Realizing the Petz Recovery Map on an NMR Quantum Processor

This paper reports the first experimental realization of the Petz recovery map on a nuclear magnetic resonance quantum processor using duality quantum computing, demonstrating how tuning the reference state enables effective, noise-adapted recovery from amplitude and phase damping errors and validating the map's physical implementability beyond its theoretical formulation.

The Gist

The Big Idea: Fixing a Broken Message with a "Custom Key"

Imagine you are trying to send a secret message written on a piece of paper. But, the paper has to travel through a stormy tunnel (the "noise") that smudges the ink, tears the edges, or changes the colors. By the time the paper arrives, the message is hard to read.

In the world of quantum computers, this "storm" is called decoherence. It's the natural tendency of quantum bits (qubits) to lose their special properties and get messed up by their environment.

Scientists have long known about a mathematical tool called the Petz Recovery Map. Think of this map as a "magic eraser" or a "repair kit" designed to un-smudge the paper and restore the original message. However, there's a catch: this repair kit isn't a one-size-fits-all tool. It is a custom-made key. To work perfectly, the repair kit must be built specifically for the type of damage the paper suffered and for the specific style of the original message.

The Problem: Until now, this "magic repair kit" existed mostly on paper (in math equations). No one had successfully built it in a real lab to prove it actually works.

The Solution: This paper reports the first successful experiment of building and testing this Petz Recovery Map on a real quantum machine (an NMR processor). They proved that if you tune the "repair kit" correctly, you can fix the damage. If you tune it wrong, it makes things worse.

---

How They Did It: The "Shadow Puppet" Trick

Building a quantum repair kit is tricky because the damage (noise) isn't a simple flip of a switch; it's a messy, non-reversible process. You can't just "undo" it with a standard quantum move.

To get around this, the researchers used a clever technique called Duality Quantum Computing (DQC).

• The Analogy: Imagine you want to simulate a messy spill of water (the noise) on a table. You can't just pour water and expect to clean it up easily. Instead, you use a system of mirrors and shadows (ancilla qubits). You set up a complex dance of shadows that looks like the water spilled, even though the actual table remains dry and controlled.
• The Experiment: They used a molecule (diethyl fluoromalonate) dissolved in liquid as their quantum computer. This molecule has three tiny magnets (nuclei) acting as their "bits."
  • One bit was the Message (the system qubit).
  • Two other bits were the Helpers (ancilla qubits) used to create the "shadow puppet" simulation of the noise and the repair.

They simulated two specific types of "storms":

1. Amplitude Damping: Like a battery losing energy. The message naturally wants to fall asleep (go to a "zero" state).
2. Phase Damping: Like a spinning top wobbling until it loses its rhythm. The message loses its timing and rhythm, but not its energy.

The "Tuning Knob" Experiment

The most important part of this experiment was testing the Reference State.

Think of the Petz Recovery Map as a pair of noise-canceling headphones.

• If you are listening to Rock music (a specific type of noise), you need headphones tuned to cancel out Rock frequencies.
• If you put those same Rock-headphones on while listening to Jazz, they won't work; they might even make the sound worse.

In the experiment, the researchers acted as the "tuner." They built the repair kit based on a specific "Reference State" (a guess about what the original message looked like).

What they found:

1. The Match: When the "Reference State" they used to build the repair kit matched the actual message they were trying to save, the repair worked beautifully. The message was restored with high clarity.
2. The Mismatch: When they used a "Reference State" that didn't match the message, the repair failed. In fact, the "repair kit" actually made the message more garbled than the noise did.

Example from the paper:

• If they tried to fix a message that had lost its energy (Amplitude Damping) using a repair kit designed for a message that was already asleep, it worked great.
• But if they tried to use that same kit on a message that was still awake, it failed.

Why This Matters (According to the Paper)

This paper doesn't claim to have built a perfect quantum computer that can fix all errors yet. Instead, it proves a fundamental concept:

The Petz Recovery Map is a real, physical thing, not just a math trick.

It shows that:

• You can physically build a device that reverses noise.
• But you must know exactly what kind of noise happened and what the original message looked like to build the right "key."
• It bridges the gap between abstract math theory and a real, working laboratory experiment.

Summary

The researchers took a complex mathematical idea for fixing broken quantum information, built a physical version of it using a liquid molecule and magnetic pulses, and proved that it works—but only if you tune it to the specific type of damage and the specific message you are trying to save. If you guess the wrong settings, the "fix" actually breaks the message further. This is a major step in understanding how to protect quantum information in the real world.

Technical Summary

Technical Summary: Realizing the Petz Recovery Map on an NMR Quantum Processor

Problem Statement
The Petz recovery map is a fundamental construct in quantum information theory designed to reverse the effects of noise on quantum systems. Unlike standard error correction or diagnostic tools like quantum process tomography, the Petz map is intrinsically dependent on a chosen reference state. While theoretically motivated by the saturation of the data-processing inequality for quantum relative entropy, its physical implementation has remained challenging due to its explicit dependence on the noise channel, its adjoint, and a tunable reference state. Prior to this work, experimental realizations were limited, with only a recent implementation in trapped-ion platforms reported. The central challenge addressed is bridging the gap between the abstract, information-theoretic formulation of the Petz map and its practical realization on a realistic quantum platform, specifically demonstrating its state-adapted nature.

Methodology
The authors experimentally realized Petz recovery maps on a nuclear magnetic resonance (NMR) quantum processor using a three-qubit system composed of $^{1}$H, $^{19}$F, and $^{13}$C nuclei in a diethyl fluoromalonate molecule. The system qubit was $^{19}$F, while $^{1}$H and $^{13}$C served as ancillary qubits.

To implement the non-unitary noise channels and the corresponding recovery maps, the team employed the Duality Quantum Computing (DQC) algorithm. This approach allows for the coherent implementation of linear combinations of unitary operators, effectively simulating non-unitary Kraus operators by decomposing them into unitary components acting on an ancillary register.

• Noise Models: Two paradigmatic single-qubit noise channels were investigated:
  1. Amplitude Damping (AD): Modeling energy dissipation, driven by Kraus operators $K_0$ and $K_1$.
  2. Phase Damping (PD): Modeling loss of coherence without population change, driven by Kraus operators $K_0$ and $K_1$.
• Reference States: The Petz maps were constructed for specific reference states ($\sigma$) tailored to the noise structure:
  • For AD: A diagonal state $\sigma = (1-\epsilon)|0\rangle\langle 0| + \epsilon|1\rangle\langle 1|$.
  • For PD: A state diagonal in the Pauli-X basis, $\sigma = (1-\epsilon)|+\rangle\langle +| + \epsilon|-\rangle\langle -|$.
• Experimental Protocol: The system was initialized in a pseudopure state (PPS). The DQC algorithm was used to simulate the damping channel followed immediately by the Petz recovery map. The output state was reconstructed via quantum state tomography (using a reduced set of pulses to trace out the ancilla), and fidelity was calculated against the original input state.

Key Contributions

1. First NMR Realization: This work reports the first experimental realization of Petz recovery maps on an NMR quantum processor, extending the scope of such implementations beyond trapped-ion systems.
2. Demonstration of State-Adapted Recovery: The study explicitly demonstrates that the Petz map is not a universal inverse but a state-adapted recovery channel. The authors systematically tuned the reference state parameter ($\epsilon$) to show that recovery fidelity is maximized when the reference state aligns with the input state's characteristics and degrades when they are mismatched.
3. Algorithmic Implementation: The paper successfully adapts the DQC algorithm, previously used for simulating noise channels, to implement the inverse recovery operations, proving that ancilla-assisted unitaries can effectively reverse non-unitary open-system dynamics.

Results
The experimental results showed close quantitative agreement with theoretical predictions for both AD and PD channels across various input states and reference state parameters ($\epsilon = 0.2, 0.5, 0.8$).

• Amplitude Damping: For an input state $|1\rangle$, recovery fidelity increased as the reference state parameter $\epsilon$ increased (shifting the reference toward $|1\rangle$). Conversely, for an input state $|0\rangle$, higher $\epsilon$ values led to fidelity degradation. The $|0\rangle$ state remained robust against AD noise regardless of recovery, as the channel naturally drives systems to $|0\rangle$.
• Phase Damping: For input states $|+\rangle$ and $|-\rangle$, the recovery fidelity was highly sensitive to the choice of $\epsilon$. A reference state close to the input state (e.g., $\epsilon=0.2$ for $|+\rangle$) yielded high recovery fidelity, while a mismatched reference (e.g., $\epsilon=0.8$ for $|+\rangle$) resulted in lower fidelity. Notably, when $\epsilon=0.5$ (maximally mixed reference), the Petz map acted as a PD channel itself, further reducing fidelity rather than recovering it.
• Mismatched Inputs: For input states that did not overlap significantly with the chosen reference basis (e.g., computational basis states under a PD map constructed with X-basis references), no significant recovery was observed, confirming the channel-specific and state-specific nature of the Petz map.

Significance
The paper claims that this work establishes an experimental benchmark for testing noise-adapted recovery strategies on near-term quantum devices. By validating the Petz map as a physically realizable, reference-state-dependent channel rather than a purely formal inverse, the study bridges the gap between abstract information theory and laboratory implementation. The authors posit that these results provide direct evidence for the operational interpretation of the Petz map and offer a framework for exploring the fundamental limits of information recovery in open quantum systems. The work suggests that such channel-adapted strategies can be extended to larger systems and more complex noise models, providing insights for developing noise mitigation techniques in future quantum technologies.