The Petz recovery map for optical losses

This paper investigates the Petz recovery map as an approximate method for correcting optical losses in quantum information processing, demonstrating that for Gaussian states it utilizes either beam-splitters or amplifiers to achieve near-optimal performance that outperforms simple state re-preparation but may underperform compared to doing nothing when the reference state is inaccurate, while also extending these findings to two-mode systems.

The Gist

The Big Picture: Fixing a Broken Signal

Imagine you are trying to send a secret message using a beam of light (like a laser pointer). As the light travels through the air or a fiber optic cable, it doesn't just stay perfect; it gets "leaky." Some of the light scatters away into the room, and the signal gets weaker and fuzzier. In the quantum world, this is called optical loss.

The big problem is that once that light scatters into the room (an "unmonitored mode"), it's gone. You can't just grab it back. The paper asks: If we know what the message should have looked like before it got messed up, can we build a machine to fix the signal as best as possible?

The answer is "yes, but not perfectly." The authors investigate a specific mathematical tool called the Petz recovery map to see how well it can "undo" the damage.

---

The Setup: The Leaky Bucket Analogy

To understand the problem, imagine your information is water in a bucket.

• The Loss: As you carry the bucket, it has a hole in the bottom. Water leaks out, and the bucket gets mixed with some dirty water from the floor (the "environment").
• The Goal: You want to pour the remaining water back into a clean bucket that looks exactly like the original.
• The Catch: You can't just pour the water back; you have to use a specific set of tools (like funnels, pumps, or filters) to try to restore it.

The paper focuses on a specific type of leak: Gaussian loss. This is a very common, smooth type of noise found in real-world optics, where the signal gets dimmer and noisier in a predictable way.

The Tool: The "Petz Map"

The authors are testing a specific recipe for fixing the signal called the Petz recovery map. Think of this map as a "smart guesser."

1. The Reference State (The Prior): Before the leak happens, you have a "best guess" of what the water in the bucket looked like. Maybe you know it was usually warm and slightly salty. This is your Reference State.
2. The Magic: The Petz map uses this reference state to figure out how to reverse the leak. It's like a Bayesian update (a fancy way of saying "updating your belief based on new evidence"). It asks, "Given that I see this messy, leaky water, and I know what the clean water usually looks like, what is the most likely way to fix it?"

What Did They Find?

1. The Fixer Changes Its Shape

The most surprising finding is that the "fixer" (the Petz map) doesn't always look the same. Depending on how much light was lost and what the "reference" water looked like, the fixer becomes one of two things:

• A Beam Splitter (The Filter): If the loss isn't too bad and the reference is similar to the loss, the fixer acts like a filter. It just separates the good signal from the noise.
• An Amplifier (The Booster): If the loss is severe or the reference is different, the fixer acts like a volume knob turned up. It boosts the weak signal.
  • Why? Because when light is lost, the remaining signal is very quiet. To hear it, you have to turn up the volume (amplify it), even though turning up the volume also makes the background hiss (noise) louder. The Petz map calculates the perfect amount of volume boost to recover the message without making the noise too terrible.

2. Is It Better Than Doing Nothing?

The authors compared the Petz map to two other simple strategies:

• Strategy A (Do Nothing): Just keep the messy, leaky water.
• Strategy B (Throw it Away): Ignore the messy water entirely and just pour in a fresh bucket of the "reference" water you guessed earlier.

The Results:

• Petz vs. Throw Away: The Petz map is always better than just throwing the signal away and replacing it with a guess. It actually uses the information that is still there.
• Petz vs. Do Nothing: This depends on your guess.
  • If your "Reference State" (your guess of the original) was close to the real signal, the Petz map works wonders and is much better than doing nothing.
  • If your guess was wildly wrong (e.g., you thought the water was hot, but it was actually freezing), the Petz map might make things worse. In that case, it's actually better to just leave the messy signal alone than to try to "fix" it with a bad guess.

3. Is It the Best Possible Fix?

The paper shows that while the Petz map isn't a "perfect" fix (you can't get 100% of the lost light back), it is near-optimal.

• Among a whole class of possible fixing machines, the Petz map is usually the best or very close to the best.
• The less the signal was damaged to begin with, the closer the Petz map gets to being perfect.

4. Two Buckets at Once (Two Modes)

Finally, the authors looked at what happens if you are sending two beams of light at the same time that are entangled (linked together like a pair of magic dice).

• Local Fix: You try to fix each beam separately.
• Global Fix: You treat the two beams as a single unit and fix them together.

The Result: The Global Fix is better at saving the "connection" (entanglement) between the two beams. However, if the beams are very bright or the connection isn't too strong, fixing them separately works just as well. It's like trying to untangle two knots: sometimes you need to look at the whole rope, but sometimes you can just fix each knot individually.

Summary

This paper is about finding the best way to repair a damaged quantum light signal.

• They found a mathematical recipe (Petz map) that acts like a smart filter or a smart amplifier.
• It works best when you have a good idea of what the signal looked like before it got damaged.
• It is generally better than ignoring the damage or just guessing a new signal, but it can't fix everything if your guess was terrible.
• When dealing with multiple signals, fixing them together is usually better for keeping their special quantum connections intact.

The paper doesn't claim this will immediately fix all quantum computers, but it provides a very strong, near-perfect tool for handling the inevitable "leaks" in optical quantum systems.

Technical Summary

Technical Summary: The Petz Recovery Map for Optical Losses

Problem Statement
Optical systems are a primary platform for quantum information processing, yet they suffer from information loss due to scattering into unmonitored modes. In the context of continuous-variable (CV) quantum information, where information is encoded in the state of a mode rather than the mode itself, these losses are modeled as state-independent beam-splitter interactions with a thermal environment (often approximated as the vacuum). It is a established result that perfect error correction for such Gaussian lossy channels is impossible using only Gaussian operations. Consequently, the paper investigates the Petz recovery map as a method for approximate recovery of states degraded by optical losses.

Methodology
The authors analyze the Petz recovery map, $\mathcal{P}_{\mathcal{N},\sigma}$, defined for a quantum channel $\mathcal{N}$ and a reference state $\sigma$. The map is constructed as a quantum analog of a Bayesian update, utilizing the prior $\sigma$ to reverse the channel's action.

• Framework: The study focuses on single-mode and two-mode Gaussian systems. The forward loss channel $\mathcal{N}$ is modeled as a beam splitter with transmissivity $\eta$ coupling the signal to a Gaussian environment $\xi$.
• Derivation: Assuming both the reference state $\sigma$ and the environment $\xi$ are Gaussian, the authors derive the specific parameters (transformation matrix $X_P$, noise matrix $Y_P$, and displacement $d_P$) of the resulting Petz recovery map, which is itself a Gaussian channel.
• Performance Metrics: The recovery performance is evaluated using the fidelity between the initial input state $\rho$ and the recovered state. The Petz map is compared against two trivial protocols:
  1. $R^{(0)}$: Doing nothing (keeping the noisy state).
  2. $R^{(1)}_\sigma$: Discarding the noisy state and re-preparing the reference state $\sigma$.
• Optimality Analysis: The authors restrict the search space to a class of recovery protocols $\mathcal{R}_{\mathcal{N},\sigma}$ that perfectly recover the reference state $\sigma$ (satisfying specific covariance and displacement constraints) and compare the Petz map's fidelity against the optimal protocol within this class.
• Extension: The analysis is extended to two-mode lossy channels, comparing "local" recovery (independent single-mode Petz maps) against "global" recovery (a joint two-mode Petz map utilizing an entangled reference state).

Key Contributions and Results

1. Characterization of the Single-Mode Petz Map:

   • For single-mode losses with Gaussian reference states, the Petz recovery map is found to be implementable as either a beam splitter or a phase-insensitive amplifier, depending on the parameters.
   • Thermal Environment Limit: When the environment is the vacuum ($n_\xi=0$), the Petz map acts as a beam splitter only if the reference state is also the vacuum ($n_\sigma=0$). For all other thermal reference states, the Petz map acts as a phase-insensitive amplifier. This suggests the map attempts to amplify the remaining signal information rather than merely restoring the average photon number.
   • Generalized Transmissivity: The authors define a generalized transmissivity $\eta'$, which can take values in $[0, \infty)$. If $\eta' > 1$, the map is an amplifier; if $0 \le \eta' < 1$, it is a beam splitter.

2. Comparison with Trivial Protocols:

   • Vs. Re-preparation ($R^{(1)}_\sigma$): The Petz map always outperforms the protocol that simply discards the noisy state and re-prepares the reference state, provided the input is a thermal state.
   • Vs. Doing Nothing ($R^{(0)}$): The Petz map does not always outperform doing nothing. The authors derive a condition (Result 2) showing that if the reference state $\sigma$ is significantly different from the true input state $\rho$, the Petz recovery may yield lower fidelity than simply retaining the noisy state.

3. Near-Optimality:

   • Within the class of recovery protocols that perfectly recover the reference state $\sigma$, the Petz map demonstrates near-optimal performance.
   • In the specific case where the environment is the vacuum and the reference is thermal, the Petz map saturates the complete positivity (CP) bound, effectively choosing the largest possible transmissivity to preserve the maximum amount of information from the input while introducing minimal noise.
   • Numerical results indicate that the relative fidelity difference between the Petz map and the optimal recovery in this class is generally below 25%, dropping below 15% when the input state shares the same structure (zero mean vector) as the reference state.

4. Two-Mode Extension and Entanglement Recovery:

   • For two-mode systems, the authors compare local recovery (separable reference) with global recovery (entangled reference).
   • Entanglement: Global Petz recovery maps generally outperform local maps in recovering quantum correlations (measured by logarithmic negativity). The use of an entangled reference state introduces correlation-preserving structures directly into the recovery channel.
   • Trade-offs: While global recovery is superior for correlations, the authors note that increasing the squeezing parameter or mean photon number of the reference state can degrade overall state fidelity. Thus, a careful trade-off is required to balance correlation recovery against state fidelity.

Significance and Claims
The paper claims that the Petz recovery map provides a robust, near-optimal strategy for approximating the reversal of optical losses in Gaussian systems without requiring the complex encoding stages typically associated with quantum error correction.

• It establishes that the Petz map is not merely a theoretical construct but has a concrete physical implementation (beam splitters or amplifiers) for Gaussian channels.
• It clarifies the limitations of the approach: the map is highly dependent on the choice of the reference state. If the prior is inaccurate (far from the true state), the recovery can be detrimental compared to inaction.
• The work highlights the advantage of global operations in multi-mode settings for preserving entanglement, suggesting that exploiting shared quantum correlations during the recovery process is beneficial, though parameter optimization is necessary to maintain high state fidelity.

The authors do not claim to have achieved perfect correction or to have proposed a new experimental architecture beyond the theoretical implementation of the map using standard optical components (beam splitters, amplifiers, and ancilla states). The significance lies in characterizing the performance and physical nature of the Petz map in a realistic optical loss scenario.