Integrability breaking in semiclassical strings in Koopman-Krylov space

This paper introduces a Koopman-Krylov framework based on the Koopman-von Neumann description and gEDMD approximation to characterize non-integrable semiclassical string dynamics, demonstrating how integrability-breaking deformations induce observable-dependent delocalization and spectral weight redistribution in Krylov space.

The Gist

Imagine you are watching a perfectly choreographed dance troupe. In an integrable system (a perfectly ordered one), every dancer moves in a predictable, repeating pattern. If you know the steps of one dancer, you can predict exactly where they will be a minute from now. The whole group moves like a well-oiled machine, never colliding or getting lost. In physics, this is like a string vibrating in a perfect, unchanging universe.

But what happens if you slightly nudge the music? Maybe you change the tempo just a tiny bit, or add a new instrument. This is integrability breaking. The dance doesn't immediately turn into a chaotic mosh pit. Instead, it becomes "messy" in a very specific way: some dancers stay in their perfect lines, while others start to drift, wobble, or get stuck in small, confusing loops.

This paper is about a new way to measure how messy that dance gets, specifically for "strings" in theoretical physics (which are tiny, vibrating loops that make up the universe in string theory).

The Problem with Old Tools

Traditionally, physicists tried to spot chaos by watching individual dancers (trajectories). They asked: "If I move this dancer's hand a millimeter to the left, do they end up in a completely different spot?" If yes, it's chaos.

But this is like trying to understand a traffic jam by watching a single car. It tells you if that car is stuck, but it doesn't tell you why the whole gridlock happened or which lanes are clogged. In the "weakly chaotic" systems the authors study, the messiness is subtle. It's not a total crash; it's a slow, confusing drift that depends entirely on which dancer you are watching.

The New Tool: The "Koopman-Krylov" Lens

The authors introduce a new method called Koopman-Krylov. Let's break this down with an analogy:

1. The Koopman Shift: From Dancers to the Music
Instead of watching the dancers move, imagine you are watching the music itself.

• In the old way, you track the dancer's position (non-linear, hard to predict).
• In the Koopman way, you track how the song changes as the dance happens. Even if the dance is wild and unpredictable, the way the song evolves can be described by simple, linear rules. It's like shifting your focus from the chaotic crowd to the steady beat of the drum.

2. The Krylov Space: The "Echo Chamber"
Now, imagine you take a single note from that song (an "observable") and let it bounce around a room.

• Integrable (Perfect Order): The note bounces around and creates a clean, repeating echo. It stays in a small, tidy corner of the room.
• Integrability Breaking (The Mess): As you nudge the system, that note starts to spread out. It hits different walls, bounces into new corners, and mixes with other sounds. It "delocalizes."

The Krylov method is a way to measure exactly how far that note spreads and how much it mixes with other sounds.

What They Found

The authors applied this "Echo Chamber" test to three different types of "string dances" that were slightly broken:

1. The "Spin Chain" Strings (SU(2) & SU(3)):
   Imagine a string made of beads. In the perfect version, the beads spin in perfect sync. When they added a "deformation" (a slight glitch in the physics), they found that the "echo" of the music didn't spread everywhere at once.

   • The Surprise: If you listened to the energy of the beads, the echo spread a lot. But if you listened to the momentum (how fast they were spinning), the echo stayed mostly in the corner.
   • The Lesson: Chaos isn't a universal "on/off" switch. It depends entirely on what you are measuring. Some parts of the system get messy; others stay calm.

2. The "Winding" Strings (AdS5 × T1,1):
   Here, they looked at strings wrapped around a donut-shaped space.

   • The Result: When they broke the integrability, the "echo" didn't get louder or faster in a dramatic way. Instead, the shape of the echo changed. The frequencies of the sound shifted and rearranged themselves.
   • The Lesson: The "messiness" was hiding in the spectrum (the mix of frequencies), not in the speed of the spread. It was a subtle reshuffling of the musical notes rather than a loud crash.

The Big Picture: Why This Matters

Think of the universe as a giant, complex machine. For a long time, physicists thought, "If it's not perfectly ordered, it must be total chaos."

This paper says: "Not so fast."

There is a huge middle ground between "Perfect Order" and "Total Chaos." In this middle ground:

• The system is mostly ordered, but with thin layers of confusion.
• The "confusion" only affects specific parts of the system, depending on how you look at it.
• Old tools (like watching a single dancer) miss these subtle shifts.
• The new Koopman-Krylov tool acts like a high-tech sound engineer, allowing us to hear exactly which notes are getting out of tune and how the music is rearranging itself.

In a Nutshell

This paper teaches us that when the universe gets a little "broken," it doesn't just fall apart. It reorganizes. By listening to the "music" of the system (using Koopman theory) and measuring how the notes spread out (using Krylov space), we can map out exactly where the order is breaking down and where it is still holding strong. It's a new way to understand the subtle, messy beauty of a universe that isn't perfectly perfect.

Technical Summary

Here is a detailed technical summary of the paper "Integrability breaking in semiclassical strings in Koopman-Krylov space" by Rathindra Nath Das and Saskia Demulder.

1. Problem Statement and Motivation

The paper addresses the challenge of characterizing non-integrable dynamics in semiclassical string theory, specifically within the context of the planar AdS/CFT correspondence.

• Context: While integrable sectors of string theory (e.g., planar $\mathcal{N}=4$ SYM) are well-understood, they are fragile. Small deformations (e.g., higher-loop corrections, geometric deformations of the background) typically break integrability.
• The Gap: Standard tools for detecting non-integrability, such as Poincaré sections and Lyapunov exponents, provide binary indicators (chaotic vs. regular) or measure trajectory divergence. They often fail to capture the mechanism of how integrability breaks down in "weakly" chaotic regimes where phase space is a mix of regular islands and thin chaotic layers.
• Goal: The authors aim to develop a systematic probe to understand how spectral weight redistributes and how dynamical structures reorganize when integrability is weakly broken, moving beyond simple trajectory divergence to an observable-based perspective.

2. Methodology: The Koopman-Krylov Framework

The authors introduce a novel framework that bridges classical mechanics and quantum many-body complexity theory by adapting Krylov methods to classical systems via the Koopman-von Neumann (KvN) formalism.

A. Koopman-von Neumann Formulation

Instead of tracking phase-space points $x(t)$, the dynamics are lifted to the evolution of observables (functions on phase space) under a linear operator.

• Koopman Operator ($U_\sigma$): Evolves an observable $g$ via composition: $(U_\sigma g)(x) = g(F^\sigma x)$.
• Koopman Generator ($K$): The infinitesimal generator of this flow, defined as $K = -i\{ \cdot, H \}$, where $\{ \cdot, H \}$ is the Poisson bracket. This maps the nonlinear Hamiltonian flow to a linear, infinite-dimensional evolution problem on the Hilbert space of observables $L^2(X, \mu)$.

B. Krylov Construction

The authors construct a Krylov subspace generated by repeated action of the Koopman generator $K$ on a seed observable $g_0$:
$$\mathcal{K} = \text{span}\{ g_0, K g_0, K^2 g_0, \dots \}$$
This mimics the "operator growth" seen in quantum chaos but applies to classical observables.

C. Numerical Implementation: gEDMD

Since the Koopman generator is infinite-dimensional, the authors use Generator Extended Dynamic Mode Decomposition (gEDMD) to approximate it:

1. Dictionary Selection: A finite basis of functions (e.g., sines, cosines, polynomials) is chosen to span the observable space.
2. Data Sampling: Trajectories are sampled from the phase space (restricted to a narrow energy shell to ensure fair comparison).
3. Projection: The generator is projected onto the dictionary to obtain a finite matrix approximation $\hat{K}$.
4. Lanczos Tridiagonalization: The matrix $\hat{K}$ is tridiagonalized to extract the Krylov basis and the associated spectral measure (eigenvalues and weights).

D. Diagnostics

The paper defines several metrics to quantify integrability breaking:

• Krylov Spread ($C_K$): The mean Krylov index explored, measuring the "drift" of spectral weight.
• Inverse Participation Ratio (IPR): Measures the localization of the Krylov wavefunction. A decay in IPR indicates delocalization (mixing).
• Wasserstein-1 Distance ($W_1$): A metric comparing the spectral measures of integrable vs. deformed systems. It quantifies the "transport" of spectral weight, distinguishing between rigid shifts and genuine reshaping of the spectrum.
• Sector-Resolved Leakage: Measures how much spectral weight leaks from "protected" (integrable) sectors into "deformation-activated" sectors.

3. Key Contributions

1. Formalism Extension: Successfully extends Krylov complexity methods, previously restricted to quantum systems, to classical Hamiltonian systems using the KvN formalism.
2. Observable-Dependent Chaos: Demonstrates that in weakly non-integrable systems, chaos is not universal. Different observables probe different resonant channels, leading to distinct signatures of integrability breaking.
3. Spectral vs. Trajectory Analysis: Shows that spectral transport (reshaping of the Koopman spectrum) can be a more sensitive and informative probe of weak chaos than Lyapunov exponents, which may remain small or zero in near-integrable regimes.

4. Results: Three Case Studies

The framework is applied to three distinct semiclassical string models:

Case 1: Two-Loop SU(2) Dilation Operator (Landau-Lifshitz Limit)

• Setup: A deformation of the integrable SU(2) spin chain via higher-derivative terms (Ostrogradsky formalism).
• Mechanism: The deformation enlarges the phase space (introducing higher-derivative degrees of freedom).
• Findings:
  • Delocalization: The deformation induces a genuine broadening of the Krylov distribution and increased entropy.
  • Observable Dependence: Observables coupling to the new transverse degrees of freedom show significant leakage into non-protected sectors, while others remain confined.
  • Spectral Transport: The Wasserstein distance grows systematically with the deformation parameter $a_1$, indicating resonance-driven redistribution.

Case 2: Leigh-Strassler Deformed SU(3) Dilation Operator

• Setup: A deformation of the SU(3) sector involving complex parameters ($\beta, \kappa$).
• Mechanism: The deformation introduces a non-separable sextic potential on the same phase space (no new dimensions).
• Findings:
  • Energy Dependence: The signatures of chaos are highly energy-dependent. Low-energy shells show stronger sensitivity to deformation than high-energy shells.
  • Resonant Coupling: The mixing observable ($g_{mix}$) shows stronger delocalization than angular observables, confirming that chaos is driven by specific resonant couplings rather than global mixing.
  • Spectral Reshaping: The Wasserstein distance reveals a continuous redistribution of spectral weight away from discrete peaks, even when Lyapunov exponents are small.

Case 3: Near-Penrose Limit of $AdS_5 \times T^{1,1}$

• Setup: Two scenarios: (A) Motion with an AdS radial component (Integrable) and (B) Motion restricted to the internal $T^{1,1}$ sector (Non-integrable).
• Findings:
  • Integrable Case (A): The Koopman-Krylov diagnostics show near-rigidity. Despite the Hamiltonian being deformed, the spectral measures and Krylov spread remain virtually unchanged, validating the method's specificity to integrability breaking.
  • Non-Integrable Case (B): The restricted $T^{1,1}$ motion exhibits weak chaos.
    • Krylov spread and delocalization are enhanced but remain strongly observable-dependent.
    • The Wasserstein distance is the most robust signature, showing significant spectral transport.
    • This confirms that in this regime, chaos is encoded at the spectral level (reshaping of frequencies) rather than through rapid global phase-space diffusion.

5. Significance and Conclusion

• Nuanced View of Chaos: The paper argues that "weak chaos" in classical strings is not a uniform phenomenon. It is organized by resonant channels and is highly dependent on the choice of observable.
• Diagnostic Power: The Koopman-Krylov framework provides a granular view of how integrability erodes:
  • It distinguishes between deformations that merely shift frequencies (integrable) and those that reshape the spectrum (non-integrable).
  • It identifies which dynamical directions (sectors) are activated by the breaking of integrability.
• Future Directions: The authors suggest applying this framework to LLM microstate geometries (to study trapping) and pseudointegrable billiards (to probe phase-space topology).

In summary, this work establishes a rigorous, observable-based methodology to detect and characterize the onset of non-integrability in semiclassical strings, revealing that the breakdown of integrability is a spectral transport phenomenon driven by specific resonances, rather than a universal exponential instability.