Pith. sign in

REVIEW 2 major objections 5 minor 48 references

Negative association of Busemann functions in exponential last-passage percolation

T0 review · 2 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read In exponential last-passage percolation, Busemann increments across arbitrary directions are negatively associated: raising one set reduces any monotone function of a disjoint set.

desk verdict The paper proves a plausible and new negative-association result for Busemann increments in exponential LPP, but the proof of the key finite-volume lemma appears to have a sign error that reverses the required inequality; the main theorem is unsupported as written. read the letter →

arxiv 2608.07236 v1 pith:FLMMP4NW submitted 2026-08-07 math.PR math-phmath.MP

classification math.PRmath-phmath.MP MSC 60K3560K37
keywords last-passagepercolationKardar–Parisi–ZhangBusemannfunctionnegativeassociationrandomgrowthqueueingmapsBurkepropertyexponentialweights
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

In exponential last-passage percolation, Busemann increments along a down-right path are independent only when the asymptotic direction parameters are ordered a certain way. This paper shows that the dependence appearing for arbitrary directions is one-sided: any coordinate-wise monotone function of one group of Busemann increments is negatively correlated with any monotone function of a disjoint group. That negative-association inequality is strong enough to dominate the joint moment-generating function by a product of marginals, and hence to yield exponential concentration at the diffusive scale $\sqrt n$ for sums of Busemann increments with mixed directions. The result matters because negative correlation, rather than independence, is the property that Busemann functions are expected to have in general, and it is what a heuristic argument for the KPZ fluctuation relation $\xi = 2 \chi$ requires.

What carries the argument

The engine is the description of the joint Busemann process as increments of an LPP with an exponential boundary. The upward queueing map $\sigma_m$, a Pitman-type transform, swaps the rates of two adjacent exponential levels by the Burke property, and the paper applies a sequence of these maps to produce a reference environment $h^*$ in which the distinguished increment becomes a single boundary weight. Braid relations let the maps be rearranged so that all other Busemann increments become monotone functions of that weight, and the Harris–FKG inequality then converts this monotone dependence into the negative-association estimate. All steps except the Burke property are deterministic and work for arbitrary weights.

What would settle it

Using the explicit two-edge joint law in the proof of Proposition 1.10 (with independent $X_1, X_2 \sim \mathrm{Exp}(\rho)$, $I_1, I_2 \sim \mathrm{Exp}(\lambda)$, $J \sim \mathrm{Exp}(\rho-\lambda)$), verify whether $\mathbb{P}(B^\lambda_{-e_1,0} > s, B^\rho_{0,e_1} > t)$ is at most the product of its marginals for every $s,t > 0$; a single pair $(s,t)$ with the reverse strict inequality would contradict Theorem 1.1. The same check can be run on simulated non-exponential weights to see whether the Burke-property restriction is actually needed.

Watch

Extended reading notes

Core claim

Theorem 1.1 states that if $B_1,\ldots,B_n$ are Busemann increments along a down-right path in exponential LPP, with arbitrary direction parameters in $(0,1)$, and $R_1,\ldots,R_k$ are Busemann functions whose endpoints lie in rectangles between those edges, then for any non-negative functions $f_i$ and $g$ that are coordinate-wise monotone in the same direction, $\mathbb{E}[g(R)\prod_i f_i(B_i)] \le \mathbb{E}[g(R)]\prod_i \mathbb{E}[f_i(B_i)]$. The inequality says that raising one set of Busemann increments can only reduce any monotone functional of a disjoint set. The paper also proves strictness: for $0<\lambda<\rho<1$, $\mathrm{Cov}(B^\lambda_{-e_1,0}, B^\rho_{0,e_1})<0$, so the correlation is genuinely nonzero, and the theorem yields a weaker form of negative association between a single increment and a whole block of other Busemann functions.

Load-bearing premise

The exponential distribution of the vertex weights is the load-bearing assumption: only the Burke property, which swaps and preserves the exponential rates across a queueing map, makes the rate-swapping construction of the environment $h^*$ valid, and without it the reduction to the finite-volume boundary LPP collapses.

Editorial extensions

If this is right

  • Joint moment-generating functions of Busemann increments along a down-right path are dominated by the product of the individual MGFs (Corollary 1.3).
  • Sums of Busemann increments with arbitrary directions concentrate exponentially on the $\sqrt n$ scale, with constants depending only on the minimum rate $\varepsilon$ (Corollary 1.4).
  • Busemann increments along a down-right path satisfy negative orthant dependence for upper and lower tails (Corollary 1.5).
  • A Busemann increment in direction $\lambda$ is independent of the $\sigma$-algebra of all Busemann functions whose directions and positions fall on the appropriate side of it (Theorem 1.8).
  • The dependence is nonzero: adjacent increments in two distinct directions have strictly negative covariance (Proposition 1.10).

Reading between the lines

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

  • Because every ingredient except the Burke property is weight-independent, the same argument should yield negative association in other exactly solvable models with a Burke-type rate-swapping mechanism, such as inverse-gamma polymers or the KPZ equation.
  • The negative-orthant-dependence corollary is the kind of one-sided concentration that the heuristic for $\xi = 2\chi$ needs; this paper supplies it for mixed-direction increments, not just the equal-direction case.
  • The finite-volume inhomogeneous LPP whose increments match the Busemann process must inherit the negative-association inequality; a direct proof there would bypass Busemann limits and could extend to finite-size or perturbative settings.
  • Corollary 1.4 is an exponential tail bound on sums of increments without independence; checking whether the constants are optimal would test how tight the negative-association inequality is.
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 / 5 minor

Summary. The paper studies exponential last-passage percolation and proves that Busemann increments along a down-right path with arbitrary direction parameters are negatively associated with Busemann functions located in the intervening rectangles. The main theorem (Theorem 1.1) states that for coordinate-wise monotone functions f_i and g of the same direction, E[g(R)\prod f_i(B_i)] ≤ E[g(R)]\prod E[f_i(B_i)]. From this the authors derive product-form domination of moment generating functions, a diffusive exponential concentration inequality for sums of Busemann increments, negative orthant dependence, and an improved independence statement for a single Busemann function. The proof passes through a finite-volume boundary LPP model: a rate-swapping environment h* is built by queueing maps using the Burke property, the target increments are rewritten in h*, and monotonicity lemmas plus the Harris–FKG inequality yield the covariance inequality. The exponential distribution enters only through the Burke property; the remaining ingredients are deterministic.

Significance. If correct, the result provides a systematic one-sided dependence structure for multi-directional Busemann processes in an exactly solvable model, going beyond the known independence for monotone direction parameters. The applications, especially the concentration inequality at diffusive scale, are concrete and would be useful for KPZ fluctuation-exponent arguments. Strengths of the paper include a self-contained development of the queueing-map toolkit, explicit deterministic monotonicity lemmas, a clear separation of the exponential-specific Burke step, and several falsifiable corollaries. The main caveat is a sign inconsistency in the key monotonicity lemma, which appears correctable but must be fixed before the proof is valid.

major comments (2)
  1. [Section 3.1, Lemma 3.2] The horizontal-edge case contains a sign error that is load-bearing for Proposition 3.1. For an edge with d−c=e1 and c=(i−1,t), d=(i,t), the target inequality (3.20) is G^S_{−j,c}(dec)−G^S_{−j,d}(dec) ≥ G^S_{−j,c}(h)−G^S_{−j,d}(h), which is equivalent to −I^{S,−j}_{(i,t)}(dec) ≥ −I^{S,−j}_{(i,t)}(h), i.e. I^{S,−j}_{(i,t)}(dec) ≤ I^{S,−j}_{(i,t)}(h). The proof, however, states 'I^{S,−j}_{(i,t)}(π_{t,n−1}eh_dec) ≥ I^{S,−j}_{(i,t)}(π_{t,n−1}eh) for each i∈Z_{>0}' and calls this '(3.20)'. The corner-flipping relation (2.15), together with the increased J-increment in (3.23), in fact yields the required ≤ direction, so the intended argument is recoverable. As written, the displayed conclusion is reversed and the monotonicity feeding into the FKG step (3.18) has the wrong sign; without correcting this, Proposition 3.1 and hence Theorem 1.1 are unsupported.
  2. [Section 2.5, Proposition 2.13] The proof of the braid relation is central because identity (3.8) is used repeatedly in the proof of Proposition 3.1, but the long calculation for (σ1σ2h)_1 contains several reindexing steps (for example, changes between z2≤z1 and z2≤z1−1, and conversions between suprema and infima) that are not fully explained. Please expand these steps or provide a precise reference that covers the vertex-weight, centered-Pitman convention used here, so that the equality chain can be verified without reverse-engineering the indices.
minor comments (5)
  1. [Section 3.1] After correcting the sign in the horizontal-edge case, the sentence 'This is exactly (3.20)' should be rechecked: with I^{S,−j}_{(i,t)}(dec) ≤ I^{S,−j}_{(i,t)}(h), the displayed inequality is indeed equivalent to (3.20), but currently the text asserts the opposite inequality.
  2. [Section 3.2] There is a typo: 'by defintion' should be 'by definition'.
  3. [Section 4.1] In the proof of Theorem 1.1, after reducing general endpoints to unit-edge increments, the text says 'Proposition 3.1 can also be applied'; this should be 'Proposition 4.2'.
  4. [Section 1.3] Reference [9] is listed as 'To appear' with a DOI; if the final publication data are available, please update the citation.
  5. [General] The notation for the operators π_{m,n−1} and π^{m,n−1} is introduced only inside the proof of Proposition 3.1; since the same notation is used in the statement of the lemmas, it would help to define it just before Proposition 3.1 or in Section 2.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: all load-bearing stochastic inputs are external theorems (Burke property, joint Busemann law) whose assumptions do not include the target inequality.

full rationale

The paper's central claim, Theorem 1.1, is a negative association inequality for Busemann functions. The proof chain has two genuinely external stochastic inputs: Proposition 2.15 (Burke property, cited to [48]) and Proposition 4.1 (joint distribution of Busemann functions, cited to [29]). Neither input is a restatement of the target result. The Burke property states that, for independent exponential levels with rates lambda < rho, the upward queueing map preserves independence and swaps the rates; this is a standard M/M/1 queueing fact whose stated assumptions do not include negative association or any conclusion about monotone functions of Busemann increments. It is used only to transport rate parameters in the construction of h*, leading to the distributional identity (3.17). The joint-distribution result [29] is an external representation theorem identifying the Busemann process with increments of boundary LPP; it is not derived from or equivalent to Theorem 1.1. The remaining proof ingredients—queueing-map braid relations, monotonicity, and the Harris–FKG inequality—are deterministic or classical and do not smuggle in the desired conclusion. There are no fitted parameters labeled predictions, no self-defined quantities, and no uniqueness theorem imported from the authors' own work. The co-author's prior paper [48] is cited for the Burke property and for a baseline independence result, but this is real independent support: the Burke property is externally falsifiable and does not contain the target theorem. A skeptical sign-error allegation concerning Lemma 3.2 would be a mathematical correctness concern, not circularity, and is therefore out of scope for this pass. Consequently, the derivation is self-contained relative to external benchmarks and the circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard domain assumptions for exponential LPP and on three cited results: existence of Busemann functions, the joint-distribution identity of [29], and the Burke property of [48]. No parameters are fitted to data and no new entities are introduced. The Burke property is the only exponential-specific axiom; the other ingredients are deterministic or hold for arbitrary weights.

assumptions (4)
  • domain assumption Existence of Busemann functions as almost sure limits in exponential LPP (equation (1.3)), established in [33] and [42, Theorem 4.2(iii)].
    Invoked throughout, e.g., in (1.4) for marginal distributions; the paper does not re-prove existence.
  • domain assumption The joint distribution of the Busemann process equals the increments of an LPP model with exponential boundary conditions (Proposition 4.1, from [29, Theorem 3.2 and Lemma 3.3]).
    This identity is the bridge that allows the main theorem to be reduced to the finite-volume boundary LPP analyzed in Section 3.
  • domain assumption Burke property for exponential M/M/1 queues: for two independent exponential levels with rates lambda < rho, the upward queueing map sigma_m preserves independence and swaps the rates (Proposition 2.15, from [48, Proposition 3.5]).
    The rate-swapping construction of h* in Section 3 and the distributional equality (3.17) depend entirely on this property; it is the unique exponential-specific input.
  • domain assumption Monotonicity of LPP increments with respect to the environment (Proposition 2.14, from [4, Lemma B.1]): raising a boundary weight raises the horizontal I-increments and lowers the vertical J-increments.
    Used in the proofs of Lemmas 3.2 and 3.3 to compare the original and decreased environments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Negative association of Busemann functions in exponential last-passage percolation." pith.science (2026). https://pith.science/paper/FLMMP4NW

@misc{pith2026260807236,
  author       = {Pith},
  title        = {Pith review of: Negative association of Busemann functions in exponential last-passage percolation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FLMMP4NW}},
  note         = {Machine review of arXiv:2608.07236}
}
read the original abstract

One hallmark of exactly solvable KPZ random growth models is product-form invariant measures. In the setting of exponential last-passage percolation (LPP), this corresponds to the independence of Busemann increments along any down-right path. However, this independence breaks down when multiple asymptotic directions are considered simultaneously, owing to the fact that jointly invariant measures are not jointly product-form. This paper shows that the failure of independence is one-sided: Busemann increments across arbitrary directions are negatively associated. As an application, we derive an exponential concentration inequality for sums of Busemann increments on the diffusive scale, even when the increments are not independent. While our argument relies on a Burke property that is special to exponential weights, all other proof ingredients-including hidden LPP monotonicities and braid relations for queueing maps-hold for arbitrary weights.

Figures

Figures reproduced from arXiv: 2608.07236 by the authors.

Figure 1
Figure 1. A special case of our main result. Consider Busemann increments [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Setting of Theorem 1.1, in the case n = 2. Increasing the Busemann increments B1 and B2 (shown in blue), tends to probabilistically decrease the Busemann functions within any of the gray regions. This diagram shows k = 7 such Busemann functions (in light red), which can have any directions and any vertex pairs (aj , bj ) that respect the down-right order ⪯. Remark 1.2 (Terminology disclaimer). In general, random var… view at source ↗
Figure 3
Figure 3. Illustration of Remark 1.7. On the left, the direction parameters are non-increasing (ρi ≥ ρi+1), meaning the asymptotic slopes are non-decreasing and thus cross one another. On the right, the direction parameters are allowed to be non-monotone. Finally, we state a proposition demonstrating that the inequality in Theorem 1.1 can be strict, which shows the correlation structure of Busemann functions across different … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The passage value of a path is preserved when switching from vertex [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: The setup of the LPP with boundary in Proposition [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 6
Figure 6. Figure 6: Weights used in “corner-flipping” induction. The [PITH_FULL_IMAGE:figures/full_fig_p025_6.png]
Figure 7
Figure 7. Figure 7: Rewriting the two-line LPP (shown in dashed red) from [PITH_FULL_IMAGE:figures/full_fig_p026_7.png]
Figure 8
Figure 8. Figure 8: If the weight on the red dashed edge increases, then the blue vertical [PITH_FULL_IMAGE:figures/full_fig_p028_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 41 canonical work pages

  1. [1]

    Alevy, I., and Krishnan, A.Negative correlation of adjacent Busemann increments.Ann. Inst. Henri Poincar´ e Probab. Stat. 58, 4 (2022), 1942–1958.https://doi.org/10.1214/21-aihp1236. 3, 7

  2. [2]

    Auffinger, A., and Damron, M.The scaling relation χ = 2ξ− 1 for directed polymers in a random environment.ALEA Lat. Am. J. Probab. Math. Stat. 10, 2 (2013), 857–880. https://alea.impa.br/ articles/v10/10-36.pdf. 2

  3. [3]

    Auffinger, A., and Damron, M.A simplified proof of the relation between scaling exponents in first-passage percolation.Ann. Probab. 42, 3 (2014), 1197–1211. https://doi.org/10.1214/13-AOP854. 2

  4. [4]

    Sigma 8(2020), article no

    Bal´azs, M., Busani, O., and Sepp ¨al¨ainen, T.Non-existence of bi-infinite geodesics in the exponential corner growth model.Forum Math. Sigma 8(2020), article no. e46, 34 pp. https: //doi.org/10.1017/fms.2020.31. 7, 17

  5. [5]

    Theory Related Fields 180, 1-2 (2021), 113–162

    Bal´azs, M., Busani, O., and Sepp ¨al¨ainen, T.Local stationarity in exponential last-passage percolation.Probab. Theory Related Fields 180, 1-2 (2021), 113–162. https://doi.org/10.1007/ s00440-021-01035-7. 7

  6. [6]

    Bal´azs, M., Cator, E., and Sepp ¨al¨ainen, T.Cube root fluctuations for the corner growth model associated to the exclusion process.Electron. J. Probab. 11(2006), article no. 42, 1094–1132. https://doi.org/10.1214/EJP.v11-366. 2, 6

  7. [7]

    Bal´azs, M., Quastel, J., and Sepp ¨al¨ainen, T.Fluctuation exponent of the KPZ/stochastic Burgers equation.J. Amer. Math. Soc. 24, 3 (2011), 683–708. https://doi.org/10.1090/S0894-0347-2011- 00692-9. 3

  8. [8]

    Bal´azs, M., and Sepp ¨al¨ainen, T.Order of current variance and diffusivity in the asymmetric simple exclusion process.Ann. of Math. (2) 171, 2 (2010), 1237–1265. https://doi.org/10.4007/annals. 2010.171.1237. 2

Show all 48 references
  1. [9]

    Theory Related Fields, 66 pp

    Bates, E., Emrah, E., Martin, J., Sepp ¨al¨ainen, T., and Sorensen, E.Permutation invariance in last-passage percolation and the distribution of the Busemann process.Probab. Theory Related Fields, 66 pp. To appear (available online). https://doi.org/https://doi.org/10.1007/s00...

  2. [10]

    L., and Sepp¨al¨ainen, T.Intertwining the Busemann process of the directed polymer model.Electron

    Bates, E., F an, W.-T. L., and Sepp¨al¨ainen, T.Intertwining the Busemann process of the directed polymer model.Electron. J. Probab. 30(2025), article no. 50, 80 pp. https://doi.org/10.1214/25- ejp1310. 8

  3. [11]

    Biane, P., Bougerol, P., and O’Connell, N.Littelmann paths and Brownian paths.Duke Math. J. 130, 1 (2005), 127–167.https://doi.org/10.1215/S0012-7094-05-13014-9. 14

  4. [12]

    Busani, O.Diffusive scaling limit of the Busemann process in last passage percolation.Ann. Probab. 52, 5 (2024), 1650–1712.https://doi.org/10.1214/24-aop1681. 8

  5. [13]

    L.Universality of the geodesic tree in last passage percolation.Ann

    Busani, O., and Ferrari, P. L.Universality of the geodesic tree in last passage percolation.Ann. Probab. 50, 1 (2022), 90–130.https://doi.org/10.1214/21-aop1530. 7

  6. [14]

    Busani, O., Sepp ¨al¨ainen, T., and Sorensen, E.Scaling limit of multi-type invariant measures via the directed landscape.Int. Math. Res. Not. IMRN, 17 (2024), 12382–12432. https://doi.org/10. 1093/imrn/rnae168. 8

  7. [15]

    Busani, O., Sepp ¨al¨ainen, T., and Sorensen, E.The stationary horizon and semi-infinite geodesics in the directed landscape.Ann. Probab. 52, 1 (2024), 1–66. https://doi.org/10.1214/23-aop1655. 8

  8. [16]

    Cator, E., and Groeneboom, P.Second class particles and cube root asymptotics for Hammersley’s process.Ann. Probab. 34, 4 (2006), 1273–1295.https://doi.org/10.1214/009117906000000089. 2 35

  9. [17]

    Cator, E., and Pimentel, L. P. R.Busemann functions and equilibrium measures in last passage percolation models.Probab. Theory Related Fields 154, 1-2 (2012), 89–125. https://doi.org/10.1007/ s00440-011-0363-6. 6

  10. [18]

    Chatterjee, S.The universal relation between scaling exponents in first-passage percolation.Ann. of Math. 177, 2 (2013), 663–697.https://doi.org/10.4007/annals.2013.177.2.7. 2

  11. [19]

    Electron

    Chaumont, H., and Noack, C.Characterizing stationary 1 + 1 dimensional lattice polymer models. Electron. J. Probab. 23(2018), article no. 38, 19 pp.https://doi.org/10.1214/18-EJP163. 3

  12. [20]

    Chaumont, H., and Noack, C.Fluctuation exponents for stationary exactly solvable lattice polymer models via a Mellin transform framework.ALEA Lat. Am. J. Probab. Math. Stat. 15, 1 (2018), 509–547. https://doi.org/10.30757/alea.v15-21. 3

  13. [21]

    Ciech, F., and Georgiou, N.Order of the variance in the discrete Hammersley process with boundaries.J. Stat. Phys. 176, 3 (2019), 591–638. https://doi.org/10.1007/s10955-019-02314-3 . 3

  14. [22]

    Corwin, I., Gu, Y., and Sorensen, E.Periodic Pitman Transforms and Jointly Invariant Measures. Comm. Math. Phys. 407, 3 (2026), Paper No. 49, 157. https://doi.org/10.1007/s00220-025-05541-w . 8

  15. [23]

    Das, S., Sorensen, E., and Yang, Z.Invariant measures for half-space geometric LPP: classification and the one force–one solution principle. 2026. Preprint, 61 pp. https://doi.org/10.48550/arXiv. 2605.22787. 8

  16. [24]

    229, 2 (2022), 201–285.https://doi.org/10.4310/acta.2022.v229.n2.a1

    Dauvergne, D., Ortmann, J., and Vir ´ag, B.The directed landscape.Acta Math. 229, 2 (2022), 201–285.https://doi.org/10.4310/acta.2022.v229.n2.a1. 13

  17. [25]

    Preprint, 75 pp.https://arxiv.org/abs/2405.00194

    Dauvergne, D., and Vir ´ag, B.The directed landscape from Brownian motion, 2024. Preprint, 75 pp.https://arxiv.org/abs/2405.00194. 14

  18. [26]

    Dauvergne, D., and Zhang, L.The directed landscape in half-space. 2026. Preprint, 94 pp. https://doi.org/10.48550/arXiv.2604.10020. 8

  19. [27]

    Emrah, E., Janjigian, C., and Sepp ¨al¨ainen, T.Anomalous Geodesics in the Inhomogeneous Corner Growth Model.Comm. Math. Phys. 406, 12 (2025), Paper No. 316, 65. https://doi.org/10.1007/ s00220-025-05470-8. 7

  20. [28]

    G., and Tanny, S

    Epstein, L. G., and Tanny, S. M.Increasing generalized correlation: A definition and some economic consequences.Can. J. Econ. 13, 1 (1980), 16–34.https://doi.org/10.2307/134617. 6

  21. [29]

    L., and Sepp¨al¨ainen, T.Joint distribution of Busemann functions in the exactly solvable corner growth model.Probab

    F an, W.-T. L., and Sepp¨al¨ainen, T.Joint distribution of Busemann functions in the exactly solvable corner growth model.Probab. Math. Phys. 1, 1 (2020), 55–100. https://doi.org/10.2140/pmp.2020. 1.55. 7, 13, 17, 29, 33, 34

  22. [30]

    A., and Martin, J

    Ferrari, P. A., and Martin, J. B.Multi-class processes, dual points and M/M/1 queues.Markov Process. Related Fields 12, 2 (2006), 175–201. https://math-mprf.org/journal/articles/id1076/. 7

  23. [31]

    A., and Martin, J

    Ferrari, P. A., and Martin, J. B.Stationary distributions of multi-type totally asymmetric exclusion processes.Ann. Probab. 35, 3 (2007), 807–832.https://doi.org/10.1214/009117906000000944. 7

  24. [32]

    A., and Martin, J

    Ferrari, P. A., and Martin, J. B.Multiclass Hammersley-Aldous-Diaconis process and multiclass- customer queues.Ann. Inst. Henri Poincar´ e Probab. Stat. 45, 1 (2009), 250–265. https://doi.org/10. 1214/08-AIHP168. 7

  25. [33]

    A., and Pimentel, L

    Ferrari, P. A., and Pimentel, L. P. R.Competition interfaces and second class particles.Ann. Probab. 33, 4 (2005), 1235–1254.https://doi.org/10.1214/009117905000000080. 4

  26. [34]

    Theory Related Fields 169, 1-2 (2017), 177–222

    Georgiou, N., Rassoul-Agha, F., and Sepp ¨al¨ainen, T.Stationary cocycles and Busemann functions for the corner growth model.Probab. Theory Related Fields 169, 1-2 (2017), 177–222. https://doi.org/10.1007/s00440-016-0729-x. 4 36

  27. [35]

    Groathouse, S., Janjigian, C., and Rassoul-Agha, F.Existence of generalized Busemann functions and Gibbs measures for random walks in random potentials.Trans. Amer. Math. Soc. 378, 12 (2025), 8487–8563.https://doi.org/10.1090/tran/9452. 4

  28. [36]

    Theory Related Fields 192, 1-2 (2025), 303–372

    Groathouse, S., Rassoul-Agha, F., Sepp ¨al¨ainen, T., and Sorensen, E.Jointly invariant measures for the Kardar–Parisi–Zhang equation.Probab. Theory Related Fields 192, 1-2 (2025), 303–372. https://doi.org/10.1007/s00440-025-01362-z. 8

  29. [37]

    Janjigian, C., and Rassoul-Agha, F.Busemann functions and Gibbs measures in directed polymer models onZ 2.Ann. Probab. 48, 2 (2020), 778–816.https://doi.org/10.1214/19-AOP1375. 4

  30. [38]

    Janjigian, C., Rassoul-Agha, F., and Sepp ¨al¨ainen, T.Geometry of geodesics through Busemann measures in directed last-passage percolation.J. Eur. Math. Soc. (JEMS) 25, 7 (2023), 2573–2639. https://doi.org/10.4171/jems/1246. 7

  31. [39]

    Joag-Dev, K., and Proschan, F.Negative association of random variables, with applications.Ann. Statist. 11, 1 (1983), 286–295.https://doi.org/10.1214/aos/1176346079. 6

  32. [40]

    Multivariate Anal

    Joe, H.Multivariate concordance.J. Multivariate Anal. 35, 1 (1990), 12–30. https://doi.org/10. 1016/0047-259X(90)90013-8. 6

  33. [41]

    R., Sepp ¨al¨ainen, T., and V alk´o, B.Fluctuation exponents for directed polymers in the intermediate disorder regime.Electron

    Moreno Flores, G. R., Sepp ¨al¨ainen, T., and V alk´o, B.Fluctuation exponents for directed polymers in the intermediate disorder regime.Electron. J. Probab. 19(2014), no. 89, 28. https: //doi.org/10.1214/EJP.v19-3307. 3

  34. [42]

    Lecture notes, 63 pp

    Sepp¨al¨ainen, T.Variational formulas, Busemann functions, and fluctuation exponents for the corner growth model with exponential weights. Lecture notes, 63 pp. https://doi.org/10.48550/arXiv. 1709.05771. 4, 6, 11

  35. [43]

    Sepp¨al¨ainen, T.Scaling for a one-dimensional directed polymer with boundary conditions.Ann. Probab. 40, 1 (2012), 19–73. Corrected version available at arXiv:0911.2446. https://doi.org/10.1214/10- AOP617. 3

  36. [44]

    Sepp¨al¨ainen, T., and Shen, X.Coalescence estimates for the corner growth model with expo- nential weights.Electron. J. Probab. 25(2020), Paper No. 85, 31. Corrected version available at arXiv:1911.03792.https://doi.org/10.1214/20-ejp489. 7

  37. [45]

    Sepp¨al¨ainen, T., and Sorensen, E.Busemann process and semi-infinite geodesics in Brownian last-passage percolation.Ann. Inst. Henri Poincar´ e Probab. Stat. 59, 1 (2023), 117–165. https: //doi.org/10.1214/22-aihp1245. 8

  38. [46]

    Sepp¨al¨ainen, T., and Sorensen, E.Global structure of semi-infinite geodesics and competition interfaces in Brownian last-passage percolation.Probab. Math. Phys. 4, 3 (2023), 667–760. https: //doi.org/10.2140/pmp.2023.4.667. 8

  39. [47]

    Sepp¨al¨ainen, T., and V alk´o, B.Bounds for scaling exponents for a 1 + 1 dimensional directed polymer in a Brownian environment.ALEA Lat. Am. J. Probab. Math. Stat. 7(2010), 451–476. https://alea.impa.br/articles/v7/07-20.pdf. 3

  40. [48]

    Shen, X.Independence property of the Busemann function in exactly solvable KPZ models.Ann. Appl. Probab. 35, 1 (2025), 458–480.https://doi.org/10.1214/24-aap2118. 6, 12, 17 37

Pith tools

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