Abstract:We show that the directed landscape is the unique coupling of the KPZ fixed point from all initial conditions at all times satisfying three natural properties: independent increments, monotonicity, and shift commutativity. Equivalently, we show that the directed landscape is the unique directed metric on $\mathbb R^2$ with independent increments and KPZ fixed point marginals. This gives a framework for proving convergence to the directed landscape given convergence to the KPZ fixed point. We apply this framework to prove convergence to the directed landscape for a range of models, some without exact solvability: asymmetric exclusion processes with potentially non-nearest neighbour interactions, exotic couplings of ASEP, the random walk and Brownian web distance, and directed polymer models. All of our convergence theorems are new except for colored ASEP and the KPZ equation, where we provide alternative proofs.
What problem does this paper attempt to address?
### The problems the paper attempts to solve
The paper aims to solve the problem of how to construct and characterize the directed landscape from the KPZ fixed point. Specifically, the author hopes to prove that the directed landscape is the only coupling that satisfies certain natural properties, which include independent increments, monotonicity, and shift commutativity. In addition, the paper also provides a framework for proving that a series of models converge to the directed landscape after appropriate rescaling, and some of these models do not have exact solvability.
### Background and problem description
#### KPZ fixed point and directed landscape
The KPZ (Kardar - Parisi - Zhang) universality class refers to a class of one - dimensional random interface growth models and two - dimensional random metric or polymer models that exhibit the same behavior under rescaling. In the past 25 years, significant progress has been made in the research of the KPZ universality class, especially through the discovery of some exactly solvable models, such as the exclusion process, the directed polymer, the last - passage percolation model, and the KPZ equation itself.
The KPZ fixed point is the universal limit scale of the KPZ - class growth models, and the directed landscape is the universal limit scale of the metric and polymer models. The KPZ fixed point can be constructed as a marginal distribution of the directed landscape, and vice versa: the author hopes to construct the directed landscape from the KPZ fixed point. This is the main objective of this paper.
#### Exclusion process and KPZ fixed point
To describe the KPZ fixed point, consider a specific family of growth models: the exclusion process on the integer lattice. Let \(p\) be a measure on the integer lattice with finite support. The exclusion process allows each particle to attempt to jump to another position, and all particle jumps are independent. The height function \(h_t(x)\) describes the state change of the exclusion process.
For the asymmetric exclusion process (AEP), the height function decreases at a linear rate, so the process \((h_t)_{t\geq0}\) generates a growth interface constrained by the local structure. Matetski, Quastel and Remenik [MQR21] showed that the height function of TASEP converges to a limit Markov process, namely the KPZ fixed point, under rescaling.
#### Random metric and coupled random growth
The KPZ universality class also includes two - dimensional random metric models, which can be naturally constructed by a set of independent random variables or noise on the plane. For example, consider the distance process from a point to a set of time - indexed lines, which will generate couplings of many different random growth processes. If a particularly natural coupling is given, this coupling can be realized through the random metric.
### Main results
The main results of the paper show that the directed landscape is the unique random function on the space of bounded upper - semicontinuous functions that satisfies the following conditions:
1. **Triangle inequality**: For any \(o = (x, s), p=(y, r), q=(z, t)\in Q^2\) and \(s < r < t\), we have \(M(o; p)+M(p; q)\leq M(o; q)\).
2. **Independent increments**: For any \(s_1 < s_2 <\cdots< s_k\in Q\), the increments \(M(\cdot, s_i; \cdot, s_{i + 1})\colon Q^2\to R\) are independent.
3. **KPZ fixed point marginal distribution**: For any \(p=(x, t)\in Q^2\), and \(y_1,\ldots, y_k, s_1,\ldots, s_k\in Q\) and each \(s_i > 0\), we have
\[
\{M(p; p+(y_i, s_i))\}_{i = 1}^k\stackrel{d}{=}\{h^{s_i}_{\delta_0}(y_i)\}_{i = 1}^k.
\]
Here \(\delta_0\) is the narrow - wedge initial condition, defined as \(\delta_0(0) = 0\) and \(\delta_0(x)=-\infty\) for \(x\neq0\).