Pith.

Dynamical Systems · arXiv:2611.17219 v1 · submitted 12 Nov 2026 · 15 references

Structure-Preserving Leaky Integrate-and-Fire Dynamics on Arbitrary Time Scales

Billur Kaymakçalan (University of Turkish Aeronautical Association), Behiye İlayda Selçuk (Bilkent University)

The paper says

“The leaky integrate-and-fire (LIF) neuron combines a passive membrane law with threshold-triggered firing and reset. We formulate this hybrid neuron model, under constant input, on an arbitrary Hilger time scale 𝕋.”

more of the abstract

Our central result is that, after the first spike, the firing times satisfy tn + 1 = Π𝕋+(tn + Tc), where Tc is the classical reset-to-threshold travel time. This recursion requires no periodicity, homogeneity, or translation invariance of 𝕋; it holds for every time scale that is unbounded above. To preserve the passive membrane physics, we use the inverse cylinder transformation to construct a positively regressive coefficient pm. The resulting delta-dynamic equation is exactly the restriction to 𝕋 of the classical continuous-time LIF trajectory. By contrast, the direct transcription VΔ = −(V−V∞)/τm exhibits graininess-dependent sign reversal and amplification at coarse right-scattered points. The first firing time is the forward projection of the classical continuous threshold-crossing time onto 𝕋, and, when the graininess is bounded by μ∗, the per-event forward-projection delay – from each reset to the next realized firing time, relative to the continuous crossing time computed from that same reset – is bounded above by μ∗. On 𝕋 = hℤ we obtain the explicit lattice-quantized law $T_{\mathrm{ISI}}^{(h)}=h\lceil T_c/h\rceil$, which converges to the classical continuous-time inter-spike interval as h → 0+. The framework therefore separates preserved membrane physics from the temporal opportunity structure imposed by the time domain. To our knowledge, this is the first systematic Hilger-calculus treatment of a threshold-and-reset LIF spiking neuron on an arbitrary time scale.

from the abstract

Pith says

Machine review · deepseek-v4-flash · 13 Nov 2026 · what was read was not recorded · not checked by a person

It can be wrong. Challenge this review or reply to it.

Holds up. Clean, honest math: the recursion tn+1 = ΠT^+(tn + Tc) is correct and the paper knows its scope. The constant-input limitation is real but acknowledged. Worth a referee. This line is wrong

7 minor

How often objections from reviews like this one are right: not yet measured

Are you an author? What you can do.

Not checked: figures, code, what the 15 references say.

7 minor objections

The paper says, at Abstract

Pith says: minor objection 1 of 7

The abstract states 'the first systematic Hilger-calculus treatment of a threshold-and-reset LIF spiking neuron on an arbitrary time scale.' This is a reasonable claim given the literature review, but the phrase 'systematic Hilger-calculus treatment' could be slightly toned down to 'a Hilger-calculus treatment' to avoid any perception of overclaiming, though the authors do carefully distinguish from prior work in the introduction.

The paper says, at §4.2, Definition 5

Pith says: minor objection 2 of 7

The definition of p_m(t) for µ(t)>0 is given as (e^{-µ(t)/τ_m} - 1)/µ(t). It might be helpful to explicitly note that this is the standard inverse cylinder transformation applied to the constant -1/τ_m, which is already stated but could be emphasized for readers less familiar with time-scale calculus.

show 5 more

The paper says, at §5, Theorem 2

Pith says: minor objection 3 of 7

The proof of Theorem 2 states 'the subthreshold label is purely interpretive.' This is a bit informal; consider rephrasing to clarify that the voltage trajectory (5) is defined for all t in T regardless of whether it exceeds threshold, and the threshold condition is applied separately.

The paper says, at §6, Lemma 3

Pith says: minor objection 4 of 7

The condition 'T∩(-∞, x] ≠ ∅' is needed for the definition of δ_T(x). This is fine, but it might be worth noting that for the application to firing times, this condition holds because t_n + T_c ≥ t_0 + T_c > t_0 and T is unbounded above, so the intersection is nonempty. This is implicit but could be made explicit.

The paper says, at §7, Numerical illustrations

Pith says: minor objection 5 of 7

The paper states that all figures are produced by a single reproducible script (make_figures.py). However, the script is not included in the arXiv submission. While this does not affect the mathematical claims, including the script or a link to it would enhance reproducibility.

The paper says, at References

Pith says: minor objection 6 of 7

Reference [5] (Duque et al.) is from 2023; the paper might benefit from a brief note on how the present work differs from that example, which is already done in the introduction. No further changes needed.

The paper says, at General

Pith says: minor objection 7 of 7

The paper uses the notation Π_T^+ for forward projection. This is defined in Definition 3. It might be helpful to remind the reader of this notation when it first appears in the abstract or introduction, but it is clear enough.

What would settle it

Independently re-derive t_{n+1} by writing the restarted IVP on T∩[t_n,∞) as y^Δ=p_m y with y(t_n)=V_r−V∞, solving via Fact 3, and computing the first t∈T with y(t)≥V_th−V∞; verify that this equals Π_T^+(t_n+T_c) without using any regularity of T beyond closedness and unboundedness above. As a numerical cross-check, simulate this IVP directly on a non-translation-invariant time scale such as T={0}∪⋃_{k≥1}[a_k,b_k] with varying block lengths and gaps for 10^3 resets, and compare every firing time to the recursion.

from the skeptic's reading

How the review went

4 readings by deepseek-v4-flash · 13 Nov 2026

  1. Reader holds up confidence high
  2. Skeptic partly agrees 2 min 50 s · 13,700 tokens read, 23,817 written
  3. Editor wrote the verdict above 8 s · 14,646 tokens read, 562 written
  4. Referee minor revision: 7 minor 13 s · 14,075 tokens read, 968 written

Not run on this paper: the simulated authors' defence.

Pith did not check this review's objections against the paper.

The review in full

The review in brief, machine-written

The paper does exactly what it says: it takes the LIF neuron under constant current and writes it on an arbitrary time scale T using the inverse cylinder transformation so that the membrane trajectory is the exact restriction of the classical continuous solution. The central result is the recursion tn+1 = ΠT^+(tn + Tc) for all n ≥ 1 after the first spike, which holds for any closed T unbounded above with no periodicity or homogeneity required. The math is clean, the derivations are correct, and the authors are scrupulous about what they do and do not claim. They explicitly decline to claim the first single-neuron model on time scales (Duque et al. already have one without threshold/reset), they acknowledge that exponential propagation is not new, and they state the constant-input limitation plainly in the Limitations paragraph. The paper also gives a nice classification of the failure modes of the naive delta-LIF transcription (Theorem 1) and the hZ quantization law (Corollary 4) is a clean closed-form result.

The soft spots are real but contained. The load-bearing premise is constant input and constant parameters (τm, V_∞), which makes Tc independent of absolute time. The authors say this is the foundational case and leave time-varying input, refractory periods, adaptive thresholds, and network coupling to future work. That is an honest scope, not a hidden flaw. The only concrete missing artifact is the promised make_figures.py script, but every result is closed-form and no theorem depends on the simulations, so this is minor. The paper is not long, but it is self-contained and the arguments are checkable.

Who is this for? Mathematicians working in time-scale calculus who want a clean example of a hybrid system; computational neuroscientists who want a precise statement of what changes and what does not when you move from continuous to discrete or irregular time domains. It is not a simulation paper and it does not claim biological novelty beyond the constant-input case.

Recommendation: send to peer review. The paper is sound within its stated scope, the writing is clear, and the authors are honest about limitations. A referee can check the derivations quickly and confirm the scope. No desk reject needed.

The strongest objection

After line-by-line checking of Theorem 3 and its supporting results (Lemma 2, Corollary 1, Theorem 2, Lemma 3, Corollary 4), I find no unacknowledged load-bearing flaw in the central claim within its stated scope. The recursion t_{n+1}=Π_T^+(t_n+T_c) is justified by three facts, all present in the text: (i) the structure-preserving coefficient p_m gives V(t)=V∞+(V_0−V∞)e^{−(t−t_0)/τ_m} on T (Corollary 1), so the on-time-scale branch is the classical solution; (ii) after a reset at t_n∈T, the restarted branch is the same formula with V_r, whose continuous crossing time is t_n+T_c because τ_m and V∞ are constant; (iii) Π_T^+ maps into T whenever T is closed and unbounded above (Fact 5), and this projection argument needs no periodicity, homogeneity, or translation invariance. The only premise that could break the recursion outside its scope is time-varying input or nonconstant parameters, which the authors explicitly exclude in Assumption 1 and the Limitations paragraph; that is a scoping choice, not a hidden assumption. Lemma 3's gap bound and Corollary 4's hZ quantization also check out. The promised make_figures.py is not included, but all claims are closed-form and none depend on the simulations.

The reader's ACCEPT verdict is warranted. I examined the exact-restriction construction (Definition 5, Lemma 1, Lemma 2, Corollary 1), the firing-time theorems (Theorem 2, Theorem 3, Remark 2), the projection-gap lemma (Lemma 3, Corollary 3), and the hZ law (Corollary 4). The central recursion is a direct consequence of constant τ_m and V∞ plus Fact 5; no step uses any structural assumption on T beyond closedness and unboundedness above. The inverse-cylinder coefficient is the standard exact-exponential discretization, so the paper's claim that the membrane semigroup is preserved is correct. The naive transcription's failure modes (Theorem 1) are correctly classified. The one condition that would invalidate the recursion outside the paper's scope is time-varying input or nonconstant parameters; this is explicitly stated in Assumption 1 and the Limitations paragraph, so it is an acknowledged boundary rather than a defect. The unshipped figure script is a minor reproducibility issue, not relevant to the mathematical claim. Agreement with the reader is partial: the constant-input assumption is indeed the only conceivable failure mode, but because the paper scopes and acknowledges it, I do not treat it as a live objection.

The referee's report

If correct, the paper provides a clean separation between the passive membrane physics (which is preserved exactly) and the temporal opportunity structure imposed by the time domain. This is a conceptually useful framework for understanding how discrete or irregular time grids affect spike timing in LIF models. The results are mathematically rigorous within the stated scope (constant input, no adaptation, no refractory period). The paper is clearly written and makes appropriate use of time-scale calculus. The central recursion is simple and elegant. The paper does not claim more than it delivers; limitations are explicitly acknowledged.

Pith's reading of the paper (a machine's summary, not the authors')

The leaky integrate-and-fire (LIF) neuron combines a passive membrane leak with a threshold that, when crossed, triggers a spike and resets the voltage. Conventionally this is written as an ordinary differential equation on continuous time, but simulations often run on a clock, and experiments sample at uneven times. The paper asks whether the same membrane law can be stated on any closed subset of the real line, called a time scale, without changing the underlying biophysics.

Core claim. The central result is the recursion t_{n+1} = Π_T^+(t_n + T_c) for all n ≥ 1, where T_c is the classical reset-to-threshold travel time and Π_T^+ is forward projection onto the time scale T. The paper states this holds for every time scale that is closed and unbounded above, with no periodicity, homogeneity, or translation invariance required (Theorem 3, Remarks 2). If correct, the spike train on any such time domain is governed entirely by the geometry of T plus the single autonomous constant T_c.

The premise it rests on. The load-bearing premise is the constant-input autonomy of the subthreshold law, assumptions (A2) and (A3) that τ_m and V_∞ are constant. This is what makes the reset-to-threshold travel time T_c independent of the absolute position of the reset time, and without it the one-line recursion collapses: a time-varying input would give a different crossing time after each reset. The authors explicitly acknowledge this in the Limitations paragraph, stating that time-varying input, refractory intervals, and adaptive thresholds are left to future work, so the recursion is characterized only for the constant-input founding case.

Check this review

Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.

How often reviews like this one are right

Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.

How often objections from reviews like this one are right: not yet measured

Every value above, in the review's public record (JSON)

Reviews by people

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

The leaky integrate-and-fire (LIF) neuron combines a passive membrane law with threshold-triggered firing and reset. We formulate this hybrid neuron model, under constant input, on an arbitrary Hilger time scale 𝕋. Our central result is that, after the first spike, the firing times satisfy tn + 1 = Π𝕋+(tn + Tc), where Tc is the classical reset-to-threshold travel time. This recursion requires no periodicity, homogeneity, or translation invariance of 𝕋; it holds for every time scale that is unbounded above. To preserve the passive membrane physics, we use the inverse cylinder transformation to construct a positively regressive coefficient pm. The resulting delta-dynamic equation is exactly the restriction to 𝕋 of the classical continuous-time LIF trajectory. By contrast, the direct transcription VΔ = −(V−V∞)/τm exhibits graininess-dependent sign reversal and amplification at coarse right-scattered points. The first firing time is the forward projection of the classical continuous threshold-crossing time onto 𝕋, and, when the graininess is bounded by μ∗, the per-event forward-projection delay – from each reset to the next realized firing time, relative to the continuous crossing time computed from that same reset – is bounded above by μ∗. On 𝕋 = hℤ we obtain the explicit lattice-quantized law $T_{\mathrm{ISI}}^{(h)}=h\lceil T_c/h\rceil$, which converges to the classical continuous-time inter-spike interval as h → 0+. The framework therefore separates preserved membrane physics from the temporal opportunity structure imposed by the time domain. To our knowledge, this is the first systematic Hilger-calculus treatment of a threshold-and-reset LIF spiking neuron on an arbitrary time scale.

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-11-14

The authors' main claim, as the classifier picked it out 1.00

“Our central result is that, after the first spike, the firing times satisfy $t_{n+1}=\Pi_{\mathbb{T}}^{+}(t_n+T_c)$, where $T_c$ is the classical reset-to-threshold travel time.”

How the main claim is stated: asserted 1.00

What the abstract itself states, as probabilities: claims first 0.96 · states limitation 0.44 · human participants 0.02 · controlled comparison 0.18 · randomized assignment 0.01 · states quantitative result 0.94 · compares with prior results 0.83 · mentions code or data availability 0.02

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
The leaky integrate-and-fire (LIF) neuron combines a passive membrane law with threshold-triggered firing and reset. background 1.00 0.05 0.01 asserted 1.00
We formulate this hybrid neuron model, under constant input, on an arbitrary Hilger time scale $\mathbb{T}$. contribution 0.65 0.95 0.03 asserted 0.99
Our central result is that, after the first spike, the firing times satisfy $t_{n+1}=\Pi_{\mathbb{T}}^{+}(t_n+T_c)$, where $T_c$ is the classical reset-to-threshold travel time. result 1.00 0.94 0.95 asserted 1.00
This recursion requires no periodicity, homogeneity, or translation invariance of $\mathbb{T}$; it holds for every time scale that is unbounded above. result 0.88 0.91 0.14 asserted 1.00
To preserve the passive membrane physics, we use the inverse cylinder transformation to construct a positively regressive coefficient $p_m$. method 0.99 0.93 0.03 asserted 1.00
The resulting delta-dynamic equation is exactly the restriction to $\mathbb{T}$ of the classical continuous-time LIF trajectory. result 0.91 0.93 0.09 asserted 1.00
By contrast, the direct transcription $V^\Delta=-(V-V_\infty)/\tau_m$ exhibits graininess-dependent sign reversal and amplification at coarse right-scattered points. result 0.66 0.79 0.03 asserted 1.00
The first firing time is the forward projection of the classical continuous threshold-crossing time onto $\mathbb{T}$, and, when the graininess is bounded by $\mu^\ast$, the per-event forward-projection delay -- from each reset to the next realized firing time, relative to the continuous crossing time computed from that same reset -- is bounded above by $\mu^\ast$. result 0.99 0.91 0.10 asserted 1.00
On $\mathbb{T}=h\mathbb{Z}$ we obtain the explicit lattice-quantized law $T_{\mathrm{ISI}}^{(h)}=h\lceil T_c/h\rceil$, which converges to the classical continuous-time inter-spike interval as $h\to0^+$. result 1.00 0.92 0.06 asserted 1.00
The framework therefore separates preserved membrane physics from the temporal opportunity structure imposed by the time domain. interpretation 1.00 0.79 0.15 asserted 1.00
To our knowledge, this is the first systematic Hilger-calculus treatment of a threshold-and-reset LIF spiking neuron on an arbitrary time scale. contribution 0.65 0.64 0.06 asserted 0.95

What kind of paper the classifier reports this is: theoretical result 0.99 · evidence from proof or derivation 1.00

What was examined, and what was not

Sources read

  • arXiv listing abstract v1

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-14: 1 request, 8,537 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Figures (5)
Figure 1 of the paper
Figure 1. Exact restriction of the classical LIF trajectory to different time scales. The classical continuous-time passive, unreset trajectory V (t) (drawn beyond thresh￾old, since Corollary 1 places no threshold restriction on its validity) is shown together with the structure-preserving solution (5) evaluated on three representative time scales: (A) the uniform lattice T = 2Z ms, (B) the coarser lattice T = 8Z ms, and (C) a deterministic irregular time scale. In every case the time-scale solution lies exactly on the classical continuous trajectory at all t ∈ T, illustrating Corollary 1.…
Figure 2 of the paper
Figure 2. Graininess-dependent failure of the naive delta-LIF formulation. Naive and structure-preserving LIF trajectories are compared on uniform time scales with three values of the dimensionless graininess ratio ρ := µ/τm. (A) For ρ = 0.5 < 1, both formulations decay toward V∞, although the naive trajectory does not exactly reproduce the continuous membrane semigroup. (B) For ρ = 1.5 ∈ (1, 2), the naive multiplier becomes negative, producing sign-alternating overshoots around V∞, whereas the structure-preserving trajectory remains monotone (Corollary 2). (C) For ρ = 2.5 > 2, the magnitude…
Figure 3 of the paper
Figure 3. Local dynamical regimes of the naive delta-LIF multiplier. The naive right-scattered update (4) is governed by the dimensionless multiplier M(ρ) = 1 − ρ, ρ := µ/τm. (A) Signed multiplier M(ρ). (B) Magnitude |M(ρ)|. For 0 < ρ < 1 the update is sign-preserving and contractive; for 1 < ρ < 2 it is sign-alternating but contractive; at ρ = 2, |M| = 1; and for ρ > 2, |M| > 1, resulting in local amplification. The vertical boundaries at ρ = 1 and ρ = 2 correspond to the regime transitions characterized in Theorem 1. the on-time-scale firing semantics of Section 5: the continuous threshold…
Figure 4 of the paper
Figure 4. Forward projection of the continuous threshold-crossing time onto a time scale. The vertical line marks the classical continuous-time threshold-crossing time t ∗ . For each temporal domain, the filled marker denotes the realized on-time-scale firing time tf = Π+ T (t ∗ ) (Theorem 2). For T = R, the projection is exact and tf = t ∗ . On uniform lattices hZ, firing is delayed until the first admissible time-scale point at or after t ∗ . The corresponding projection error δT(t ∗ ) = tf − t ∗ is annotated for each time scale. For the uniform lattices, the shaded interval repre
Figure 5 of the paper
Figure 5. Quantization of the inter-spike interval and firing rate on hZ. The genuine staircase quantity is the integer step count N(h) := ⌈Tc/h⌉. The plotted quantities T (h) ISI = hN(h) and fh = 1000/(hN(h)) Hz are therefore lattice-quantized rather than step functions themselves: on intervals where N(h) is constant they vary linearly and hyperbolically in h, respectively (with h in ms, for the representative parameters above), with a jump discontinuity each time N(h) decreases. (A) T (h) ISI (Corollary 4); the dashed horizontal line indicates the continuous reset-to-threshold travel time

Figures are the authors', reproduced from the paper.

References (15)
  1. [1] S. Hilger, Analysis on measure chains — a unified approach to continuous and discrete calculus,Results Math.18 (1990), 18–56 on Pith
  2. [2] S. Hilger, Differential and difference calculus — unified!,Nonlinear Analysis: Theory, Methods & Applications 30(5) (1997), 2683–2694 on Pith
  3. [3] V. Lakshmikantham, S. Sivasundaram, and B. Kaymak¸ calan,Dynamic Systems on Measure Chains, Math- ematics and Its Applications, vol. 370, Kluwer Academic Publishers, 1996. https://doi.org/10.1007/ 978-1-4757-2449-3 on Pith
  4. [4] M. Bohner and A. Peterson,Dynamic Equations on Time Scales: An Introduction with Applications, Birkh¨ auser, 2001 on Pith
  5. [5] C. Duque, H. Leiva, R. Gallo, A. Tridane, On the Existence and Stability of Bounded Solutions for Abstract Dynamic Equations on Time Scales,Int. J. Differential Equations, vol. 2023, Article ID 8489196 on Pith
  6. [6] F. Pasemann, Dynamics of a single model neuron,Int. J. Bifurcation and Chaos3(2) (1993), 271–278 on Pith
  7. [7] S. Rotter, M. Diesmann, Exact digital simulation of time-invariant linear systems with applications to neuronal modeling,Biol. Cybernetics81(1999), 381–402 on Pith
  8. [8] R. Brette, Exact simulation of integrate-and-fire models with synaptic conductances,Neural Comput.18(8) (2006), 2004–2027 on Pith
  9. [9] R. Brette, Exact simulation of integrate-and-fire models with exponential currents,Neural Comput.19(10) (2007), 2604–2609 on Pith
  10. [10] W. Teka, T. M. Marinov, F. Santamaria, Neuronal spike timing adaptation described with a fractional leaky integrate-and-fire model,PLoS Comput. Biol.10(3) (2014), e1003526 on Pith
  11. [11] Y. Li, Y. Hua, and Y. Fei, Global exponential stability of delayed Cohen–Grossberg BAM neural networks with impulses on time scales,Journal of Inequalities and Applications2009(2009), Article 491268 on Pith
  12. [12] A. Chen and D. Du, Global exponential stability of delayed BAM network on time scale,Neurocomputing 71(16–18) (2008), 3582–3588 on Pith
  13. [13] Y. Li and S. Gao, Global exponential stability for impulsive BAM neural networks with distributed delays on time scales,Neural Processing Letters31(1) (2010), 65–91 on Pith
  14. [14] Y. Li and C. Wang, Almost periodic solutions to dynamic equations on time scales and applications,Journal of Applied Mathematics2012(2012), Article 463913 on Pith
  15. [15] Z. He, C. Li, L. Chen, and Z. Cao, Dynamic behaviors of the FitzHugh–Nagumo neuron model with state-dependent impulsive effects,Neural Networks121(2020), 497–511. University of Turkish Aeronautical Association, Ankara 06790, T¨urkiye Email address:bkaymakcalan@thk.edu.tr Bilke... on Pith

Challenges and discussion

Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.

No comments yet.

Sign in to comment