⚛️ quantum physics

Chasing shadows with Gottesman-Kitaev-Preskill codes

This paper proposes a shadow tomography protocol for logical subsystems defined by Gottesman-Kitaev-Preskill codes that utilizes measurement twirling to extract encoded information from arbitrary input states, demonstrating specific applications for heterodyne and photon parity measurements to enable efficient estimation of bounded observables via Gaussian decompositions and Wigner sampling.

Original authors: Jonathan Conrad, Jens Eisert, Steven T. Flammia

Published 2026-01-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jonathan Conrad, Jens Eisert, Steven T. Flammia

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a very complex, high-dimensional sculpture made of light (a quantum state). You want to know what it looks like, but you can't just take a photograph because the sculpture is too delicate and the camera is too blurry. This is the challenge of "quantum tomography": trying to figure out the shape of a quantum object by taking measurements.

This paper introduces a clever new way to take "snapshots" of these quantum sculptures, specifically those built using a special type of error-correcting code called GKP codes. Think of GKP codes as a way to hide a simple, logical message (like a single bit of information) inside a messy, infinite ocean of physical noise.

Here is the core idea, broken down with simple analogies:

1. The Problem: The "Shadow" is Blurry

Usually, to understand a quantum state, you have to measure it. But if you measure it directly, you might destroy the information or get a result that is too noisy to make sense of.
The authors use a technique called "Shadow Tomography." Imagine you are trying to guess the shape of an object in a dark room by throwing darts at it and seeing where they land. Instead of trying to reconstruct the whole object perfectly, you just want to know specific things about it (e.g., "Is it round?" or "How heavy is it?").

2. The Trick: "Twirling" the Mess

The paper's main innovation is a mathematical trick called "twirling."

The Analogy: Imagine you have a messy, tangled ball of yarn (the noisy physical quantum state). You want to find a specific pattern hidden inside (the logical information).
The Action: Instead of trying to untangle the yarn perfectly, you spin the ball of yarn rapidly in random directions (this is the "twirl").
The Result: When you spin it fast enough, the messy parts average out, and the core pattern becomes visible in a very specific, predictable way. In the paper, they "twirl" the measurement process using random operations (Gaussian unitaries) that are natural to the system. This turns a messy, complex measurement into a clean, simple one that still tells you about the hidden logical message.

3. Two Ways to Take the Snapshot

The paper shows how to do this "twirling" with two different types of cameras (measurements):

A. The "Heterodyne" Camera (The Gaussian Decomposition)

How it works: This camera takes a picture that looks like a fuzzy cloud (a Gaussian state).
The Magic: The authors show that if you take many of these fuzzy pictures after "twirling" the system, you can mathematically combine them to recreate the logical information of the original state.
The Benefit: It's like taking a blurry photo of a complex machine and realizing that if you overlay enough of these blurry photos, you can mathematically reconstruct a clear blueprint of the machine's logic, even if the machine itself is physically messy. This allows scientists to simulate how these quantum machines would behave using standard computers.

B. The "Photon Parity" Camera (The Wigner Sampling)

How it works: This camera counts whether there is an even or odd number of photons (light particles).
The Magic: This is related to a famous mathematical map called the "Wigner function," which is like a topographical map of the quantum state.
The Benefit: The paper shows that by randomly choosing where to look on this map (based on the GKP code's structure), you can estimate the properties of the state without needing to map the entire thing. It's like estimating the average height of a mountain range by randomly sampling a few points, rather than measuring every single rock.

4. The "Random Code" Superpower

Finally, the paper takes this a step further. Usually, these methods rely on knowing exactly which "code" (the specific pattern of the GKP lattice) you are using.

The Innovation: The authors show that if you randomly choose a different code every time you take a measurement, you can build a "universal" shadow.
The Result: You can estimate properties of any quantum state, not just ones that fit a specific code. It's like having a universal translator that works regardless of the specific language the quantum state is "speaking," as long as you randomize your approach enough.

Summary

In short, this paper provides a new toolkit for scientists working with continuous quantum systems (like light or sound waves). It shows how to:

Spin the measurement process to filter out noise.
Convert messy physical data into clean, classical descriptions (like Gaussian states or Wigner maps).
Randomize the process so it works for any quantum state, not just the perfect ones.

This allows researchers to "chase shadows"—gathering just enough information to understand the logical heart of a quantum system without needing to perfectly reconstruct the entire, infinitely complex physical object.

Technical Summary: Chasing Shadows with Gottesman–Kitaev–Preskill Codes

Problem Statement
The paper addresses the challenge of performing shadow tomography on logical subsystems within continuous variable (CV) quantum systems, specifically those encoded using Gottesman–Kitaev–Preskill (GKP) error-correcting codes. While classical shadow tomography has been successfully established for discrete variable (qubit) systems using Clifford twirling, extending these protocols to infinite-dimensional CV systems is non-trivial due to the inaccessibility of random operations over the full infinite Hilbert space. Furthermore, logical observables in GKP systems are often not directly measurable; they require indirect inference via physical measurements (e.g., heterodyne or photon parity) that may not align naturally with the logical structure. The authors seek to develop a protocol that extracts classical shadows of logical information from arbitrary physical input states without requiring the input to be a perfect code state, while providing rigorous bounds on sample complexity.

Methodology
The core methodology relies on twirling techniques adapted for CV systems and GKP codes. The protocol operates by applying random logical Clifford gates (implemented as Gaussian unitary operations) to an input state before performing a physical measurement. This process effectively "twirls" the measurement channel, projecting it onto a depolarizing channel relative to the logical code space.

Key technical components include:

Logical Clifford Twirling: The authors define a twirl over the group of logical Clifford automorphisms of the GKP code. For GKP codes, these logical gates correspond to symplectic lattice automorphisms, which can be implemented via Gaussian unitaries.
Displacement Twirling: To handle the infinite-dimensional nature of the system, the protocol incorporates a displacement twirl over the symplectic dual lattice ( $L^\perp$ ) of the GKP code. This suppresses non-stabilizer coherences, effectively projecting the state or channel onto the code space (or a depolarized version thereof).
POVM Twirling: The protocol applies these twirls to specific physical Positive Operator Valued Measures (POVMs). The authors analyze two primary cases:
- Heterodyne Measurement: The pointer states are Gaussian. The protocol yields a probabilistic decomposition of the input state into an ensemble of Gaussian states that reproduce the logical information of the input.
- Photon Parity Measurement: This is linked to Wigner function sampling. The protocol samples the Wigner function of the input state at random points determined by the GKP lattice structure.
Randomized Lattice Selection: To move beyond a specific code and estimate arbitrary CV observables, the authors introduce a protocol that averages over a random choice of GKP codes (random symplectic lattices). This utilizes the Haar measure over the space of symplectic lattices (Siegel's mean value theorem) to ensure the shadow protocol is not biased toward a specific code structure.

Key Contributions and Results

GKP Logical Shadow Protocol: The authors establish a framework for shadow tomography where the "shadow" is a classical description of the logical state. They prove that for a fixed GKP code, the protocol yields an estimator for logical expectation values with sample complexity bounds derived from the "shadow norm," analogous to the qubit case but adapted for the logical subspace.
Gaussian Decomposition (Heterodyne): For heterodyne measurements, the protocol provides a rigorous method to decompose an arbitrary physical CV state into a convex combination of Gaussian states. The authors prove bounds on the "Gaussian compressibility," showing that a finite number of Gaussian samples suffices to faithfully retain the logical content of the input state, even without prior knowledge of the state's analytical form.
Wigner Sampling (Photon Parity): For photon parity measurements, the protocol is shown to be equivalent to a randomized Wigner sampling scheme. The authors derive sample complexity bounds for estimating bounded observables, demonstrating that the required number of samples depends on the variance of the observable in phase space.
Random Code Shadow Tomography: By randomizing over the choice of the GKP lattice, the authors demonstrate how to construct a shadow protocol for the full physical CV Hilbert space. This allows for the estimation of any sufficiently bounded observable without restricting the analysis to a specific code space. The sample complexity bounds for this general case are derived using second moments of functions over the space of symplectic lattices.

Significance and Claims
The paper positions itself as a bridge between continuous variable physics and randomized tomographic methods originally developed for discrete systems. The authors claim that their work provides:

Experimental Handles: A practical "black-box" procedure to convert arbitrary physical input states (including noisy or imperfect GKP states) into classical descriptions (Gaussian mixtures or Wigner samples) that capture logical information.
Robustness: The protocol is inherently robust to imperfections in the POVM, as the twirling process projects noisy measurements onto a depolarizing channel with modified parameters, which can be accounted for in post-processing.
Pedagogical Insight: The work aims to refine the general understanding of classical shadow protocols by viewing them through the lens of GKP codes, highlighting the interplay between physical structure (CV systems) and logical content (error-corrected information).
Simulation Utility: The Gaussian decomposition result offers a potential tool for classically simulating GKP error correction and assessing the performance of quantum computations using real GKP states, as Gaussian states are efficiently simulable.

The authors remain modest regarding the scaling of sample complexity, acknowledging that while bounds are derived, they may scale exponentially with system size (particularly for full state tomography or high-dimensional lattices). However, they emphasize the value of obtaining rigorous, assumption-free bounds on the number of samples needed to estimate logical observables, a capability previously lacking for general CV states in this context.

1. The Problem: The "Shadow" is Blurry

2. The Trick: "Twirling" the Mess

3. Two Ways to Take the Snapshot

4. The "Random Code" Superpower

Summary

More like this