Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

On the localization regime of high-dimensional directed polymers in time-correlated random field

T0 review · 3 major / 3 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read This paper proves that a high-dimensional directed polymer in a time-correlated random field obeys a law of large numbers, has well-defined free energies at all temperatures, and localizes only beyond a positive critical inverse…

desk verdict A promising but unfinished extension of directed polymer results to time-correlated fields; the main LLN is unproven because its claimed limit is never shown to exist. read the letter →

arxiv 2412.14712 v1 pith:GW76JWGT submitted 2024-12-19 math.PR

classification math.PR MSC 60K3582B4460F1060G42
keywords directedpolymertime-correlatedrandomfieldlawoflargenumbersquenchedfreeenergyannealedlocalization-delocalizationtransitionMarkoviancone-mixingcondition
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 studies a directed polymer chain in a random environment that varies with time as well as position, under two exponential-mixing conditions on the field. It claims that in transient dimensions ($d\ge 3$), for every inverse temperature $\beta\ge 0$, the time-averaged environment seen along the polymer path obeys a law of large numbers and converges to a deterministic constant. It further claims that the quenched and annealed free energies exist at all temperatures, that the annealed free energy is differentiable, and that the difference $\rho(\beta)-\lambda(\beta)$ is continuous and non-increasing, so a critical $\beta^*$ is well defined. The main new result is that $\beta^*>0$: weak disorder always survives for small positive $\beta$, and under an entropy-type condition one gets $0<\beta^*<\infty$, meaning a genuine delocalized-localized transition. This matters because time-correlated disorder is substantially harder than the i.i.d. case and its localization picture was previously unresolved.

What carries the argument

The load-bearing construction is a pair of auxiliary random fields plus a block-regeneration scheme that disconnects the time correlation. For an independent field $\epsilon_n\in\{-1,0,1\}$, the paper sets $\eta_{n,z}=2\omega_{n,z}\mathbf{1}_{\{\epsilon_n=0\}}$, so that $E_Q[\eta_{n,z}]=\omega_{n,z}$, and $\xi^{(l)}_{n,z}=-\beta l\mathbf{1}_{\{\epsilon_n=\pm1\}}+\log(2e^{\beta\omega_{n,z}}-e^{-\beta l})\mathbf{1}_{\{\epsilon_n=0\}}$, so that $E_Q[Z^{1,\xi^{(l)}}_N]=Z^{\beta,\omega^{(l)}}_N$ for the truncated field $\omega^{(l)}=\max\{\omega_{n,z},-l\}$. Random times $\tau_n^{(L)}$, defined by long runs of $+1$ followed by $-1$ or $0$, split the path into blocks whose conditional laws are shown, via cone-mixing estimates, to be within $e^{-gtL}$ in total variation of a fixed law. This yields the law of large numbers. For localization, the normalized block partition function $H_{n,\beta}$ is exhibited as a nonnegative martingale; an $L^2$ estimate, a zero-one law, and comparison with $L_{n,\beta}$ give $\rho(\beta)=\lambda(\beta)$ for small $\beta$.

What would settle it

Compute $\gamma_L := E E^S_0[4^{-L}\sum_{k=1}^{\tau_1^{(L)}}\eta_{k,S_k}]$ and $\beta_L := E_Q[4^{-L}\tau_1^{(L)}]$ for the paper's Gaussian example with covariance $\exp(-\|x-y\|_1)G(x,y)$, and check whether $\gamma_L/\beta_L$ converges as $L\to\infty$; if two subsequences give different limits, the deterministic $\ell$ in Theorem 2.3 does not exist. A Monte-Carlo simulation of the polymer path showing that $N^{-1}\sum_{k=1}^N\omega_{k,S_k}$ fails to concentrate across independent environment samples as $N$ grows would likewise disprove the law of large numbers.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that time-correlated disorder does not destroy the basic thermodynamic and path-level structure of directed polymers. Under either the time-correlated condition (TC) or Guo's time-correlated condition (TCG), and for a transient reference random walk, the paper proves that for every fixed $\beta\ge 0$ the moving average $N^{-1}\sum_{k=1}^N\omega_{k,S_k}$ converges $P\otimes P^S_0$-almost surely to a deterministic constant $\ell$; that the quenched free energy $\rho(\beta)$ and annealed free energy $\lambda(\beta)$ exist for all $\beta\ge 0$, with $\lambda$ differentiable on $[0,\infty)$ and $\rho-\lambda$ continuous and non-increasing; and that the critical inverse temperature $\beta^*=\inf\{\beta>0:\rho(\beta)<\lambda(\beta)\}$ is strictly positive. The paper also gives sufficient conditions under which $\beta^*=\infty$ and under which $0<\beta^*<\infty$, and it exhibits a Gaussian field with covariance $\exp(-\|x-y\|_1)G(x,y)$ that realizes the latter case.

Load-bearing premise

The argument's load-bearing premise is that the ratio $\gamma_L/\beta_L$ of expected block averages converges as $L\to\infty$; the paper defines $\ell$ as this limit but never proves it exists, and without that convergence the claimed almost-sure law of large numbers can fail.

Editorial extensions

If this is right

  • For fixed $\beta\ge 0$, the almost-sure limit $\ell=\lim_{N\to\infty}N^{-1}\sum_{k=1}^N\omega_{k,S_k}$ makes the environment self-averaging along the polymer path despite temporal correlations.
  • The free energies $\rho(\beta)$ and $\lambda(\beta)$ exist for every $\beta\ge 0$, with $\lambda$ differentiable and $\rho-\lambda$ continuous and non-increasing, so the weak-disorder/strong-disorder dichotomy is governed by a single critical value $\beta^*$.
  • The inequality $\beta^*>0$ guarantees that in transient dimensions small disorder is always in the delocalized regime under either (TC) or (TCG).
  • If $\lim_{\beta\nearrow\infty}\Lambda(\beta)<K$, the paper's criterion gives $\beta^*=\infty$, meaning the polymer remains delocalized at every temperature.
  • Under the entropy-type condition involving $P(\omega_{1,0}=\hbar)$ and the walk's entropy, $0<\beta^*<\infty$; the Gaussian field with covariance $\exp(-\|x-y\|_1)G(x,y)$ is shown to satisfy this condition.

Reading between the lines

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

  • Editorial inference: one could test numerically whether $\gamma_L/\beta_L$ converges for the paper's Gaussian example, since the proof defines $\ell$ as that limit; an explicit computation showing oscillatory behavior would pinpoint exactly where the law of large numbers needs an additional hypothesis.
  • Editorial inference: the block-regeneration construction is not tied to nearest-neighbor walks and could be extended to $\alpha$-stable long-range walks by replacing the single space-time cone with a countable superposition of cones, yielding analogous LLN and free-energy statements for L\'evy directed polymers.
  • Editorial inference: the constants $\kappa_1,\kappa_2,K'$ in the localization criteria are crude, and the paper itself notes that $K'$ vanishes in the i.i.d. limit; sharpening them might produce a closed-form threshold for $\beta^*$ in terms of single-site exponential moments and walk entropy.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper studies directed polymers in a Markovian random field on N×Z^d satisfying one of two time-correlation conditions, (TC) or (TCG). It claims a law of large numbers for the polymer moving average (Theorem 2.3), existence of quenched and annealed free energies with differentiability of the annealed free energy (Theorem 2.4), a positive-temperature localization regime with β*>0 (Theorem 2.5), and entropy-type criteria for delocalization and 0<β*<∞ (Theorem 2.6). The proof strategy introduces auxiliary fields and regeneration times to isolate blocks of the polymer path and uses cone estimates to control the time-correlated environment.

Significance. The topic is timely and the auxiliary-field/block decomposition is an original technical idea; if the results were fully established, they would be a substantial extension of directed-polymer results beyond independent environments. The paper does not fit parameters to data, and the main claims are stated as falsifiable mathematical theorems. However, several load-bearing steps are incomplete or incoherent as written, so the manuscript in its current form does not establish its central claims.

major comments (3)
  1. [Section 4, after Eq. (4.5)] The proof of Theorem 2.3 defines the deterministic limit as ℓ := lim_{L→∞} γ_L/β_L, but no argument is given for the existence of this limit; Lemmas 3.1 and 3.2 only control each fixed-L block and do not control the evolution of γ_L as L varies, and the estimate (4.5) is only for fixed L and n→∞. Passing from the fixed-L limits to the original time average also requires a uniform-in-L diagonal argument, which is absent. Since Theorem 2.5 later uses the LLN for τ_n^{(L)}/n and the block interpolation, this gap is load-bearing for the central localization claim.
  2. [Section 5, Step IV, around Eq. (5.5)] The definition of ω† is not meaningful as written, and the FKG step is not justified; the measure with density (W^{β,ω†}_N)^{-1} e^{β∑ω†}/E[Z^{β,ω†}_N] is not shown to be monotone in ω, so the claimed E[∂_β log W_N] ≤ 0 does not follow. Consequently the non-increasing property of β↦ρ(β)−λ(β) in Theorem 2.4, and hence the standard definition of β*, is not established.
  3. [Section 7, Lemma 7.1 and Theorem 2.6] The criteria for β*<∞ require ℏ=ess sup ω_{1,0} finite with P(ω_{1,0}=ℏ)>0, but the Gaussian example immediately after Theorem 2.6 has unbounded Gaussian ω_{1,0}, so the claimed 0<β*<∞ for that example does not follow from the stated theorem. The proof of Theorem 2.6 also contains a sign mismatch: the displayed hypothesis uses βλ′(β)−λ(β)>−K(S)H(S_1), whereas the subsequent derivative condition requires βλ′(β)−λ(β)>K(S)H(S_1).
minor comments (3)
  1. [Section 3, cone definition] The definition of C(k,x,γ,ζ) has |vec z−vec x|^2 on the right-hand side, which is dimensionally inconsistent with the linear left-hand side; this appears to be a typo and should be corrected or explained.
  2. [Abstract and Theorem 2.4] The abstract states that smoothness of limiting free energies is proved at all temperature, but Theorem 2.4 asserts differentiability only for the annealed free energy λ, not for the quenched free energy ρ; the wording should be adjusted.
  3. [Section 7, Lemma 7.1] The displayed chain in Lemma 7.1 should use liminf and limsup explicitly; as written, the repeated use of lim over β→∞ is ambiguous.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: main theorems are proved from the stated mixing assumptions; the terse ℓ=lim_L γ_L/β_L step in Theorem 2.3 is a rigor gap, not a circular reduction.

full rationale

I find no circular step. The LLN (Theorem 2.3) is proved via auxiliary fields η and ξ^(l), block times τ_n^(L), and approximation by i.i.d. block averages; the claimed limit ℓ is presented as ℓ = lim_{L→∞} γ_L/β_L in Lemma 4.2 after Eq. (4.5). This is a construction from the model, not a fitted parameter or an input defined by the output. The proof would benefit from an explicit diagonal argument showing the L-limit exists and commutes with the n→∞ limit; the manuscript is terse or incomplete there, but that is a correctness/rigor concern, not equivalence-by-construction. The auxiliary-field construction and Lemma 3.2 are derived from the (TC)/(TCG) assumptions and do not presuppose Theorem 2.3. Theorem 2.4 uses external results [30] plus martingale and convexity arguments; Theorem 2.5 uses square-integrable martingale estimates and an external zero-one law [6], not a self-citation chain. The only self-citation, [8], is historical ("later adopted by ... myself [8]") and is not load-bearing. No uniqueness theorem from the authors is invoked, no known result is renamed, and no parameter is fitted to a subset of data and then called a prediction. Thus there is no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 3 invented entities

The central results depend on the model assumptions (Markovian field with exponential-moment condition and either TC or TCG mixing, transient finite-range walk) and on several external probabilistic theorems taken as black boxes (quenched free energy existence [30], martingale LDP [24], Khas'minskii lemma, FKG). The paper also introduces hand-chosen constants (cone angle, cutoffs, threshold K) that are existential rather than fitted to data. The auxiliary fields eta, xi^{(l)}, and omega-dagger are internal proof devices without independent predictive content.

free parameters (3)
  • r0
    Defined in Lemma A.3 as min{l > d : (pi^2/6) l^2 e^{-g t l} K1 l^2 < 1}; a cutoff chosen by hand so that the Khas'minskii series converges. The proof of Lemma 6.1 depends on its existence, not its numerical value.
  • zeta0 (cone aperture)
    Chosen in Section 3 to be sufficiently small so that the finite-range walk satisfies S_{n+1} in C(n, S_n, gamma, zeta). Any sufficiently small positive value works; it is an existence choice.
  • Threshold K in Lemma 6.1
    An absolute constant 'specified in Appendix A' for the condition Lambda(beta) < K that guarantees L2-boundedness of the approximate martingale. Only existence is used.
assumptions (4)
  • domain assumption The random field omega is Markovian and satisfies either (TC)_{C,g} or (TCG)_{C,g} (Definitions 2.1, 2.2).
    This is the model class under study; all theorems are conditional on these mixing conditions.
  • domain assumption Exponential moments: E exp(beta omega_{n,z}) < infinity for all beta in R (condition (2.1)).
    Used throughout to justify Laplace transforms and martingale moment bounds; stated as a standing assumption.
  • domain assumption The reference random walk is transient (d >= 3) with finite-range increments (||S1||_1 < infinity).
    Transience is used in Lemma A.1 via the Green function bound; finite range gives p_S > 0 in Theorem 2.6.
  • standard math Quenched free energy existence is quoted from [30, Theorem 2.3]; martingale LDP from [24, Theorem 3.2]; Khas'minskii lemma and FKG inequality are used as black boxes.
    External theorems invoked without reproof in Sections 4, 5 and Appendix A.
invented entities (3)
  • Auxiliary field eta (3.1a)
    purpose: Randomly thins the time-correlated field (sets most entries to 0) so that E_Q[eta] = omega and block sums become nearly independent; used to prove the LLN.
    Internal proof device with no falsifiable physical prediction.
  • Auxiliary field xi^{(l)} (3.1b)
    purpose: Replaces the truncated field omega^{(l)} to connect the polymer measures at different truncation levels; central to the localization proof.
    Internal proof device.
  • Integrated field omega-dagger
    purpose: Introduced in Section 5, Step IV to cancel distant time-correlation so that an FKG inequality can be applied to prove monotonicity of rho-lambda.
    Its definition, E[omega_{n,z}|...]-style integration over other coordinates, is unclear and no independent evidence is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the localization regime of high-dimensional directed polymers in time-correlated random field." pith.science (2026). https://pith.science/paper/GW76JWGT

@misc{pith2026241214712,
  author       = {Pith},
  title        = {Pith review of: On the localization regime of high-dimensional directed polymers in time-correlated random field},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GW76JWGT}},
  note         = {Machine review of arXiv:2412.14712}
}
read the original abstract

This paper describes directed polymer on general time-correlated random field. Law of large numbers, existence and smoothness of limiting free energies are proved at all temperature. We also display the delocalized-localized transition, via separating techniques for entanglement of the random field.

Figures

Figures reproduced from arXiv: 2412.14712 by the authors.

Figure 1
Figure 1. An illustration of simple random in transient dimensions. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

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. Splitting algorithm and normed convergence for drawing the random Loewner curves

    math.PR 2025-07 reject novelty 5.0 of 10

    The Ninomiya-Victoir splitting output is claimed to converge to SLE in sup-norm and in a weak Lp sense, and is used to draw fractional and noise-reinforced Loewner curves.

Reference graph

Works this paper leans on

44 extracted references · 42 canonical work pages · cited by 1 Pith paper

  1. [1]

    Albeverio, X

    S. Albeverio, X. Y. Zhou. A martingale approach to directed polymers in a random environment. J. Theor. Probab., 9: 171–189, 1996

  2. [2]

    A. D. Barbour, L. Holst, S. Janson. Poisson Approximation. Oxford Studi. Probab., Oxford Univ. Press, 1st ed., 1992

  3. [3]

    Bazaes, C

    R. Bazaes, C. Mukherjee, A. F. Ram ´ ırez, S. Saglietti. The effect of disorder on quenched and averaged large deviations for random walks in random environments: Boundary behavior. Stoch. Process. Appl., 158: 208–237, 2023. 28

  4. [4]

    H. Berbee. Convergence rates in the strong law for bounded mixing sequences. Probab. Theor. Relat. Field., 74: 255–270, 1987

  5. [5]

    Bezerra, S

    S. Bezerra, S. Tindel, F. Viens. Superdiffusivity for a Brownian polymer in a continuous Gaussian environment. Ann. Probab., 36 (5): 1642–1675, 2008

  6. [6]

    Bolthausen

    E. Bolthausen. A note on the diffusion of directed polymers in a random environment. Commun. Math. Phys., 123: 529–534, 1989

  7. [7]

    Carmona, Y

    P. Carmona, Y. Hu. On the partition function of a directed polymer in a Gaussian random environment. Probab. Theor. Relat. Field., 124: 431–457, 2002

  8. [8]

    J. Chen. Large deviations of quenched and annealed random walk in mixing random environment. arXiv preprint, arXiv: 2409.06581, 2024

Show all 44 references
  1. [9]

    F. Comets. Weak disorder for low dimensional polymers: The model of stable laws. Markov Process. Relat. Field., 13: 681–689, 2006

  2. [10]

    F. Comets. Directed Polymers in Random Environments. Springer ´Ecole d’ ´Et´ e de Probab. de Saint- Flour XL VI,2017

  3. [11]

    Comets, T

    F. Comets, T. Shiga, N. Yoshida. Directed polymers in a random environment: path localization and strong disorder. Bernoulli, 9 (4): 705–723, 2003

  4. [12]

    Comets, N

    F. Comets, N. Yoshida. Directed polymers in random environment are diffusive at weak disorder. Ann. Probab., 34 (5): 1746–1770, 2006

  5. [13]

    Comets, O

    F. Comets, O. Zeitouni. A law of large numbers for random walks in random mixing environments. Ann. Probab., 32 (1B): 880–914, 2004

  6. [14]

    Cover, J

    T. Cover, J. A. Thomas. Elements of Information Theory. Wiley Ser. Telecommun. Signal Process., 2nd ed., 2006

  7. [15]

    Da Prato, J

    G. Da Prato, J. Zabczyk. Ergodicity for Infinite Dimensional Systems. London Math. Soc. Lect. Note Ser., 229, 1996

  8. [16]

    Dembo, O

    A. Dembo, O. Zeitouni. Large Deviations Techniques and Applications. Springer Stoch. Model. Appl. Probab., 2nd ed., 38, 2010

  9. [17]

    Gˆ ateaux

    R. Gˆ ateaux. Fonctions d’une infinit´ e de variables ind´ ependantes.Bull. Soc. Math. France, 47: 70–96, 1919

  10. [18]

    Guerra Aguilar

    E. Guerra Aguilar. On the transient (T) condition for random walk in mixing environment. Ann. Probab., 47 (5): 3003–3054, 2019

  11. [19]

    Guerra Aguilar, A

    E. Guerra Aguilar, A. F. Ram ´ ırez. Asymptotic direction for random walks in mixing random environ- ments. Electron. J. Probab., 22 (92): 1–41, 2017

  12. [20]

    X. Guo. On the limiting velocity of random walks in mixing random environment. Ann. Inst. Henri Poincar´ e, Probab. Stat.,50 (2): 375–402, 2014

  13. [21]

    D. A. Huse, C. L. Henley. Pinning and roughening of domain walls in Ising systems due to random impurities. Phys. Rev. Lett., 54: 2708, 1985

  14. [22]

    G. L. Karakostas. An extension of H¨ older’s inequality and some results on infinite products. Indian J. Math., 50 (2): 303–307, 2008

  15. [23]

    H. Lacoin. Influence of spatial correlation for directed polymers. Ann. Probab., 39 (1): 139–175, 2011

  16. [24]

    Lesigne, D

    E. Lesigne, D. Voln´ y. Large deviations for martingales. Stoch. Process. Appl., 96 (1): 143–159, 2001

  17. [25]

    T. M. Liggett. Interacting Particle Systems. Grundlehren Math. Wiss., 276, 1985. 29

  18. [26]

    Magdziarz, W

    M. Magdziarz, W. Szczotka, P. ˙Zebrowski. Asymptotic behaviour of random walks with correlated temporal structure. Proc. Royal Soc. A, 46920130419, 2013

  19. [27]

    Medina, T

    E. Medina, T. Hwa, M. Kardar, Y.-C. Zhang. Burgers equation with correlated noise: Renormalization- group analysis and applications to directed polymers and interface growth. Phys. Rev. A, 39: 3053, 1989

  20. [28]

    Miura, Y

    M. Miura, Y. Tawara, K. Tsuchida. Strong and weak disorder for L´ evy directed polymers in random environment. Stoch. Analy. Appl., 26 (5): 1000–1012, 2008

  21. [29]

    G. Rang, J. Song, M. Wang. Scaling limit of a long-range random walk in time-correlated random environment. arXiv preprint, arXiv: 2210.01009, 2022

  22. [30]

    Rassoul-Agha, T

    F. Rassoul-Agha, T. Sepp¨ al¨ ainen, A. Yilmaz. Quenched free energy and large deviations for random walks in random potentials. Commun. Pure Appl. Math., 66: 202–244, 2013

  23. [31]

    Rassoul-Agha, T

    F. Rassoul-Agha, T. Sepp¨ al¨ ainen. A Course on Large Deviations with an Introduction to Gibbs Mea- sures. Grad. Studi. Math., 162, 2015

  24. [32]

    Rovira, S

    C. Rovira, S. Tindel. On the Brownian-directed polymer in a Gaussian random environment. J. Funct. Analy., 222 (1): 178–201, 2005

  25. [33]

    R. T. Rockafellar. Convex Analysis. Princeton Math. Ser., 2nd ed., 1970

  26. [34]

    M. I. Smith, V. Bertola. Effect of polymer additives on the wetting of impacting droplets. Phys. Rev. Lett., 104: 154502, 2010

  27. [35]

    Str¨ omberg

    T. Str¨ omberg. A note on the differentiability of conjugate functions. Arch. der Math., 93: 481–485, 2009

  28. [36]

    Sznitman, M

    A.-S. Sznitman, M. P. W. Zerner. A law of large numbers for random walks in random environment. Ann. Probab., 27 (4): 1851–1869, 1999

  29. [37]

    Thorisson

    H. Thorisson. Coupling, Stationarity, and Regeneration. Probab. Its Appl., 1st ed., 2000

  30. [38]

    R. Viveros. Directed polymer for very heavy tailed random walks. Ann. Appl. Probab., 33 (1): 490–506, 2023

  31. [39]

    Vysotsky

    V. Vysotsky. When is the rate function of a random vector strictly convex? Electron. Commun. Probab., 26: 1–11, 2021

  32. [40]

    Werner, E

    W. Werner, E. Powell. Lecture notes on the Gaussian free field. Soc. Math. France, Cours Sp´ ecial., 28, 2021

  33. [41]

    Zagrodny

    D. Zagrodny. On the weak ∗ convergence of subdifferentials of convex functions. J. Convex Analy., 12 (1): 213–219, 2005

  34. [42]

    M. P. W. Zerner, F. Merkl. A zero-one law for planar random walks in random environment. Ann. Probab., 29 (4): 1716–1732, 2001

  35. [43]

    M. A. Zorn. Characterization of analytic functions in Banach spaces. Ann. Math., 46 (4): 585–593, 1945

  36. [44]

    M. A. Zorn. Derivatives and Fr´ echet differentials. Bull. American Math. Soc., 52: 133–137, 1946. 30

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.