Periodic KPZ fixed point with general initial conditions

This paper establishes the periodic KPZ fixed point with general upper-semicontinuous initial conditions by deriving the large-time limit of the rescaled space-time multipoint distribution for the periodic totally asymmetric simple exclusion process, utilizing a novel hitting expectation representation of the energy and characteristic functions.

The Gist

Imagine you are watching a massive, endless line of people trying to get through a narrow hallway. This is a bit like the Totally Asymmetric Simple Exclusion Process (TASEP). In this world, people (particles) can only move forward, never backward. They can't jump over each other, and they can't occupy the same spot. If the person in front of you is standing still, you have to wait.

Now, imagine this hallway isn't just a straight line; it's a giant loop (like a racetrack). If you run far enough to the right, you end up back where you started. This is the Periodic version of the problem.

The Big Question: How does the crowd move over time?

Scientists have been studying how "height" (the number of people who have passed a certain point) fluctuates in these systems. They found that if you wait a long time, the crowd's movement follows a universal pattern called the KPZ Fixed Point. Think of this as the "DNA" of growth: whether it's bacteria growing on a petri dish, a fire spreading, or people in a hallway, they all eventually move in the same statistical way.

But there's a catch. The "DNA" (the KPZ Fixed Point) was only fully understood for infinite, straight lines. What happens when the system is on a loop?

The "Relaxation" Moment

The authors of this paper asked: What happens when the loop is huge, and we wait a very specific amount of time?

• Too short: The loop is so big it feels like an infinite line. The crowd behaves normally.
• Too long: The crowd has circled the track so many times that the starting point doesn't matter anymore. Everyone is just jiggling randomly like a Brownian motion (like dust motes dancing in a sunbeam).
• Just Right (The Sweet Spot): There is a specific "Goldilocks" time scale (called the relaxation time) where the system is transitioning. It's no longer acting like an infinite line, but it hasn't forgotten its starting shape yet. This is where the magic happens.

The Main Discovery: A New "Universal Map"

The authors successfully calculated exactly what the crowd looks like at this "Just Right" moment, no matter how the crowd started.

Previously, they could only solve this for very simple starting lines (like everyone starting in a perfect row, or everyone starting in a tight wedge). This paper solves it for any starting shape.

The Analogy:
Imagine you are a weather forecaster.

• Old papers: Could only predict the weather if the day started with a perfectly clear sky or a perfectly stormy sky.
• This paper: Can predict the weather for any starting condition—whether it's a mix of sun and clouds, a foggy morning, or a chaotic storm. They found a universal formula that works for every possible "initial mood" of the system.

How Did They Do It? (The "Guess-and-Check" Magic)

The math behind this is incredibly complex, involving things called "Bethe roots" (which are like hidden keys to the system's energy) and "Fredholm determinants" (a fancy way of calculating probabilities).

The authors' biggest breakthrough was finding a new way to describe these complex formulas using Random Walks.

• The Old Way: Trying to solve a giant, tangled knot of algebra.
• The New Way: They realized these knots could be untangled by imagining a random walker (a drunk person stumbling around) trying to hit a specific target.
  • They found that the "energy" of the system could be described by the expected time it takes for this random walker to hit a wall.
  • They found that the "characteristic function" (another key piece of the puzzle) could be described by the probability of the walker hitting a wall before a certain time, but after another time.

This is like realizing that instead of calculating the exact path of every single person in the crowd, you can just ask: "If I drop a single person in the crowd and let them wander randomly, how likely are they to bump into the front of the line?"

Why Does This Matter?

1. Universality: It proves that this "Periodic KPZ Fixed Point" is a real, universal object. It exists independently of the specific details of the system, just like the standard KPZ fixed point.
2. The Bridge: It connects two different worlds: the world of "infinite lines" (where things are chaotic and growing) and the world of "loops" (where things eventually settle down). This new "Fixed Point" is the bridge between them.
3. Future Tools: By translating complex algebra into "hitting probabilities" (random walks), they gave future mathematicians a much easier toolkit to study these systems. It's like giving them a map instead of a maze.

In a Nutshell

The authors took a complex, looping traffic jam problem, waited for the perfect moment in time, and figured out exactly how the traffic flows for any starting arrangement. They did this by inventing a clever new way to look at the problem: instead of tracking the traffic, they tracked a single, wandering ghost to see where it would bump into the crowd. This revealed a new, universal law of nature for how things grow and move in loops.

Technical Summary

1. Problem Statement and Context

The KPZ Universality Class:
The Kardar-Parisi-Zhang (KPZ) universality class encompasses a wide range of stochastic growth models (e.g., TASEP, random polymers, stochastic PDEs). A central conjecture is that under the KPZ scaling ($t^{1/3}$ height fluctuations, $t^{2/3}$ spatial correlations), these models converge to a universal limiting object called the KPZ fixed point.

The Periodic Setting:
While the KPZ fixed point on the infinite line was constructed recently (MQR21), the behavior of these systems on a periodic domain (a ring of size $L$) remains a critical area of study.

• Regimes:
  • $t \ll L^{3/2}$: Finite-size effects are negligible; the system behaves like the infinite-space KPZ fixed point.
  • $t \gg L^{3/2}$: Spatial correlations become trivial; the system behaves like Brownian motion.
  • Relaxation Time Scale ($t = O(L^{3/2})$): This is the critical scaling window where a non-trivial limiting field, the Periodic KPZ Fixed Point, is expected to emerge, interpolating between the KPZ fixed point and Brownian motion.

The Gap:
Previous rigorous results (specifically [BL19, BL21]) established the limiting distributions for the Periodic Totally Asymmetric Simple Exclusion Process (PTASEP) only for special initial conditions (e.g., periodic narrow wedge, flat, or half-flat). The behavior for general upper-semicontinuous initial conditions remained an open problem.

Objective:
This paper aims to derive the large-time limit of the rescaled space-time multipoint distribution of the height function for PTASEP with arbitrary initial conditions in the relaxation time scale, thereby rigorously defining the Periodic KPZ Fixed Point for general initial data.

---

2. Methodology

The authors employ a combination of integrable probability, asymptotic analysis, and probabilistic representations.

A. Starting Point: Algebraic Formulas

The authors begin with the exact multipoint distribution formula for PTASEP with general initial conditions derived in [BL21]. In this formula, the initial condition is encoded via two spectral functions:

1. The Energy Function ($E_Y$): Related to the Bethe roots of the system.
2. The Characteristic Function ($pch_Y$): A modified periodic characteristic function.

B. Key Technical Innovation: Probabilistic Representations

The core novelty of the paper is the derivation of hitting expectation representations for the Energy and Characteristic functions. This transforms algebraic objects into probabilistic ones suitable for asymptotic analysis.

• Energy Function: The authors establish a Fredholm determinant representation for $E_Y$ involving a kernel defined by the hitting time of a geometric random walk to the strict epigraph of the initial condition profile. This was achieved through an extensive "guess-and-check" exploration, as no prior Fredholm-type representation existed for this symmetric function of Bethe roots.
• Characteristic Function: They derive a probabilistic representation for $pch_Y$ using random walk hitting expectations. This is a periodic analogue of recent results for the infinite domain [LL25, MQR21], involving a difference of two hitting times ($\tau$ and $\tau^*$).

C. Asymptotic Analysis (Steepest Descent)

With the probabilistic representations in hand, the authors perform a rigorous asymptotic analysis as $L \to \infty$ (with $t \sim L^{3/2}$):

1. Scaling: They rescale the Bethe roots and parameters to the KPZ scaling window.
2. Kernel Convergence: They prove that the discrete kernels in the Fredholm determinants (involving geometric random walks) converge to continuous kernels involving Brownian motion and its hitting times.
3. Trace Norm Convergence: They establish uniform bounds on the trace norms of the kernels to ensure the convergence of the Fredholm determinants.
4. Series Expansion: They analyze the series expansion of the distribution function, proving pointwise convergence and uniform bounds to apply the Dominated Convergence Theorem.

---

3. Key Contributions and Results

A. Definition of the Periodic KPZ Fixed Point

The paper defines the limiting multipoint distribution function $F_h^{(p)}$ for a general initial condition $h \in UC_p$ (the space of $p$-periodic upper-semicontinuous functions).

• Formula: The limit is expressed as a contour integral involving two main components:
  1. $C_h(z)$: A term involving a Fredholm determinant of an operator $K_{en}^h$ (related to Brownian hitting times) and exponential factors involving polylogarithms.
  2. $D_h(z)$: A Fredholm determinant (or series expansion) involving a "periodic characteristic function" $X_h$, which is also defined via Brownian hitting expectations.
• Consistency: The authors prove that these finite-dimensional distributions form a consistent family (Kolmogorov consistency), thereby defining a unique random field, the Periodic KPZ Fixed Point $H_{KPZ}^{(p)}(\alpha, \tau; h)$.

B. Main Theorem (Theorem 1.2)

For any general initial condition $h \in UC_p$, the rescaled height function of the PTASEP converges in distribution to the Periodic KPZ Fixed Point:
$$\lim_{L \to \infty} P\left( \bigcap_{i=1}^m \{ \text{Rescaled } H(x_i, t_i) \leq \beta_i \} \right) = F_h^{(p)}(\beta_1, \dots, \beta_m; (\alpha_1, \tau_1), \dots, (\alpha_m, \tau_m))$$
where the scaling is $t \sim L^{3/2}$ and $x \sim L$.

C. Scaling Invariance and Limits

The paper establishes the scaling properties of the fixed point:

• Scaling Invariance: The fixed point satisfies a specific scaling relation involving the period $p$, linking the general case to the normalized period $p=1$.
• Infinite Period Limit ($p \to \infty$): The authors conjecture (and prove for specific cases) that as the period goes to infinity, the periodic fixed point converges to the standard KPZ fixed point on the infinite line.
• Small Period Limit ($p \to 0$): They conjecture that as the period vanishes, the system converges to a standard Brownian motion, reflecting the loss of spatial information.

D. Connection to Directed Landscape

The paper discusses the open problem of constructing a Periodic Directed Landscape $L_p$, which would provide a variational formula for the fixed point analogous to the infinite case: $H_{KPZ}^{(p)} \stackrel{d}{=} \max_{\alpha'} \{ h(\alpha') + L_p(\alpha', 0; \alpha, \tau) \}$.

---

4. Significance

1. Completeness of the KPZ Theory: This work completes the rigorous construction of the KPZ fixed point for periodic systems, extending it from special initial conditions to the full space of upper-semicontinuous functions. This parallels the recent completion of the infinite-space theory.
2. Methodological Breakthrough: The derivation of the hitting expectation representation for the energy function is a significant technical achievement. It bridges the gap between algebraic Bethe ansatz formulas and probabilistic path integrals, offering a new toolkit for analyzing integrable systems with general initial data.
3. Universality: By proving that the limit depends only on the initial condition through a specific functional form (involving Brownian hitting times), the paper reinforces the universality of the KPZ class in periodic domains.
4. Foundation for Future Work: The established formulas and the conjectured connection to the Periodic Directed Landscape provide a rigorous foundation for future studies on the sample path properties (e.g., Hölder continuity) and the geometric structure of the periodic KPZ fixed point.

In summary, this paper resolves a major open problem in the KPZ universality class by rigorously defining the periodic KPZ fixed point for general initial conditions, utilizing novel probabilistic representations of spectral functions to bridge discrete integrable models and continuous stochastic fields.