A tridiagonal matrix-valued process with stochastic resetting for arbitrary Dyson index $\beta>0$

This paper introduces a symmetric tridiagonal matrix-valued process with stochastic resetting, demonstrating that simultaneous resetting yields an analytically solvable stationary eigenvalue distribution identical to resetting Dyson Brownian motion, while independent resetting produces a distinct ensemble that is studied numerically and applied to compute the annealed partition function of a disordered quantum system.

The Gist

Imagine a crowded dance floor where $N$ dancers are moving around. In the world of this paper, these dancers aren't just people; they represent the "eigenvalues" (special numbers) of a giant, complex machine called a matrix. Usually, these dancers push each other away (repulsion) while being gently pulled back toward the center of the room (a trap). This specific dance is known as Dyson Brownian Motion.

For a long time, scientists knew exactly how this dance looked when the dancers were special types of people (specifically for three mathematical "flavors" called $\beta = 1, 2, 4$). They could describe the dance by imagining the dancers were actually the shadows of a giant, shifting machine. But for any other "flavor" of dancer ($\beta > 0$), no one knew what the underlying machine looked like.

This paper introduces a new, clever way to build that machine for any type of dancer, and then adds a twist: Stochastic Resetting.

Here is the breakdown of their discovery using everyday analogies:

1. Building the Machine (The $\beta$-TMP)

To make the dancers move correctly for any type, the authors built a specific kind of machine: a Tridiagonal Matrix. Think of this machine as a long, narrow hallway with rooms only next to each other (no diagonal shortcuts).

• The Walls (Diagonal Entries): The walls of the rooms move back and forth randomly, like a drunk person stumbling in a straight line but always trying to return to the center. In math, this is called an Ornstein-Uhlenbeck process.
• The Doors (Off-Diagonal Entries): The doors connecting the rooms are trickier. They can't just be negative numbers; they must be positive. The authors made these doors move like a Cox-Ingersoll-Ross (CIR) process. Imagine a door that swings open and closed, but the harder it swings, the more likely it is to be pushed back. It's a "bouncing" motion that stays positive.

By carefully tuning how the walls and doors move, the authors proved that the shadows cast by this machine (the eigenvalues) perfectly match the complex dance of the particles, no matter what "flavor" ($\beta$) they are.

2. The Twist: Stochastic Resetting

Now, imagine a game master standing in the corner with a stopwatch. Every now and then, the game master yells "RESET!"

• The Rule: When the game master yells, everything stops. Every dancer is instantly teleported back to their starting line (the origin), and the game starts over from scratch. This happens randomly, like a clock ticking at a steady average rate.
• The Result: Even though the dancers are constantly being thrown back to the start, they eventually settle into a new, stable pattern of movement called a Non-Equilibrium Stationary State (NESS). They don't stop moving, but their overall distribution of positions becomes predictable and unchanging over time.

3. Two Ways to Reset

The paper explores two different ways the game master can yell "RESET":

• Scenario A: The "Simultaneous" Reset (SRTMP)
  The game master yells, and every single dancer is teleported back to the start at the exact same moment.

  • The Finding: The authors found a beautiful, exact mathematical formula for where the dancers end up in this scenario. Surprisingly, this formula works for any type of dancer ($\beta > 0$). It turns out this new pattern is the same as the one found in a previous study for the special "flavors" of dancers. This proves that their new machine works perfectly for the whole universe of these particles.

• Scenario B: The "Independent" Reset (IRTMP)
  The game master yells, but this time, each dancer has their own private timer. Dancer A might get reset, while Dancer B keeps dancing, and then Dancer C gets reset later. They are reset independently.

  • The Finding: This is much messier. Because the dancers are reset at different times, they don't share a "history" of being thrown back together. The authors couldn't find a simple math formula for where these dancers end up. However, they used computers to simulate this scenario.
  • The Surprise: When they compared the computer simulation of the "Independent" reset dancers to the "Simultaneous" reset dancers, the patterns were completely different. The "Independent" group looked nothing like the "Simultaneous" group, proving that how you reset the system changes the final outcome drastically.

4. A Real-World Application: The Disordered Lattice

Finally, the authors showed how this math applies to a real physics problem: a single quantum particle hopping along a one-dimensional ring (like a bead on a wire) where the "hopping rates" (how easily it jumps between spots) are random and disordered.

• They used their "Simultaneous Reset" machine to model the disorder in the wire.
• Because they had the exact formula for the dancers' positions (the energy levels of the particle), they could calculate the average energy (free energy) of the system perfectly.
• They discovered that in the limit of a very long wire, the energy of the system is dominated by the disorder itself, and the temperature of the system barely matters.

Summary

In short, this paper built a universal "machine" (a specific type of matrix with moving walls and doors) that generates the correct behavior for a complex system of interacting particles for any parameter. They then showed that if you constantly reset this system, you get a stable, predictable pattern. They proved this works perfectly if you reset everyone at once, but if you reset everyone individually, the pattern changes completely, and we still don't have a simple formula to describe it. This new understanding allows physicists to calculate the energy of disordered quantum systems with perfect precision.

Technical Summary

Technical Summary: A Tridiagonal Matrix-Valued Process with Stochastic Resetting for Arbitrary Dyson Index $\beta > 0$

Problem Statement
The paper addresses the challenge of constructing an underlying matrix-valued process whose eigenvalues evolve according to the $\beta$-Dyson Brownian motion ($\beta$-DBM) for arbitrary Dyson index $\beta > 0$, specifically in the context of stochastic resetting. While it is well-established that for the special cases $\beta = 1, 2, 4$, the $\beta$-DBM corresponds to the eigenvalues of a Gaussian matrix with independent Ornstein-Uhlenbeck (OU) entries, a general matrix construction for arbitrary $\beta$ had not been explicitly defined. Furthermore, the authors investigate how this matrix process behaves under stochastic resetting—where the system is interrupted at random times and restarted from an initial condition—and whether a "resetting matrix ensemble" exists that reproduces the known stationary distribution of the resetting $\beta$-DBM (resetting Dyson Brownian motion, $\beta$-RDBM) for general $\beta$.

Methodology
The authors employ a combination of stochastic process theory, random matrix theory, and renewal theory:

1. Construction of the $\beta$-TMP: The authors define a symmetric tridiagonal matrix-valued process, $\beta$-TMP, $H(t)$.

   • Diagonal entries: Evolve as independent OU processes starting from the origin.
   • Off-diagonal entries: Evolve as independent Cox-Ingersoll-Ross (CIR) processes (a specific case of Feller processes) starting from the origin. Crucially, the parameters of the CIR processes depend on the row index $k$, specifically the shape parameter $q = \beta(N-k)/2$.
   • The authors derive the exact time-dependent joint probability density function (JPDF) of the matrix entries and, via a change of variables involving the Vandermonde determinant, show that the JPDF of the eigenvalues matches the known propagator of the $\beta$-DBM for all $t$ and arbitrary $\beta > 0$.

2. Stochastic Resetting Framework: The paper utilizes the renewal equation formalism to introduce stochastic resetting with rate $r$. Two distinct resetting protocols are applied to the $\beta$-TMP:

   • $\beta$-SRTMP (Simultaneous Resetting): All matrix entries are reset to the origin simultaneously at random times drawn from an exponential distribution with rate $r$.
   • $\beta$-IRTMP (Independent Resetting): Each matrix entry resets to the origin independently with rate $r$.

3. Analytical and Numerical Analysis:

   • For the $\beta$-SRTMP, the authors analytically compute the stationary JPDF of the matrix entries by integrating the time-dependent propagator over the exponential distribution of resetting times. They then derive the stationary JPDF of the eigenvalues.
   • For the $\beta$-IRTMP, the stationary distribution of the matrix entries is derived as a product of independent distributions (since entries reset independently). However, the authors note that the JPDF of the eigenvalues for this ensemble cannot be easily computed analytically. Consequently, they generate the stationary $\beta$-IRTMP ensemble numerically using the inverse cumulative distribution function (CDF) method to compute the average eigenvalue density.

4. Application: The stationary $\beta$-SRTMP ensemble is applied to a disordered quantum tight-binding Hamiltonian on a 1D lattice. The authors compute the annealed partition function and free energy exactly, leveraging the known stationary eigenvalue distribution.

Key Contributions and Results

• Generalization of Matrix-Valued Processes: The primary contribution is the explicit construction of the $\beta$-TMP, a tridiagonal matrix process with mixed OU and CIR dynamics that yields the $\beta$-DBM eigenvalue statistics for arbitrary $\beta > 0$. This generalizes the known results for $\beta = 1, 2, 4$ (which rely on standard Gaussian matrices) to the full range of $\beta$.
• Exact Stationary Distribution for $\beta$-SRTMP: The authors prove that the $\beta$-SRTMP process, where all entries reset simultaneously, reaches a non-equilibrium stationary state (NESS). The JPDF of the eigenvalues in this NESS is shown to coincide exactly with the known stationary distribution of the $\beta$-RDBM (Eq. 8 in the paper) for any $\beta > 0$. This confirms that the $\beta$-SRTMP is the correct underlying matrix process for the resetting Dyson Brownian motion in the general $\beta$ case.
• Distinction Between Resetting Protocols: The paper highlights a fundamental difference between simultaneous and independent resetting:
  • In $\beta$-SRTMP, the matrix entries become highly correlated in the stationary state due to the shared history of resetting events. However, the eigenvalue statistics are analytically tractable.
  • In $\beta$-IRTMP, the matrix entries remain independent in the stationary state (similar to Wigner matrices), making numerical generation straightforward. However, the resulting eigenvalue statistics are analytically intractable. Numerical results demonstrate that the average eigenvalue density of the $\beta$-IRTMP ensemble differs significantly from the analytical density of the $\beta$-SRTMP ensemble.
• Application to Disordered Systems: The paper provides a concrete application by computing the annealed free energy of a disordered tight-binding model where hopping rates and on-site potentials are drawn from the stationary $\beta$-SRTMP ensemble. The authors derive an exact expression for the annealed free energy, showing that in the large $N$ limit, it scales as $\sqrt{N}$ and becomes independent of temperature, dominated by the disorder/energy landscape.

Significance
The paper claims significance in providing a unified framework for understanding the $\beta$-DBM and its resetting variants through a matrix-valued process. By constructing the $\beta$-TMP, the authors bridge the gap between the stochastic dynamics of interacting particles (Dyson Brownian motion) and matrix ensembles for arbitrary $\beta$. The introduction of stochastic resetting to this matrix framework allows for the study of non-equilibrium stationary states (NESS) in random matrix theory. The work generalizes previous results limited to $\beta = 1, 2, 4$ to the entire range $\beta > 0$, offering a new tool for analyzing disordered quantum systems and stochastic many-body systems. The distinction drawn between simultaneous and independent resetting protocols offers new insights into how correlation structures emerge in non-equilibrium matrix ensembles.