The five vertex model on a rectangle

arXiv:2606.22004 · math-ph, math.CO, math.MP, math.PR · Submitted 2026-06-20 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "The five vertex model on a rectangle".

Mira: Here's the gist of the paper. The paper derives an explicit determinantal expression for the partition function of an inhomogeneous asymmetric five vertex model on a rectangular geometry with general boundary conditions at the bottom and top Model Definition The model considers an M × N grid where configurations consist of k non-intersecting lattice paths flowing along up-right directions from the bottom to the top The partition function is defined as Gx/y (UV) = X C Y M i=1 Y N j=1 w C i,j (uj/vi), (4)> Determinantal Representation Theorem 2.1 states that the partition function is expressible as a determinant: Gx/y(UV) = det(Qi,j (xi, yj)) 1≤i,j≤k Orthogonality Relations The paper establishes an orthogonality relation between Fx and Fe using a complex contour integral Theorem 3.5 (Orthogonality) states that for sequences x, x' of positions with length k, we have 1/k!

Kai: I C dz1 2πi · · · I C dzk 2πi ρ(Z) ∆(Z) 2 Fx(UZ)Fe x'(UZ) = C −1 k δx,x' Homogeneous Limit Results In the homogeneous limit ui, vi → 1, α → 1, the partition function simplifies to a determinant involving a change of variables Explicit Evaluation The explicit evaluation of the diagonal entries Bi,i when xi ≥ yi is shown to be exactly the sum of up-right paths from yi to xi with lattice height M, weighted r per corner The proof shows that the determinant Bi,j in (57) is equal to δx,x' Connection to Vertex Models The five-vertex model is closely connected to the stochastic six-vertex model underlying the asymmetric exclusion process (ASEP) via transformations that preserve the Yang–Baxter equation The L-matrices in (98) are identified with those used in

MS13: The resulting L and R matrices coincide with those used in the present text, namely (1) and (12), respectively, and satisfy the same Yang–Baxter equation (13)> Appendix Details The appendix details how the five-vertex model arises from the q = 0 limit of the stochastic six-vertex model The L and R matrices in (91) are represented graphically as βq(1−v/u) 1−qv/u v/u 1−v/u β(1−qv/u) 1 α α (1−q) √v/u 1−qv/u (1-q) √v/u The resulting L and R matrices coincide with those used in the present text, namely (1) and (12), respectively, and satisfy the same Yang–Baxter equation (13)> References The paper cites numerous foundational works including

Bax85: R.

Mira: J.

Kai: Baxter.

Title and authors: Kai: So we're looking at this paper today about "The five vertex model on a rectangle". It deals with how you can find a determinantal expression for an inhomogeneous version of this five vertex model when you have rectangular geometry and general boundary conditions at the bottom and top.

Mira: That sounds like a lot of technical machinery, Kai. What does that mean for us as we look at what they actually did? It suggests that instead of just using standard non-intersecting lattice path approaches, which they say aren't immediate, they found a determinantal form.

Lev: A determinantal form is usually great for computation because it means you can calculate the partition function by finding the determinant of some matrix. But for this to work out cleanly, we need specific conditions on the model setup.

Kai: Exactly. The paper sets up this framework using a grid of M times N vertices and spectral parameters U and V, which are just lists of numbers that define the weights at each point in that grid <ref:2606.22004#pg3>.

Mira: It’s interesting because they show how this setup relates to things like the partition function interpretation for inhomogeneous generalisations of Grothendieck polynomials, which is a deep connection in combinatorics.

Lev: And they establish an orthogonality relation between these related objects, which is usually a big deal for understanding the structure of these systems.

Kai: So, essentially, this paper moves us away from just looking at the path counting and towards using linear algebra tools to describe the system's partition function <ref:2606.22004#pg2>. That sets up what they can do next when we talk about specific results.

The paper's summary: Mira: What the paper really boils down to, Kai, is that they prove this determinantal expression for the partition function of an inhomogeneous five vertex model on a rectangular geometry with arbitrary boundary conditions at both ends <ref:2606.22004#pg2>.

Kai: So, they're taking that general setup, the M times N grid and those spectral parameters U and V, and showing that the partition function G x over y is equal to a determinant of some matrix B with entries B i j depending on x sub i and y sub j <ref:2606.22004#pg1>.

Lev: That's the main payoff, I guess. If you have a determinant, you can use established linear algebra techniques to study things like its eigenvalues or its asymptotic behavior. But we need to make sure those matrix entries are well-defined for any boundary condition x over y <ref:2606.22004#pg3>.

Mira: They then go on to show that these matrix entries have a very specific combinatorial meaning, which is really neat because it connects the physics back to counting lattice paths.

Kai: They evaluate the diagonal entries B i i when x sub i is greater than or equal to y sub i and show that this entry is exactly the sum of all up-right paths from y sub i to x sub i with a lattice height M, weighted by r for each corner <ref:2606.22004#pg1>.

Lev: That's a concrete combinatorial interpretation. It means that the diagonal entries aren't just some abstract numbers; they count specific paths on the grid defined by the geometry and the boundary conditions.

Mira: And they show that this determinant is actually equal to delta x, x prime when we compare two different sequences of positions <ref:2606.22004#pg1>. That means if you change your path configuration slightly, it changes the determinant in a predictable way.

The paper's improvements: Kai: Now for the part where they suggest improvements, or at least what they found that is useful for other work. They discuss how this structure allows them to interpret the partition function of inhomogeneous generalisations of Grothendieck polynomials <ref:2606.22004#pg2>.

Mira: That connection is powerful because it links this specific lattice model to a much broader mathematical framework, which is what makes the result more than just a niche calculation for one type of path counting.

Lev: If the AI can compute these Grothendieck polynomial interpretations and derive their determinant expressions, it opens up new avenues for studying things like scattering amplitudes or other correlation functions in quantum field theory.

Kai: And they also establish an orthogonality relation between these Grothendieck polynomials <ref:2606.22004#pg2>. That tells us that the basis of these objects is well-behaved, which is a strong structural statement for any system we are modeling.

Mira: This orthogonality relation, combined with the Cauchy identity they find for the transfer matrix, gives us tools to analyze how these systems behave under different conditions.

Lev: From an engineering standpoint, if we can use this to derive a Cauchy identity for the transfer matrix, it means we have a way to control and simplify those massive matrices that you'd run on real hardware.

Conclusion: Kai: So, wrapping up the paper "The five vertex model on a rectangle", the main thing is that they found an explicit determinantal expression for inhomogeneous asymmetric five vertex models on rectangular geometries with arbitrary boundary conditions at the bottom and top <ref:2606.22004#pg1>.

Mira: They showed this leads to an interpretation of inhomogeneous generalisations of Grothendieck polynomials and provides an orthogonality relation between them, which is a significant structural result in combinatorics <ref:2606.22004#pg2>.

Lev: And they connected the model to the five vertex model arising from the q equals zero limit of certain stochastic six vertex models, which links it to some of the most studied integrable systems Baxeighty-five.

Kai: It’s a solid piece of work. It takes a complex physical model and translates it into a structured linear algebra problem that we can attack using established methods.

Mira: Exactly. The connection to Grothendieck polynomials is what makes this result really interesting for theoretical physics, because those polynomials are central to many areas of algebraic geometry and representation theory <ref:2606.22004#pg2>.

Lev: For our work on error correction, the implication is that if we can map these path counting problems onto determinantal forms, it might give us better ways to analyze the complexity of those error-correction spaces.

JAN DE GIER, RICHARD KENYON, MICHAEL WHEELER, YUYANG ZHOU

University of Melbourne · Yale University

math-ph, math.CO, math.MP, math.PR

Submitted: 2026-06-20

Updated: 2026-11-02

Comments: 30 pages, final accepted version

Journal ref: Ann. Inst. Henri Poincare D (2026)

DOI: 10.4171/aihpd/240

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 67/100

The gist: The paper derives an explicit determinantal expression for the partition function of an inhomogeneous asymmetric five vertex model on a rectangular geometry with general boundary conditions at the…

Key concepts

Partition Function (Gx/y(U|V))
This is a mathematical quantity that summarizes all possible configurations of non-intersecting lattice paths on an M x N grid. It is calculated by summing contributions from all valid path arrangements, which the paper shows can be expressed as a determinant.
Determinantal Representation (Theorem 2.1)
The partition function is shown to be exactly equal to the determinant of a specific matrix, denoted as det(Qi,j). This means complex counting problems related to paths can be solved by calculating the determinant of this matrix.
Orthogonality Relation (Theorem 3.5)
This theorem establishes a mathematical relationship between different path configurations (Fx and Fe) using a contour integral. It quantifies how uncorrelated these configurations are, providing a way to test if two path arrangements are independent or related.

Terminology

Summary

The paper derives an explicit determinantal expression for the partition function of an inhomogeneous asymmetric five vertex model on a rectangular geometry with general boundary conditions at the bottom and top

Model Definition

The model considers an M × N grid where configurations consist of k non-intersecting lattice paths flowing along up-right directions from the bottom to the top The partition function is defined as Gx/y (UV) = X C Y M i=1 Y N j=1 w C i,j (uj/vi), (4)>

Determinantal Representation

Theorem 2.1 states that the partition function is expressible as a determinant: Gx/y(UV) = det(Qi,j (xi, yj)) 1≤i,j≤k

Orthogonality Relations

The paper establishes an orthogonality relation between Fx and Fe using a complex contour integral Theorem 3.5 (Orthogonality) states that for sequences x, x' of positions with length k, we have 1/k! I C dz1 2πi · · · I C dzk 2πi ρ(Z) ∆(Z) 2 Fx(UZ)Fe x'(UZ) = C −1 k δx,x'

Homogeneous Limit Results

In the homogeneous limit ui, vi → 1, α → 1, the partition function simplifies to a determinant involving a change of variables

Explicit Evaluation

The explicit evaluation of the diagonal entries Bi,i when xi ≥ yi is shown to be exactly the sum of up-right paths from yi to xi with lattice height M, weighted r per corner The proof shows that the determinant Bi,j in (57) is equal to δx,x'

Connection to Vertex Models

The five-vertex model is closely connected to the stochastic six-vertex model underlying the asymmetric exclusion process (ASEP) via transformations that preserve the Yang–Baxter equation The L-matrices in (98) are identified with those used in [MS13] The resulting L and R matrices coincide with those used in the present text, namely (1) and (12), respectively, and satisfy the same Yang–Baxter equation (13)>

Appendix Details

The appendix details how the five-vertex model arises from the q = 0 limit of the stochastic six-vertex model The L and R matrices in (91) are represented graphically as βq(1−v/u) 1−qv/u v/u 1−v/u β(1−qv/u) 1 α α (1−q) √v/u 1−qv/u (1-q) √v/u The resulting L and R matrices coincide with those used in the present text, namely (1) and (12), respectively, and satisfy the same Yang–Baxter equation (13)>

References

The paper cites numerous foundational works including [Bax85] R. J. Baxter. Exactly solved models in statistical mechanics, [GS92] L. p. 2

Where the paper says it

The remainder of this paper is devoted to the proof of Theorem 2. 1 using the methods of integrable vertex models.

-H. Gwa and H. Spohn. Six-vertex model, roughened surfaces, and an asymmetric spin hamiltonian Phys. p. 0
Where the paper says it

The asymmetric five vertex model (r ̸= 1) has received recent attention in combinatorics due to its connection to Grothendieck polynomials, Demazure atoms and crystal theory [MS13, MS14, WZJ19, Iwa20, Mot20, BBBG21, BS22, GZJ23]. It can be realised as the q = 0 limit of the (stochastic) six vertex model, and therefore the totally asymmetric exclusion process (TASEP), see e.

Rev. Lett., 68:725–728, Feb 1992, and [BP17] A. Borodin and L. Petrov. Integrable probability: stochastic vertex models and symmetric functions In Stochastic Processes and Random Matrices: Lecture Notes of the Les Houches Summer School: Volume 104, July 2015, Oxford University Press, 08 2017. The paper also references [KM59] S. p. 0
Where the paper says it

In mathematics this case is known as the monotone non-intersecting lattice path (MNLP) model, or equivalently, the honeycomb dimer model. The partition function in this case may be computed as a determinant using the Karlin-MacGregor-Lindström-GesselViennot method [KM59, Lin73, GV89] and the Kasteleyn method [Kas67].

Karlin and J. p. 0
Where the paper says it

In mathematics this case is known as the monotone non-intersecting lattice path (MNLP) model, or equivalently, the honeycomb dimer model. The partition function in this case may be computed as a determinant using the Karlin-MacGregor-Lindström-GesselViennot method [KM59, Lin73, GV89] and the Kasteleyn method [Kas67].

McGregor. Coincidence probabilities Pacific Journal of Mathematics, 9(4):1141 – 1164, 1959. p. 0
Where the paper says it

The intersecting path vertex has weight 0. The symmetric five vertex model, when r = 1, has received much attention in both mathematics and physics literature.

The bibliography includes a wide range of literature on integrable systems and combinatorics. p. 0
Where the paper says it

) The symmetric model is used to analyse the asymptotics of the Plancherel measure and limit shapes and enumeration of boxed plane partitions [VK77, CMLP98, Bog05, Pro16], and the correspondence of the inhomogeneous r = 1 five vertex model with Schur polynomials is the basis for the Schur process [OR03] and the combinatorics of Littlewood-Richardson coefficients [ZJ09]. In physics the symmetric model has been used for modeling of modified potassium dihydrogen phosphate (KDP) [Wu68], the roughening transition [BH82], crystal growth and evaporation in two dimensions [Gar90, GLT90, HWKK96], and interacting domain wall systems [NK94].

--- Page 1 ---

The five-vertex model is a model of non-intersecting lattice paths on the directed square lattice <ref:2606.220041,It arises as a special case of the solvable six vertex model [Lie67, SYY67, Bax85] with local vertex configurations listed in Figure 1> The symmetric five vertex model, when r = 1, has received much attention in both mathematics and physics literature <ref:2606. p. 0

Where the paper says it

Introduction The five-vertex model is a model of non-intersecting lattice paths on the directed square lattice. It arises as a special case of the solvable six vertex model [Lie67, SYY67, Bax85] with local vertex configurations listed in Figure 1.

p. 0
Where the paper says it

Introduction The five-vertex model is a model of non-intersecting lattice paths on the directed square lattice. It arises as a special case of the solvable six vertex model [Lie67, SYY67, Bax85] with local vertex configurations listed in Figure 1.

22004#pg4> The asymmetric five vertex model (r ≠ 1) has received recent attention in combinatorics due to its connection to Grothendieck polynomials, Demazure atoms and crystal theory [MS13, MS14, WZJ19, Iwa20, Mot20, BBBG21, BS22, GZJ23> It can be realised as the q = 0 limit of the (stochastic) six vertex model

--- Page 1 ---

The five-vertex model is a model of non-intersecting lattice paths on the directed square lattice <ref:2606.220041,It arises as a special case of the solvable six vertex model [Lie67, SYY67, Bax85] with local vertex configurations listed in Figure 1> The symmetric five vertex model, when r = 1, has received much attention in both mathematics and physics literature <ref:2606.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Determinantal Expression for Inhomogeneous Models: The AI can compute partition functions for inhomogeneous five vertex models on rectangular geometries with arbitrary boundary conditions by deriving and evaluating a determinantal expression, as stated in Theorem 2.1.

  2. Grothendieck Polynomial Interpretation: The system can interpret the partition function of inhomogeneous generalisations of Grothendieck polynomials, derive their determinant expressions, establish an orthogonality relation between them, and calculate a Cauchy identity for the transfer matrix using the derived formulas (6).

  3. Path Enumeration and Combinatorial Analysis: The AI can determine the number of non-intersecting lattice paths with specific boundary conditions by evaluating the expression for diagonal entries Bi,j in Theorem 4.1, which is exactly the sum of up-right paths from yi to xi with lattice height M, weighted r per corner (80).

  4. Homogeneous Limit Calculation: The system can calculate the partition function in the homogeneous limit by using the derived determinantal formula G hom x/y = det(Bi,j) (75), which corresponds to up-right paths weighted r per corner when xi >= yi for all i.

  5. Orthogonality Testing: The AI can test if two different path configurations have non-zero correlation by evaluating the orthogonality relation in Theorem 3.5 using the derived inner product formula (53) to check if it equals one, which is guaranteed by the block triangular structure of B leading to B(m) = delta x(m), x'(m) (60).

Abstract

We derive a determinantal expression for the inhomogeneous asymmetric five vertex model in a rectangular geometry with arbitrary boundary conditions at the bottom and top. Standard non-intersecting lattice path, or free fermion, approaches are not applicable and the determinantal form thus is not immediate.

Sources

Related papers