Kirillov's conjecture on Hecke-Grothendieck polynomials

This paper utilizes algebraic methods from statistical mechanics to represent Kirillov's multi-parameter class of polynomials—including Schubert and Grothendieck polynomials—as partition functions of solvable lattice models, thereby proving positivity conjectures for Hecke-Grothendieck polynomials while revealing that the broader family can exhibit negative coefficients.

The Gist

Imagine a giant, complex puzzle made of a grid of squares. In the world of mathematics, this is called a lattice model. Usually, these models are used to describe how tiny particles interact in physics, like water molecules freezing into ice. But in this paper, a team of mathematicians uses a similar grid to solve a very different kind of puzzle: understanding complex mathematical formulas called polynomials.

Here is the story of what they did, broken down into simple concepts:

1. The Goal: Taming the "Wild" Polynomials

Mathematicians have known about certain special formulas (polynomials) for a long time. These formulas are like the "DNA" of shapes and symmetries in geometry. A mathematician named Kirillov proposed a massive, flexible family of these formulas that could do everything the older, simpler ones could do, and more. He called them twisted Kirillov polynomials.

However, Kirillov made a big guess (a conjecture): he thought that if you wrote these formulas out, all the numbers (coefficients) inside them would be positive (like 1, 2, 3) and never negative (like -1, -2). He believed this was true for a specific, important sub-group of these formulas called Hecke–Grothendieck polynomials.

2. The Tool: A New Kind of "Traffic Grid"

To prove or disprove Kirillov's guess, the authors built a new kind of mathematical machine: a solvable lattice model.

Think of this model as a traffic grid for tiny cars (which they call "paths" or "colors").

• The Grid: It's a rectangle with rows and columns.
• The Cars: Different colored cars enter from the top and must drive down and to the left, exiting out the left side.
• The Rules (Boltzmann Weights): At every intersection (vertex), there are rules about how cars can pass each other. Some intersections are "free" (cost 0), while others have a "price" (a mathematical value).
• The Magic: The authors designed these rules so that the total "cost" of all possible traffic patterns on the grid exactly matches the complex Kirillov polynomials.

3. The Big Challenge: Proving the Machine Works

For a traffic grid to be useful, it must be "solvable." This doesn't mean the traffic is easy; it means the rules are perfectly balanced. If you swap the order of two intersections, the total cost of the traffic flow shouldn't change. In physics, this is called satisfying the Yang–Baxter equation.

Usually, these grids are built using known "blueprints" from quantum physics (quantum groups). But the authors' grid was weird. It didn't fit any known blueprints. It was like building a car engine that no mechanic had ever seen before.

To prove their engine worked, they had to do a massive amount of checking. They showed that no matter how the cars (colors) arranged themselves, the rules held up. They even wrote a computer program (a SageMath script) to check thousands of tiny scenarios to ensure the math was perfect.

4. The Discovery: The Guess Was Half Right

Once they proved their grid was a valid machine, they used it to check Kirillov's guess about positive numbers.

• The Bad News: They found that Kirillov's guess was false for the general family of polynomials. If you tweak the rules just right, you can get negative numbers (like -5) in the formulas. It's like finding a traffic pattern where the "cost" becomes negative, which is weird but mathematically possible.
• The Good News: They proved that Kirillov was right for the specific sub-family he cared about most: the Hecke–Grothendieck polynomials.

Why?
When they looked at the traffic grid for this specific case, they realized something beautiful: Negative numbers can only appear if two cars try to squeeze onto the same vertical road. But in this specific version of the rules, the grid physically forbids two cars from being on the same vertical road at the same time. Since the "bad" (negative) traffic patterns are impossible, the final result is guaranteed to be made of only positive numbers.

5. The Conclusion

The paper is a success story of using a physical analogy (a traffic grid) to solve an abstract math problem.

1. They built a new, strange traffic grid that perfectly mimics a complex family of polynomials.
2. They proved the grid works by showing its rules are perfectly balanced.
3. They used the grid to show that while some of these polynomials can have negative numbers, the most important ones (Hecke–Grothendieck) are always positive.

In short, they built a new kind of "calculator" made of traffic rules that finally settled a long-standing debate about whether these specific mathematical formulas are always positive.

Technical Summary

Technical Summary: Kirillov's Conjecture on Hecke–Grothendieck Polynomials

Problem Statement
The paper addresses the representation of a multi-parameter class of polynomials in several variables, defined by A. N. Kirillov, using solvable lattice models. These polynomials arise from the largest class of divided difference operators satisfying the braid relations of Cartan type $A$. This family includes, as specializations, Schubert polynomials, Grothendieck polynomials, dual-Grothendieck polynomials, and generalized key polynomials.

A central motivation is the observation that while many special functions in symmetric function theory (such as non-symmetric Hall–Littlewood polynomials and LLT polynomials) have been successfully represented as partition functions of solvable lattice models derived from quantum group modules, it remains an open question whether all functions arising from universal divided difference operators admit such a representation. Specifically, the authors aim to construct a lattice model whose partition function matches Kirillov's "twisted Kirillov polynomials" $K^{(\alpha, \beta, \gamma)}_w(x; \lambda)$, defined via a four-parameter family of operators (reduced to three parameters $\alpha, \beta, \gamma$ in the generic case where a specific parameter $d \neq 0$).

Furthermore, the paper investigates Kirillov's conjectures regarding the non-negativity of the coefficients of these polynomials. While the conjecture holds for many known specializations (e.g., Schubert and $\beta$-Grothendieck polynomials), it was unknown whether it held for the general multi-parameter family or if counter-examples existed.

Methodology
The authors employ algebraic methods from statistical mechanics, specifically the theory of solvable lattice models. The core methodology involves:

1. Lattice Construction: They define a new family of colored rectangular lattice models. Unlike previous models associated with standard quantum group modules (e.g., $U_q(\widehat{\mathfrak{gl}}(n+1))$), the Boltzmann weights for this model are not immediately recognizable as arising from such modules. The model features $n$ rows and $N+1$ columns, with spectral parameters $x_i$ assigned to rows.
2. Boltzmann Weights: The non-zero weights are defined for vertices where horizontal edges carry a single label from $\{+, 1, \dots, n\}$ and vertical edges carry subsets of $\{1, \dots, n\}$. The weights involve the parameters $\alpha, \beta, \gamma$, the spectral parameters $x_i$, and complete homogeneous symmetric functions $h_k(\alpha, \beta)$. A key feature is the operation $x \oplus_{\beta, \gamma} 1 = 1 + (\beta + \gamma)x$, which realizes a multiplicative formal group law.
3. Solvability (Yang–Baxter Equation): To prove the model is solvable, the authors demonstrate that the Boltzmann weights satisfy the Yang–Baxter equation (specifically the RLL relation).
   • They construct an R-matrix (R-vertex) with specific weights (Figure 3.2) that depends on two spectral parameters $x_i, x_j$.
   • They prove the RLL relation by reducing the verification to a finite set of cases based on the relative ordering of horizontal edge labels and their containment in vertical edge labels.
   • Crucially, they show that the proof does not depend on the total number of colors $n$ being bounded, but rather on the combinatorics of the labels. The verification is performed via a computer-assisted symbolic computation (SageMath script provided in Appendix B) that checks the equality of partition functions for all admissible boundary conditions.
4. Boundary Conditions: They define specific boundary conditions determined by a permutation $w \in S_n$ and a partition $\lambda$. These conditions force paths to enter from the top and exit to the left, with the partition function $Z(S)$ summing the weights of all admissible states.

Key Contributions and Results

1. Main Theorem (Theorem 1.2 / 2.3): The authors prove that for any positive integer $n$, permutation $w$, and partition $\lambda$, the twisted Kirillov polynomial $K^{(\alpha, \beta, \gamma)}_w(x; \lambda)$ is exactly equal to the partition function of their constructed solvable lattice model with specific boundary conditions. This establishes a direct link between the universal divided difference operators defined by Kirillov and solvable lattice models.
2. Positivity of Hecke–Grothendieck Polynomials (Theorem 1.4): By specializing the parameter $\gamma = 0$, the authors prove that the resulting "Hecke–Grothendieck polynomials" have non-negative coefficients in $\mathbb{N}[\alpha, \beta][x_1, \dots, x_n]$.
   • Mechanism: They show via Lemma 5.5 that when $\gamma = 0$, no admissible state contains a vertex with a vertical edge labeled by a subset of size greater than 1. Consequently, the complex Boltzmann weights involving negative signs and higher-degree symmetric functions (which appear only when $\gamma \neq 0$) are never realized in the partition function. The remaining weights are manifestly non-negative.
   • Corollaries: This result recovers known positivity for $\beta$-Grothendieck polynomials ($\alpha=\gamma=0$), dual $\beta$-Grothendieck polynomials ($\beta=\gamma=0$), Schubert polynomials ($\alpha=\beta=\gamma=0$), and Di Francesco–Zinn-Justin polynomials.
3. Disproof of General Positivity Conjectures (Propositions 5.2 & 5.3): The paper provides explicit counter-examples showing that Kirillov's conjecture of non-negative coefficients fails for the general family when $\gamma \neq 0$.
   • For $n=4$, specific permutations yield polynomials with negative coefficients (e.g., terms involving $-\beta^3 \gamma$).
   • Similarly, generalized key polynomials $K^{(\alpha, \beta, \gamma)}_\zeta(x)$ can have negative coefficients even when $\zeta \subset \rho$.
4. Additional Positivity Result (Theorem 5.8): Despite the general failure of positivity, the authors identify another specific case where positivity holds: for parameters $(\alpha, \beta, \gamma) = (-\alpha, -\beta, \alpha+\beta)$, the polynomials have non-negative coefficients in $\mathbb{N}[\alpha, \beta]$. This is notable because it occurs in the regime $\gamma \neq 0$, where the lattice states are most complex.
5. Connection to Quantum Groups: The authors investigate the relationship between their R-matrix and known quantum group R-matrices. They show that in the degeneration $\alpha=\gamma=0$, their model relates to the affine quantum superalgebra $U_q(\widehat{\mathfrak{sl}}(1|n))$ via a Drinfeld twist in the limit $q \to 0$. A similar connection exists for $\beta=\gamma=0$ and $U_q(\widehat{\mathfrak{sl}}(n|1))$. However, for the general multi-parameter case, no such quantum group module origin is currently known, suggesting the model lies outside the standard framework of quasi-triangular Hopf algebras.

Significance and Claims
The paper claims to provide the first solvable lattice model representation for the universal family of divided difference operators studied by Kirillov (in the generic $d \neq 0$ case). This demonstrates that lattice models are a source for these universal functions, extending the scope beyond those derived from standard quantum group modules.

The significance of the work lies in:

• Unification: Providing a common framework (the lattice model) that generates Schubert, Grothendieck, and key polynomials as specializations.
• Resolution of Conjectures: Proving the positivity of Hecke–Grothendieck polynomials while simultaneously disproving the broader conjecture for the general family, thereby clarifying the precise conditions under which positivity holds.
• Novelty of Solvability: Demonstrating solvability for a model with "strange" Boltzmann weights that do not factorize in the standard way associated with quantum group fusion, relying instead on a reduction to finite combinatorial cases and computer verification.

The authors note that while their model handles the generic case ($d \neq 0$), the case where $d=0$ (covering key polynomials) requires a different, generalized lattice model, which is the subject of subsequent work by one of the authors. They also acknowledge that the geometric interpretation of the positivity results, particularly for the $\gamma \neq 0$ case, remains an open and interesting problem.