Correlation Number for Potentials with Entropy Gaps and Cusped Hitchin Representations

This paper introduces a correlation number for pairs of strictly positive, locally Hölder continuous potentials with strong entropy gaps on countable state Markov shifts, applying it to cusped Hitchin representations to explore its relationship with the Manhattan curve and establish associated rigidity properties.

The Gist

Imagine you are standing in a vast, infinite library. This isn't a normal library; it's a Markov Shift. Think of it as a labyrinth of corridors where every room (a "state") has doors leading to other rooms. You can walk through this library forever, but you can only move from one room to another if there's a door connecting them.

Now, imagine two different "scents" or "atmospheres" drifting through this library. Let's call them Potentials.

• Potential A might be the smell of old books.
• Potential B might be the sound of rain against the windows.

As you walk through the library, you collect these scents and sounds. If you take a specific path that loops back to where you started (a "periodic orbit"), you can measure the total "amount" of smell and sound you encountered on that loop.

The Big Question: How Do These Scents Correlate?

The authors of this paper, Kao and Martone, are asking a very specific question: If I look at all the possible loops in this library, how do the total amounts of "Smell A" and "Sound B" relate to each other?

Do they grow together? If a loop has a lot of smell, does it also have a lot of sound? Or are they completely unrelated?

To answer this, they invented a new tool called the Correlation Number.

The "Manhattan Curve": A Map of Possibilities

Imagine you plot every possible loop on a graph.

• The X-axis is the total "Smell" (Potential A).
• The Y-axis is the total "Sound" (Potential B).

If you plot thousands of loops, they don't scatter randomly. They form a beautiful, curved boundary line. The authors call this the Manhattan Curve. It looks a bit like the skyline of a city seen from a distance (hence the name).

• The Curve's Shape: This curve tells you the "limit" of what is possible. You can't have a loop with 100 units of smell and 100 units of sound if the curve says the maximum combination is only 150.
• The Slope: At any point on this curve, there is a specific "slope" or angle. This slope represents a specific balance between the two scents.

The "Correlation Number": The Speed Limit of Growth

Here is the magic trick the authors discovered.

They asked: "If I only look at loops where the smell and sound are in a specific ratio (defined by the slope of the curve), how fast does the number of such loops grow as the loops get longer?"

They found that the number of these loops grows exponentially (like compound interest). The Correlation Number is simply the rate at which this growth happens.

• Analogy: Imagine you are counting how many people in a city have a specific height and weight combination. As you look at taller and taller people, the number of people fitting a specific "height-to-weight" ratio might explode. The Correlation Number is the "explosion speed" for that specific ratio.

The paper proves that this "explosion speed" is exactly equal to a geometric property of the Manhattan Curve (specifically, a calculation involving the curve's height and slope). This is a huge deal because it connects a counting problem (how many loops?) with a geometric problem (what does the curve look like?).

The "Rigidity" Surprise

One of the most exciting findings is Rigidity.

Usually, in math, things can wiggle around. But the authors found a "tipping point." If the Correlation Number hits a specific maximum value, it forces the entire library to be perfectly rigid.

• The Metaphor: Imagine two dancers. Usually, they can improvise. But if they move in perfect, locked synchronization (the "rigid" case), it means their dance steps are mathematically identical.
• The Result: If the Correlation Number is at its absolute maximum, it proves that the "Smell" and "Sound" are actually the same thing, just scaled up or down. If they are not the same, the number must be lower. This allows mathematicians to prove that two complex systems are different just by measuring this single number.

Why Does This Matter? (The Hitchin Representations)

You might be thinking, "Okay, but this is just a fancy library. Who cares?"

The authors apply this to Cusped Hitchin Representations. This is a very advanced topic in geometry and physics (related to how space is shaped and how particles might move in higher dimensions).

• The Connection: They realized that the complex, high-dimensional shapes of these "Hitchin representations" can be translated into our "infinite library" model.
• The Application: By using their new Correlation Number, they can now compare two different geometric shapes (representations) and say, "These two shapes are fundamentally different," or "These two shapes are actually the same."

Summary in Plain English

1. The Setup: We have a complex, infinite maze (Markov shift) with two different "measurements" (potentials) running through it.
2. The Map: We draw a curve (Manhattan Curve) that shows the limits of how these two measurements can combine.
3. The Discovery: We count how many paths fit a specific ratio on this curve. The speed at which this count grows is the Correlation Number.
4. The Breakthrough: This growth speed is mathematically identical to a geometric feature of the curve.
5. The Power: This allows us to compare complex geometric shapes (Hitchin representations) and determine if they are identical or different, even when they look very complicated.

In short, the authors built a mathematical ruler that measures the "similarity" of complex geometric worlds by counting paths in an imaginary library. It's a bridge between counting, geometry, and the deep structure of the universe.

Technical Summary

1. Problem Statement and Motivation

The paper addresses the problem of quantifying the relationship between the marked length spectra of two distinct geometric structures, specifically cusped Hitchin representations.

• Context: In Teichmüller theory and hyperbolic geometry, understanding the asymptotic distribution of closed geodesics (orbits) is central. Previous work (e.g., Lalley, Sharp, Pollicott) established correlation numbers for compact hyperbolic surfaces. However, extending these results to non-compact settings (surfaces with cusps) and higher-rank representations (Hitchin representations) presents significant analytical challenges due to the lack of compactness and the complexity of the symbolic dynamics involved.
• Goal: The authors aim to:
  1. Define a correlation number for pairs of strictly positive, locally Hölder continuous potentials on a countable-state Markov shift with the BIP (Big Images and Pre-images) property, assuming the potentials have strong entropy gaps at infinity.
  2. Establish the asymptotic growth rate of the number of periodic orbits where the ergodic sums of two potentials fall within a specific "Manhattan" region.
  3. Apply these dynamical results to cusped Hitchin representations to define a correlation number for pairs of such representations, thereby capturing rigidity properties of their length spectra.

2. Methodology

The authors employ a sophisticated blend of thermodynamic formalism, symbolic dynamics, and Fourier analysis. Their approach extends Lalley's seminal work on compact shifts to the non-compact, countable-state setting.

A. Symbolic Coding and Potentials

• Symbolic Model: They utilize a topologically mixing, countable-state Markov shift $\Sigma^+$ with the BIP property to code the recurrent portion of the geodesic flow on a geometrically finite hyperbolic surface with cusps.
• Potentials: They consider a pair of potentials $f, g: \Sigma^+ \to \mathbb{R}$ that are:
  • Strictly positive and locally Hölder continuous.
  • Independent: The linear combination $af + bg$ is arithmetic only if $a=b=0$.
  • Possessing strong entropy gaps at infinity: The series $Z_1(f, s) = \sum e^{-s \sup f}$ has a finite critical exponent $d(f)$ and diverges at $s=d(f)$. This condition ensures the system behaves well despite being non-compact.

B. The Counting Problem

The core dynamical object is the set of periodic points $x \in \text{Fix}_n$ (period $n$) such that the pair of ergodic sums $(S_n f(x), S_n g(x))$ lies in a specific rectangle $I^2_{\xi, m}(t) = (t, t+\xi) \times (mt, mt+\xi)$.
They define the counting function:
$$M(t; f, m, \xi) = \sum_n \frac{1}{n} \# \{ x \in \text{Fix}_n : S_n f(x) \in I^2_{\xi, m}(t) \}$$
where $m$ is an admissible slope derived from the geometry of the Manhattan curve.

C. Analytical Tools

1. Transfer Operators: They utilize the Ruelle-Perron-Frobenius transfer operator $\mathcal{L}_h$ associated with the potentials. They establish the holomorphicity of the map $(z, w) \mapsto \mathcal{L}_{zf+wg}$ in a complex domain.
2. Manhattan Curve: They analyze the curve $\mathcal{C}(f) = \{(a, b) : P(-af - bg) = 0\}$, where $P$ is the topological pressure. The slope of the normal to this curve defines the admissible slopes $m$.
3. Saddle-Point Method & Fourier Analysis: To derive precise asymptotics, they:
   • Convert the counting problem into a Laplace-Fourier transform of the transfer operator.
   • Use the Saddle-Point Method to approximate the integral near the critical point (the maximum of the pressure function).
   • Handle the non-compactness by establishing delicate global growth rate controls (upper bounds) and proving that "negligible" parts of the orbit distribution (those far from the expected slope) decay exponentially fast.

3. Key Contributions and Results

A. Dynamical Results (Symbolic Setting)

1. Existence of Correlation Number (Theorem A):
   They prove that the counting function $M(t; f, m, \xi)$ grows exponentially. The limit
   $$\alpha_f(m) = \lim_{t \to \infty} \frac{1}{t} \log M(t; f, m, \xi)$$
   exists and is finite and positive. This limit is defined as the correlation number.

2. Local Asymptotic Expansion (Theorem B):
   They provide a precise asymptotic estimate for the count restricted to a cylinder $p$:
   $$C_1(p) \leq \lim_{t \to \infty} t^{3/2} e^{-\alpha_f(m)t} M_p(t; f, m, \xi) \leq C_2(p)$$
   Crucially, the ratio $C_1(p)/C_2(p) \to 1$ as the cylinder length $|p| \to \infty$. This local expansion is essential for applications, as global expansion in non-compact settings is harder to prove directly.

3. Relation to Manhattan Curve (Theorem C):
   The correlation number is explicitly identified with a geometric quantity on the Manhattan curve. If $(a_m, b_m)$ is the point on the curve where the normal slope is $m$, then:
   $$\alpha_f(m) = a_m + m \cdot b_m$$
   This quantity is denoted $H_f(m)$.

4. Rigidity (Corollary 1.1):
   For the specific slope $m^* = \delta_f / \delta_g$ (ratio of entropies), the correlation number satisfies $H_f(m^*) \leq \delta_f$. Equality holds if and only if the potentials are proportional (i.e., $m^* S_n f = S_n g$ for all periodic points). This establishes a rigidity result: the correlation number detects when two systems are essentially the same.

5. Bishop-Steger Entropy Relation (Theorem D):
   They relate the correlation number to the $(\alpha, \beta)$-Bishop-Steger entropy, showing an inequality that becomes an equality only when the slope $m$ matches the ratio $\alpha/\beta$.

B. Geometric Applications (Cusped Hitchin Representations)

The authors apply the above theorems to cusped Hitchin representations $\rho, \eta: \Gamma \to SL(d, \mathbb{R})$.

• Coding: Using the work of Dal'Bo, Peigné, and Ledrappier, they code the geodesic flow of the surface $H^2/\Gamma$ into the Markov shift.
• Potentials: They associate strictly positive Hölder potentials $\tau^\phi_\rho$ and $\tau^\phi_\eta$ to the representations, where the ergodic sums correspond to the length functions $\ell^\phi_\rho(\gamma)$.
• Independence: They prove (Lemma 7.3) that if $\eta \neq \rho$ and $\eta \neq (\rho^{-1})^\top$, the associated potentials are independent.
• Theorem E: They define the correlation number for the pair of representations $(\rho, \eta)$ as the limit of the growth rate of pairs of lengths $(\ell^\phi_\rho(\gamma), \ell^\phi_\eta(\gamma))$. This number equals $a_m + m b_m$ derived from the Manhattan curve of the pair.
• Corollary 1.2: They establish a rigidity result for Hitchin representations: the correlation number is maximized if and only if the length spectra are proportional.

4. Significance

1. Extension to Non-Compact Settings: This is a major breakthrough in extending Lalley's correlation number theory from compact hyperbolic manifolds to non-compact (cusped) settings. The "strong entropy gap" condition is the key technical innovation that allows the thermodynamic formalism to work in infinite-volume spaces.
2. New Invariants for Higher Rank Teichmüller Theory: The paper provides a new, computable invariant (the correlation number) for pairs of Hitchin representations. This allows for the study of the geometry of the moduli space of these representations, particularly near the boundary (diverging sequences).
3. Rigidity: The results offer new rigidity theorems. The correlation number acts as a sensitive detector: if the number takes its maximal possible value, the two representations must be related by a specific algebraic symmetry (proportionality of length spectra).
4. Methodological Advancement: The paper successfully adapts the Saddle-Point Method and Fourier analysis to countable-state Markov shifts with BIP, overcoming the lack of a renewal theorem (which is available in the compact case) by using direct estimates on the transfer operator and large deviation principles.

In summary, the paper bridges deep dynamical systems theory with modern geometric representation theory, providing rigorous tools to analyze the asymptotic behavior of length spectra in non-compact, higher-rank geometric contexts.