals of Applied Probability, 15(2):992–1018, 2005

2005
[7]

Springer, 2010

Kai Diethelm.The Analysis of Fractional Differential Equations. Springer, 2010

2010
[8]

Trefethen.Spectra and Pseudospectra: The Behavior of Nonnormal Matrices and Operators

Mark Embree and Lloyd N. Trefethen.Spectra and Pseudospectra: The Behavior of Nonnormal Matrices and Operators. Princeton University Press, 2005

2005
[9]

Charles M. Goldie. Implicit renewal theoryand tails of solutions of random equations. The Annals of Applied Probability, 1(1):126–166, 1991

1991
[10]

Cambridge University Press, 1990

Gustaf Gripenberg, Sven-Olof Londen, and Olof Staffans.Volterra Integral and Func- tional Equations. Cambridge University Press, 1990

1990
[11]

Random difference equations and renewal theory for products of random matrices.Acta Mathematica, 131:207–248, 1973

Harry Kesten. Random difference equations and renewal theory for products of random matrices.Acta Mathematica, 131:207–248, 1973

1973
[12]

Woodhead Pub- lishing, 2007

Xuerong Mao.Stochastic Differential Equations and Applications. Woodhead Pub- lishing, 2007

2007
[13]

Birkhäuser, 1993

Jan Prüss.Evolutionary Integral Equations and Applications. Birkhäuser, 1993

1993
[14]

Peter J. Schmid. Nonmodal stability theory.Annual Review of Fluid Mechanics, 39:129–162, 2007

2007
[15]

Trefethen, Anne E

Lloyd N. Trefethen, Anne E. Trefethen, Satish C. Reddy, and Tobin A. Driscoll. Hydrodynamic stability without eigenvalues.Science, 261(5121):578–584, 1993

1993
[16]

Stability of stochastic differential equations with markovian switching.Stochastic Processes and their Applications, 120(3):431–460, 2010

Chenggui Yuan and Xuerong Mao. Stability of stochastic differential equations with markovian switching.Stochastic Processes and their Applications, 120(3):431–460, 2010. 32

2010