Decoherence-free subspaces in the noisy dynamics of discrete-step quantum walks in a photonic lattice

This paper theoretically and experimentally demonstrates that while temporal noise constant within a Floquet period preserves coherence in the bulk of a discrete-step quantum walk on a photonic lattice by creating decoherence-free momentum subspaces, such protection fails for topological edge states and is completely lost under fully random noise.

The Gist

Imagine a quantum walk as a very precise, magical game of "tag" played by a particle of light. In a perfect world, this particle hops from one spot to another in a grid, following strict rules. Because it's a quantum particle, it doesn't just take one path; it takes all paths at once, creating a beautiful, complex interference pattern (like ripples in a pond overlapping) that allows it to move much faster and more efficiently than a normal particle.

However, in the real world, things aren't perfect. There is "noise"—little jitters and glitches in the environment that mess up the rules. Usually, this noise ruins the magic, turning the quantum game into a boring, slow, classical shuffle.

This paper investigates what happens to our light particle when we introduce different types of noise into its "track" (a photonic lattice made of fiber optic loops). The researchers discovered something surprising: sometimes, the noise doesn't matter at all.

Here is the breakdown of their findings using simple analogies:

1. The Two Types of Noise

The researchers tested two ways of messing with the game:

• Random Noise (The "Chaotic DJ"): Imagine a DJ who changes the beat randomly every single second. Sometimes it's fast, sometimes slow, with no pattern.

  • The Result: The quantum particle gets completely confused. The beautiful interference patterns vanish almost instantly. The particle loses its "quantumness" and starts behaving like a normal, slow-moving object. The noise destroys the magic.

• Stroboscopic Noise (The "Synchronized DJ"): Imagine a DJ who changes the beat randomly, but only once every full song cycle. For the entire duration of that song, the beat stays exactly the same, even if it's different from the previous song.

  • The Result: This is where the magic happens. The researchers found that for certain specific "directions" (momentum) the particle is traveling, the noise cancels itself out. Even though the rules changed from song to song, the particle found a "safe zone" where the noise didn't affect it at all. These are called Decoherence-Free Subspaces. It's like walking through a storm where, for a specific path, the raindrops magically stop hitting you.

2. The Edge of the Map (Topological Edge States)

The researchers also looked at what happens when the particle is stuck at the very edge of the grid (a "topological edge state"). Think of this as a particle trapped in a corner of a room that usually can't escape.

• The Result: Unlike the "safe zones" in the middle of the grid, the edge is not safe. No matter if the noise is random or synchronized, the particle eventually loses its quantum coherence. The noise always finds a way to disturb the particle when it's on the edge.

3. How They Proved It

To test this, the team built a giant, high-tech "track" using two loops of fiber optic cable (like a racetrack made of glass). They shot laser pulses into the loops.

• The loops were slightly different lengths, so the light pulse would arrive at different times, effectively simulating a grid of many steps.
• They used electronic modulators to introduce the "noise" (jittering the rules) exactly as they predicted.
• They measured the light pulses over and over again (100 times) to see the average result.

The Experiment Confirmed the Theory:

• When they used Random Noise, the interference patterns disappeared, and the light spread out chaotically.
• When they used Synchronized (Stroboscopic) Noise, the interference patterns stayed strong for specific directions, proving the existence of those "decoherence-free" safe zones.
• When they looked at the Edge, the light lost its coherence in both scenarios.

The Takeaway

The paper shows that while noise usually kills quantum effects, there is a special trick: if the noise changes in a synchronized way (stroboscopic), you can find specific paths where the noise simply doesn't exist. However, this protection doesn't work for particles trapped at the edges of the system; they remain vulnerable to any kind of noise.

This is a fundamental discovery about how quantum systems behave when they aren't perfect, showing that the timing of the noise is just as important as the noise itself.

Technical Summary

Technical Summary: Decoherence-free subspaces in the noisy dynamics of discrete-step quantum walks in a photonic lattice

Problem Statement
Discrete-step quantum walks (DSQWs) serve as a versatile framework for exploring quantum transport, designing algorithms, and simulating complex condensed-matter phenomena, including topological phases and non-Hermitian dynamics. However, real-world implementations are inevitably subject to environmental noise and decoherence, which can degrade coherence, suppress interference, and induce classical dynamics, thereby limiting the utility of quantum walks as simulators or quantum processors. While previous studies have addressed noise in quantum walks, often relying on perturbative approaches or specific noise models, there remains a need to understand how different types of temporal noise—specifically those constant within a driving period versus fully random—affect the dynamics of both bulk and topological edge states in periodically driven (Floquet) systems.

Methodology
The authors investigate the dynamics of DSQWs in a one-dimensional photonic lattice subject to temporal noise. The study employs a dual approach combining theoretical derivation and experimental realization:

1. Theoretical Framework:

   • The system is modeled as a cascade of unitary operators in a 1D lattice with a characteristic Floquet period.
   • Instead of relying on second-order perturbation theory (which yields a Lindblad form), the authors derive a non-perturbative master equation for the system's density matrix, averaged over noise realizations. This is achieved by re-summing all orders of the perturbative series.
   • Two noise scenarios are analyzed:
     • Stroboscopic Noise: Noise that is constant within a single Floquet period but uncorrelated between periods ($\theta_m \to \theta_m + \tau_M$).
     • Random Noise: Noise that changes randomly at every discrete time step ($\theta_m \to \theta_m + \tau_m$).
   • The analysis is performed for both bulk dynamics (using momentum space representation) and topological edge states (using real-space representation for finite lattices).

2. Experimental Implementation:

   • Experiments are conducted using a time-multiplexed photonic lattice realized in a double-fiber ring setup.
   • The setup consists of two coupled fiber loops of slightly different lengths, where the lattice position is encoded in the pulse arrival time.
   • A short laser pulse (1.4 ns) undergoes split-step walk dynamics via a variable beamsplitter and phase modulators.
   • Different types of temporal noise are engineered by modulating the splitting angles of the variable beamsplitter.
   • The system operates in the classical regime (coherent laser sources), but the results are applicable to quantum information processing.
   • Data processing involves heterodyne measurement to extract both amplitude and phase, followed by coherent averaging over 100 noise realizations to reconstruct the density matrix dynamics and band structures.

Key Contributions and Results

• Decoherence-Free Subspaces in the Bulk (Stroboscopic Noise):
  The study reveals that when noise is constant within a Floquet period (stroboscopic noise), the bulk dynamics exhibits decoherence-free momentum subspaces. Specifically, for momenta satisfying $k = \phi + (2p + 1)\pi$ (where $\phi$ is the phase modulation and $p \in \mathbb{Z}$), the noise-induced terms in the master equation vanish. Consequently, wavepackets with these specific momenta remain immune to temporal noise fluctuations. This is a non-perturbative effect; perturbative expansions to second order would miss the specific noise processes responsible for these subspaces.

• Destruction of Coherence by Random Noise:
  In contrast, when the noise is fully random within the Floquet period (changing at every step), the decoherence-free subspaces disappear. The master equation shows that the noise modifies the free evolution term at every step, leading to the rapid destruction of coherence and the loss of interference patterns within a few time steps.

• Decoherence of Topological Edge States:
  The authors find that topological edge states are not protected by decoherence-free subspaces, regardless of whether the noise is stroboscopic or random. The edge state population decays over time.

  • For stroboscopic noise, the decay is initially exponential but transitions to a polynomial decay at longer times. This slowing down is attributed to the dynamics of localized modes in the bulk flat bands, which feed back into the edge state.
  • This behavior contrasts with dispersive bulk bands, where edge state decay remains predominantly exponential.

• Experimental Verification:
  The experimental results confirm the theoretical predictions.

  • In the noiseless case, clear spatiotemporal interference fringes are observed.
  • Under random noise, these fringes are fully destroyed, and the signal decays rapidly.
  • Under stroboscopic noise, interference patterns persist longer, and momentum-resolved measurements show inhomogeneous broadening: minimal broadening near the decoherence-free momenta ($k = \pm \pi$) and maximal broadening at other momenta ($k=0$).
  • Edge state experiments demonstrate the predicted decay dynamics, with the return probability showing an initial exponential drop followed by a slower, polynomial decay consistent with the influence of bulk flat bands.

Significance and Claims
The paper claims that the identification of controllable decoherence-free subspaces in momentum space offers a mechanism to manipulate and filter information in noisy quantum walk architectures. The authors emphasize that their non-perturbative master equation is crucial for correctly predicting these dynamics, as standard perturbative approaches fail to capture the specific noise-induced processes that preserve coherence in certain subspaces.

The work suggests that these findings are directly relevant for architectures based on cascades of unitary operators (such as time-multiplexed fiber loops) where noise may occur on timescales longer than the Floquet period. The ability to engineer noise to protect specific momentum components could be beneficial for building robust registers and simulating open quantum systems with enhanced resilience to temporal noise. The authors maintain a modest scope, noting that while their experiments are classical, the results apply to quantum communications and photonic quantum information processing, without proposing specific new applications beyond these general domains.