Abstract:We show that under the 1:2:3 scaling, critically probing large space and time, the height function of finite range asymmetric exclusion processes and the KPZ equation converge to the KPZ fixed point, constructed earlier as a limit of the totally asymmetric simple exclusion process through exact formulas.
Consequently, based on recent results of \cite{wu},\cite{DM20}, the KPZ line ensemble converges to the Airy line ensemble.
For the KPZ equation we are able to start from a continuous function plus a finite collection of narrow wedges. For nearest neighbour exclusions, we can take (discretizations) of height functions with $h(x)\le C(1+|x|)$. For non-nearest neighbour exclusions, we are restricted at the present time to a class of (random) initial data, dense in continuous functions in the topology of uniform convergence on compacts.
The method is by comparison of the transition probabilities of finite range exclusion processes and the totally asymmetric simple exclusion processes using energy estimates.
What problem does this paper attempt to address?
### The problems the paper attempts to solve
The paper aims to prove that under the 1:2:3 scale, the height functions of the finite - range asymmetric exclusion process (AEP) and the KPZ equation converge to the KPZ fixed point. Specifically, the main contribution of the paper lies in providing an analytical method. By comparing the transition probabilities of any finite - range asymmetric exclusion process with those of the totally asymmetric simple exclusion process (TASEP), it shows their identical limit behaviors under the 1:2:3 scale. Moreover, this method also allows to prove that the transition probabilities of the KPZ equation under very general initial data converge to the transition probabilities of the KPZ fixed point.
### Main results
1. **Convergence of AEP**:
- Suppose the initial data \( f_{0,\epsilon} \) satisfies \( f_{0,\epsilon} d\nu_\epsilon \to d\rho \) and \( \epsilon^{1/4} \| f_{0,\epsilon} \|_{L^2(d\nu_\epsilon)} \to 0 \), and \( \rho \)-almost surely, any sample path \( h(\cdot) \) is bounded on any compact interval and satisfies \( |h(x)| \leq C(1 + |x|) \) (where \( C > 0 \)), then the AEP height function \( \tilde{h}_\epsilon(t, x) := h_\epsilon(2t, 2x) + \epsilon^{-1} t/2 \) starting from \( f_{0,\epsilon} d\nu_\epsilon \) converges in distribution to the KPZ fixed point starting from \( \rho \) as \( \epsilon \to 0 \), under the uniform topology on compact sets.
2. **Convergence of ASEP**:
- Suppose the initial data \( h_{0,\epsilon} \to h \) in UC, then under the same scale, the ASEP height function \( \tilde{h}_\epsilon \) converges in distribution to the KPZ fixed point starting from \( h \) as \( \epsilon \to 0 \), under the uniform topology on compact sets.
3. **Convergence of the KPZ equation**:
- Starting from the initial data of a continuous function plus a finite number of narrow wedges, the KPZ equation (2.10), as \( \delta \to 0 \), the solution plus the Ito factor \( t/12 \) converges in distribution to the KPZ fixed point, under the uniform topology on compact subsets. The KPZ line ensemble converges to the Airy line ensemble in the same sense (see Section 9).
### Key steps
- **Estimation of the difference in transition probabilities**: By comparing the transition probabilities of AEP and TASEP, use the Dirichlet form to control the difference.
- **Control of the symmetric part**: Utilize the properties of the symmetric part, especially the diffusivity of the difference between the generator of the symmetric part and the asymmetric part.
- **Strong sector condition**: Control the difference in transition probabilities through the strong sector condition.
### Method overview
The key of the paper lies in using the classical semigroup theory to control the difference in transition probabilities through generators and Dirichlet forms. The specific steps include:
1. **Definition of generators**: Define the generators of AEP and TASEP and calculate their Dirichlet forms.
2. **Difference in transition probabilities**: Estimate the difference in the transition probabilities of the two Markov processes through the Dirichlet form.
3. **Strong sector condition**: Use the strong sector condition to further control the difference in transition probabilities.
4. **Extension of initial data**: Extend from specific initial data to more general data to prove convergence.
Through these steps, the paper successfully proves the convergence of AEP and the KPZ equation under the 1:2:3 scale, providing an important theoretical basis for understanding the universality of KPZ - type models.