Pith. sign in

REVIEW 2 major objections 6 minor 2 cited by

Large Deviations in Switching Diffusion: from Free Cumulants to Dynamical Transitions

T0 review · 2 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Switching diffusion's large-time cumulants are governed by the free cumulants of the diffusivity distribution, with rate-function transitions set by the edge behavior of W(D).

desk verdict A clean, correct paper linking switching diffusion to free probability; the main results are new and the analysis supports them. read the letter →

arxiv 2501.13754 v2 pith:2GA4GR4C submitted 2025-01-23 cond-mat.stat-mech math-phmath.MPmath.PR

classification cond-mat.stat-mechmath-phmath.MPmath.PR MSC 60F1046L5482C31 PACS 05.40.-a05.40.Fb02.50.-r
keywords switchingdiffusionfreecumulantsR-transformlargedeviationsdynamicaltransitionsnon-Gaussianrenewalprocessrandomdiffusivity
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 tries to prove that a single Brownian particle whose diffusion coefficient switches at rate $r$ to values drawn from an arbitrary distribution $W(D)$ has exact finite-time moments, and that at long times its cumulants are controlled by the free cumulants of $W(D)$. The central formula is $\Psi(q)=q^2R(q^2/r)$ for the scaled cumulant generating function, which turns the position distribution's large deviations into a clean object. A sympathetic reader should care because it gives a solvable classical model that displays Brownian yet non-Gaussian diffusion, and because the same parameter-free formula predicts dynamical transitions in the rate function depending only on the edge behaviour of $W(D)$. This is backed by exact renewal calculations and by numerical evaluation of probabilities as small as $10^{-2000}$.

What carries the argument

The engine is a renewal equation for the position density $p_r(x,t)$, whose first term is the no-switch Gaussian propagator and whose second term convolves the density with one last switching event. Its bilateral Laplace transform in $x$ and Laplace transform in $t$ closes exactly: $\tilde p_r(q,s)=J_r(q,s)/(1-rJ_r(q,s))$ with $J_r(q,s)=\int_0^{D_{\max}} dD\,W(D)/(r+s-Dq^2)$. For long times the leading singularity is a pole at $s=\Psi(q)$, solving $q^2/r = g((r+\Psi(q))/q^2)$ where $g$ is the Cauchy–Stieltjes transform of $W(D)$. The free-probability identity $g(R(z)+1/z)=z$ then forces $\Psi(q)=q^2R(q^2/r)$. Whether the pole or the branch cut $[-r,-r+D_{\max}q^2]$ dominates, and the smoothness with which the pole meets the branch cut, is what separates the smooth, second-order, and first-order dynamical regimes.

What would settle it

Numerically simulate the process and measure the empirical generating function $t^{-1}\ln\langle e^{qx}\rangle$ for $W(D)$ a Wigner semicircle on $[0,D_{\max}]$; the paper predicts a cusp in its second derivative at $q_c=\sqrt{4r/D_{\max}}$, with $\Psi(q)=D_{\max}q^2-r$ for $q>q_c$ and a quartic-plus-quadratic form below. If the measured slope near $q_c$ or the small-$q$ curvature disagrees, or if the rate function shows no kink at $y_c=2D_{\max}q_c$ for $\nu=1/2$, the claim fails. A complementary check is to compare the measured fourth cumulant with the predicted linear coefficient $(2n)!\, r^{1-n} \kappa_n(D)$ for $\beta$-distributed diffusivities with $\nu=4$.

Watch

Extended reading notes

Core claim

At large times the generating function of the position obeys $\hat p(q,t) \approx e^{t\Psi(q)}$, and the scaled cumulant generating function is $\Psi(q)=q^2 R(q^2/r)$, where $R(z)$ is the R-transform of the diffusivity distribution $W(D)$. Equivalently, for $rt \gg 1$ each even cumulant is linear in time, $\langle x^{2n}(t)\rangle_c \simeq (2n)!\, r^{1-n} \kappa_n(D)\, t$, with $\kappa_n(D)$ the $n$-th free cumulant of $W(D)$. From $\Psi(q)$ and its Legendre transform the paper obtains the full large-deviation rate function $I(y=x/t)$, whose small- and large-$y$ tails are universal ($y^2/(4\langle D\rangle)$ and $r+y^2/(4D_{\max})$) but whose interior depends on how $W(D)$ vanishes at $D_{\max}$. If $W(D)\sim(D_{\max}-D)^\nu$, the rate function is smooth for $\nu\le 0$, acquires one sharp light-cone transition for $0<\nu<1$, and acquires an intermediate linear/traveling-front regime for $\nu>1$.

Load-bearing premise

The load-bearing premise is that the pole of the Laplace-transformed generating function with the largest real part controls the asymptotics up to the transition point q_c, and that the R-transform inversion identity is valid on that domain, so if another singularity overtakes the pole earlier, the central formula and its derived transitions would fail.

Editorial extensions

If this is right

  • The full position distribution at long times has an exponential large-deviation form, with a Gaussian centre and a large-x tail shifted by the switching rate.
  • For a Wigner semicircle distribution of diffusivities, all free cumulants beyond the second vanish, so the fourth cumulant grows linearly while higher cumulants do not — a sharp, testable fingerprint.
  • When W(D) vanishes at Dmax, rare and typical trajectories are separated by a light cone in the space-time plane; when it vanishes faster than linearly, an intermediate traveling-front regime appears with a well-defined velocity.
  • In more than one dimension the radial coordinate follows the same large-deviation picture, and correlations between different coordinates grow linearly in time with a coefficient set by the second free cumulant.
  • The finite-time moment formula is exact for every t and every W(D) with finite moments, so the large-time free-cumulant regime can be checked against the exact short-time crossover.

Reading between the lines

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

  • A physical route to free cumulants: if the predicted linear growth is measured, position cumulants of a switching particle provide a direct experimental estimate of kappa_n(D), bypassing the usual moment-to-cumulant inversion.
  • The renewal mechanism suggests a broader universality: any renewal process with exponential waiting times and additive increments whose per-episode variance is drawn from W(D) should share the same R-transform formula, not just Brownian episodes.
  • The edge-exponent classification invites a test of universality: beta distributions with the same exponent nu but different bulk shapes should produce rate functions that are identical up to the global scales set by the mean and the upper edge.
  • The authors' mapping to a mean-field growth model implies the same dynamical transitions should be observable in heterogeneous population growth with redistribution, where the analogue of light cones would separate large- versus small-growth trajectories.
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

2 major / 6 minor

Summary. The manuscript studies a one-dimensional Brownian particle whose diffusion coefficient D(t) switches at rate r to a new value drawn from a fixed distribution W(D). The authors derive an exact finite-time formula for the moments <x^{2n}(t)> for any W(D) with finite moments (Eq. 16), and show that for large t the position cumulants grow linearly in t with coefficients given by the free cumulants of W(D). The scaled cumulant generating function is identified as Ψ(q) = q^2 R(q^2/r), with R the R-transform of W(D). For W(D) ~ (Dmax − D)^ν near the upper edge, the rate function I(y = x/t) is obtained by Legendre transform and exhibits dynamical transitions controlled by ν: smooth behavior for −1 < ν ≤ 0, a light-cone transition for 0 < ν < 1, and an additional traveling-front phase for ν > 1. Explicit results are given for the two-state model, uniform, and Wigner-semicircle distributions, and the formulas are validated with arbitrary-precision numerics.

Significance. If correct, the link between a single-particle classical diffusion model and free probability is surprising and of wide interest; the exact finite-time moment formula is a useful standalone result; and the classification of dynamical transitions (light cone for 0 < ν < 1; traveling front for ν > 1) gives concrete, falsifiable predictions. The paper's strengths include machine-checkable derivations in the SM, no fitted parameters, explicit analytic formulas for the rate functions, and numerical checks that resolve probabilities down to scales of 10^{-2000}. The analytically delicate step—that Ψ(q) is the rightmost singularity of the Laplace-transformed generating function and that the Cauchy-Stieltjes/R-transform inversion extends up to q_c—is adequately supported by the SM's monotonicity argument on the real axis and the absence of complex poles; I found no gap there. However, two stated formulas in the main text need correction (see major comments); both are localized and fixable.

major comments (2)
  1. [Eqs. (5) and (11)] The claimed universal asymptotics Ψ(q) → Dmax q^2 − r as q → ∞ and I(y) → r + y^2/(4Dmax) as y → ∞ are stated for 'any' W(D) with finite support on [0, Dmax], but they are contradicted by the paper's own exact two-state solution. From Eq. (29), for fixed 0 < p < 1, the largest pole behaves as s_+(q) → D1 q^2 − r(1−p) for q → ∞, so the corresponding tail is I(y) ≈ r(1−p) + y^2/(4D1), not r + y^2/(4D1). The physical reason is that each switching event redraws the extremal state D1 with probability p, so the effective rate of leaving the extremal state is r(1−p) rather than r. Equations (5) and (11) should either be restricted to distributions without an atom at Dmax or amended to include the atom weight; as written they are internally inconsistent with Eq. (29).
  2. [Eq. (15) and SM Eq. (194)] The front velocity v = yc − rDmax/yc is inconsistent with the Legendre transform result I(y) = qc y − γ on the intermediate interval. Since pr(x,t) ≈ e^{−tI(x/t)} = e^{−qc x + γ t}, one must have v = γ/qc = Dmax qc − r/qc. For the two-state p → 0 example of Fig. 2 (D1 = 2, D2 = 1, r = 1), the printed formula gives v = 4 − 2/4 = 3.5, whereas Eq. (31) gives I(y) = y − 1 on the intermediate interval, i.e., v = 1. The velocity formula should be corrected in both the main text and the SM.
minor comments (6)
  1. [End Matter, Wigner example] The density is written as W(D) = 8√(D(Dmax−D))/(πDmax), but this integrates to Dmax rather than 1; the correct normalization (used in SM Eq. (179) and in the quoted cumulants) is 8√(D(Dmax−D))/(πDmax^2).
  2. [Eq. (16) description] The Bell polynomial \hat{B}_{n,m} is said to be 'of n − m variables', but it has n − m + 1 arguments; the wording should be corrected.
  3. [Eqs. (6)–(8)] The notation alternates between \hat{p}_r(q,t) and \hat{p}(q,t) for the same generating function; please use a single symbol consistently.
  4. [Abstract and Fig. 2 caption] The phrase 'accuracy up to 10^{-2000}' is ambiguous; clarify whether 10^{-2000} is the smallest probability resolved or the absolute accuracy of the computed rate function.
  5. [SM Eq. (148)] The expression ((y − y1)/(νB1))^{ν/(ν−1)} involves a fractional power of a negative quantity for y < y1; please state that the real branch is intended, or write the formula in terms of (y1 − y).
  6. [Eqs. (26)–(28)] A sentence stating explicitly that the identity g(R(z)+1/z) = z is applied on the real interval where the pole exists (q < qc), with the justification given in SM Section V A, would help the reader.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the free-cumulant/R-transform relation is an external mathematical theorem, and all model-derived quantities are obtained independently of the target claims.

full rationale

The derivation starts from a renewal equation (Eq. (17)) whose exact solution (Eq. (7)) is obtained by Laplace transform, with no fitted parameters. The SCGF is then identified with the rightmost singularity of \tilde p_r(q,s); for finite-support W(D) the supplementary material shows that the pole equation has at most one real root and that no complex poles exist, so the branch-point analysis is controlled. The step \Psi(q)=q^2 R(q^2/r) uses only the standard free-probability identity g(R(z)+1/z)=z (Eq. (27)), cited to external textbooks and theorems, not to the authors' own work; it is a mathematical equivalence rather than an input built to reproduce the result. The companion exact moment formula (Eq. (16)) gives the same free-cumulant coefficients order by order, providing an independent check, and the transition points q_c and y_c are computed from W(D) rather than fitted. Numerical validations target the full functions \Psi(q) and I(y), including probabilities as small as 10^{-2000}. The few self-citations (e.g., Ref. [65] for Weibull universality and Ref. [67] for standard large-deviation theory) are contextual and not load-bearing, so no circular step is present.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no fitted parameters or invented entities. It relies on standard renewal theory, the free-probability R-transform identity, and modeling assumptions on the tail of W(D).

assumptions (5)
  • domain assumption The renewal equation (Eq. 17) is exact for the annealed dynamics with independent exponential holding times and i.i.d. diffusion coefficients.
    The model is defined so that each switching event redraws D from W(D) independently; the renewal equation follows by conditioning on the last switching time.
  • standard math The R-transform identity g(R(z) + 1/z) = z for probability measures on [0, Dmax] (Eq. 27) is used to map the pole equation (Eq. 9) to Psi(q) = q^2 R(q^2/r).
    This is a known theorem of free probability (Voiculescu, Guionnet-Maida); the paper relies on it in Appendix D.
  • standard math The large-time limit of the inverse Laplace transform is dominated by the rightmost singularity (pole or branch point) of the transformed generating function, and the saddle-point inversion of the bilateral Laplace transform is valid.
    Standard Tauberian and saddle-point arguments used to obtain Eq. (8) and the rate function via Legendre transform.
  • domain assumption For the classification of dynamical transitions, W(D) behaves as (Dmax - D)^nu near the upper edge, with nu > -1, and the Cauchy-Stieltjes transform analysis near Dmax is valid.
    The transition scenarios in Eqs. (12)-(15) and Section V of the supplement rest on this local tail behavior.
  • domain assumption All moments of W(D) exist.
    The exact moment formula requires finite moments <D^n> for all n.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Large Deviations in Switching Diffusion: from Free Cumulants to Dynamical Transitions." pith.science (2026). https://pith.science/paper/2GA4GR4C

@misc{pith2026250113754,
  author       = {Pith},
  title        = {Pith review of: Large Deviations in Switching Diffusion: from Free Cumulants to Dynamical Transitions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2GA4GR4C}},
  note         = {Machine review of arXiv:2501.13754}
}
abstract

We study the diffusion of a particle with a time-dependent diffusion constant $D(t)$ that switches between random values drawn from a distribution $W(D)$ at a fixed rate $r$. Using a renewal approach, we compute exactly the moments of the position of the particle $\langle x^{2n}(t) \rangle$ at any finite time $t$, and for any $W(D)$ with finite moments $\langle D^n \rangle$. For $t \gg 1$, we demonstrate that the cumulants $\langle x^{2n}(t) \rangle_c$ grow linearly with $t$ and are proportional to the free cumulants of a random variable distributed according to $W(D)$. For specific forms of $W(D)$, we compute the large deviations of the position of the particle, uncovering rich behaviors and dynamical transitions of the rate function $I(y=x/t)$. Our analytical predictions are validated numerically with high precision, achieving accuracy up to $10^{-2000}$.

Figures

Figures reproduced from arXiv: 2501.13754 by the authors.

Figure 1
Figure 1. FIG. 1. Trajectory of a switching diffusion process in one [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Plot of log(ˆp [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (9 more)
Figure 1
Figure 1. Figure 1: FIG. 1. Schematic description of a typical time evolution of [PITH_FULL_IMAGE:figures/full_fig_p012_1.png]
Figure 2
Figure 2. Figure 2: FIG. 2. Using the explicit formulas for the moments given in Eq ( [PITH_FULL_IMAGE:figures/full_fig_p017_2.png]
Figure 3
Figure 3. Figure 3: FIG. 3. We show here a plot of the rate function in the two-state mo [PITH_FULL_IMAGE:figures/full_fig_p020_3.png]
Figure 4
Figure 4. Figure 4: FIG. 4. Here, we show a plot of the derivative of Ψ( [PITH_FULL_IMAGE:figures/full_fig_p021_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. We illustrate here the two situations described in Eqs. ( [PITH_FULL_IMAGE:figures/full_fig_p026_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Illustration of the singularities of the function ˜p [PITH_FULL_IMAGE:figures/full_fig_p027_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. We show here theoretical predictions (solid lines) agai [PITH_FULL_IMAGE:figures/full_fig_p029_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Plot of [PITH_FULL_IMAGE:figures/full_fig_p031_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9. Plot of [PITH_FULL_IMAGE:figures/full_fig_p035_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Exact height distribution in one-dimensional Edwards-Wilkinson interface with diffusing diffusivity

    cond-mat.stat-mech 2025-02 conditional novelty 7.0 of 10

    For an Edwards-Wilkinson interface with D(t)=B^2(t), the single-point height distribution scales as t^{3/4} and its exact scaling function has exponential tails, a robust feature for general z>1.

  2. Nonequilibrium steady state of Brownian motion in an intermittent potential

    cond-mat.stat-mech 2024-12 conditional novelty 6.0 of 10

    For a Brownian particle in a rapidly switching intermittent trap, the far-tail distribution is a universal exponential and periodic traps show a first-order dynamical phase transition without drift.

Reference graph

Works this paper leans on

26 extracted references · 25 canonical work pages · cited by 2 Pith papers

  1. [1]

    The case −1 < ν ≤ 0 17

  2. [2]

    The rate function I(y) 20 VI

    The case ν > 0 18 B. The rate function I(y) 20 VI. The case where W (D) has an infinite support [0 , +∞) 22 VII. Calculation of the pre-exponential factor of ˆ pr(q, t) and the O(1) corrections to the cumulants 23 VIII. Two specific examples of W (D) with a finite support 24 A. Uniform distribution - Case ν = 0 24 B. Wigner semi-circle distribution - ν = 1/2 25

  3. [3]

    autonomous ratcheting by stochastic resetti ng

    Correction of order O(1) to the cumulants 27 IX. Discussion of the dynamical transitions for bounded W (D) 27 X. Switching diffusion in higher dimensions 28 A. Renewal equation and explicit solution in d dimension 28 B. Mixed cumulants: example in the case d = 2 29 XI. Numerical method 30 A. Numerical evaluation of Ψ( q) 30 B. Numerical evaluation of I(y) ...

  4. [4]

    The case −1 < ν ≤ 0 Hence, for ν < 0, the equation in ( 114) admits a solution s = s∗ for any q and therefore, ˜pr(q, s) has a pole for any q at this value s = s∗ (see Fig. 6). Note in addition that s∗ ≥ q2Dmax − r and in this case the large time behavior of ˆpr(q, t) in Eq. ( 109) is dominated by this pole at s = s∗. Therefore, we have ˆpr(q, t) ≈ t→∞ Re...

  5. [5]

    Hidalgo-Soria, E

    M. Hidalgo-Soria, E. Barkai, S. Burov, Cusp of the non-Gaussian density of particles for a diffusing diffusivity model , Entropy 23, 231 (2021)

  6. [6]

    Indeed, for ν > 0, the function ˜pr(q, s), as a function of s, exhibits a pole only for q2 ≤ q2 c = rgc where gc is given in Eq

    The case ν > 0 This case turns out to be more interesting. Indeed, for ν > 0, the function ˜pr(q, s), as a function of s, exhibits a pole only for q2 ≤ q2 c = rgc where gc is given in Eq. ( 115) while there is no pole for q2 > q 2 c = rgc and in that case the large t limit is instead dominated by the branch point at sb = −r + Dmaxq2 (see Fig. 6). This imp...

  7. [7]

    M. R. Evans, S. N. Majumdar, Diffusion with stochastic resetting , Phys. Rev. Lett. 106, 160601 (2011)

  8. [8]

    near-extreme

    Correction of order O(1) to the cumulants We can use the formula ( 167) to compute the O(1) corrections to the cumulants in the large time limit. Indeed using that the R-transform of the Wigner distribution on [0 , Dmax] is given by R(z) = Dmax 2 + z D2 max 16 , (187) one obtains from ( 167) ˜pr(q, t) ≈ ( 1 − q4 r2 D2 max 16 ) e t ( q2 Dmax 2 +q4 D2 max 1...

Show all 26 references
  1. [9]

    org/ A263633 32

    OEIS Foundation, A263633: https: // oeis. org/ A263633 32

  2. [10]

    Comtet, Advanced Combinatorics: The art of finite and infinite expans ions, Springer Science & Business Media (2012)

    L. Comtet, Advanced Combinatorics: The art of finite and infinite expans ions, Springer Science & Business Media (2012)

  3. [11]

    wikipedia

    https: // en. wikipedia. org/ wiki/ Bell_ polynomials

  4. [12]

    /suppress Luczka, M

    J. /suppress Luczka, M. Niemiec, P. H¨ anggi,First-passage time for randomly flashing diffusion , Phys. Rev. E 52, 5810 (1995)

  5. [13]

    S. N. Majumdar, S. Sabhapandit, G. Schehr, Dynamical transition in the temporal relaxation of stochast ic processes under resetting, Phys. Rev. E 91, 052131 (2015)

  6. [14]

    D. S. Grebenkov, Time-averaged mean square displacement for switching diffu sion, Phys. Rev. E 99, 032133 (2019)

  7. [15]

    5 2 2.5 3 q 0 2 4 6 8log( ˆpr(q, t ))/t Numerics, ν = 4 FIG

    0 log( ˆpr(q, t ))/t Numerics, ν = 0 Ψ( q) 0 1 qc = 2 3 4 q 0.0 4.0 8.0 12.0 16.0log( ˆpr(q, t ))/t Numerics, ν = 1/ 2 Ψ( q) 0 0.5 1 qc = 1. 5 2 2.5 3 q 0 2 4 6 8log( ˆpr(q, t ))/t Numerics, ν = 4 FIG. 7. We show here theoretical predictions (solid lines) agai nst numerical re...

  8. [16]

    M. R. Evans, S. N. Majumdar, G. Schehr, Stochastic resetting and applications , J. Phys. A: Math. Theor. 53, 193001 (2020)

  9. [17]

    Gupta, A

    D. Gupta, A. Pal, A. Kundu Resetting with stochastic return through linear confining p otential, J. Stat. Mech. 043202 (2021)

  10. [18]

    P. K. Ghosh, S. Nayak, J. Liu, Y. Li, F. Marchesoni, Autonomous ratcheting by stochastic resetting , J. Chem. Phys. 159, 031101 (2023)

  11. [19]

    P. C. Bressloff, Switching diffusions and stochastic resetting , J. Phys. A: Math. Theor. 53, 275003 (2020)

  12. [20]

    Santra, U

    I. Santra, U. Basu, S. Sabhapandit, Effect of stochastic resetting on Brownian motion with stoch astic diffusion coefficient , J. Phys. A: Math. Theor. 55, 414002 (2022)

  13. [22]

    Guionnet, M

    A. Guionnet, M. Ma¨ ıdaA Fourier view on the R-transform and related asymptotics of spherical integrals. J. Funct. Anal., 222, 435 (2005)

  14. [23]

    Banica, Methods of free probability , preprint arXiv:2208.07515 (2022)

    T. Banica, Methods of free probability , preprint arXiv:2208.07515 (2022)

  15. [24]

    A. K. Hartmann, S. Majumdar, G. Schehr, The distribution of the maximum of independent resetting Br ownian motions , In Target Search Problems (pp. 357-389). Cham: Springer Nature Sw itzerland (2024)

  16. [25]

    L. N. G. Filon, On a quadrature formula for trigonometric integrals , P. Roy. Soc. Edinb. A. 49, 38 (1930)

  17. [26]

    Bernard, J

    M. Bernard, J. P. Bouchaud, P. L. Doussal, A mean-field theory for heterogeneous random growth with red istribution, arXiv preprint arXiv:2503.23189 (2025)

  18. [120]

    Below, in Section VIII A, we compute Ψ( q) explicitly in the case where W (D) is the uniform distribution over [0, Dmax], corresponding to ν = 0

    that Ψ( q) and its derivatives are continuous functions of q for all real q. Below, in Section VIII A, we compute Ψ( q) explicitly in the case where W (D) is the uniform distribution over [0, Dmax], corresponding to ν = 0

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.