A Time-Frequency Framework for GKP Codes

Technical Summary: A Time–Frequency Framework for GKP Codes

Problem Statement
The paper addresses the mathematical representation of Gottesman–Kitaev–Preskill (GKP) codes, which encode finite-dimensional logical information into infinite-dimensional bosonic oscillator states. While ideal GKP codewords are well-understood as distributional grid states, a rigorous framework connecting these ideal states to finite-dimensional logical vectors via localized phase-space coefficients has been lacking. Specifically, the authors seek to answer how the finite-dimensional logical information of an ideal GKP codeword is represented by localized time–frequency coefficients and how stably this information can be reconstructed from a finite set of such coefficients. Furthermore, the work aims to bridge the gap between ideal distributional states and normalizable physical states (finite-energy approximations) within a unified coefficient model.

Methodology
The authors develop a framework grounded in time–frequency analysis, specifically utilizing modulation spaces, Gabor analysis, and the Zak transform.

1. Functional Setting: Ideal codewords are realized in the modulation space $M^\infty(\mathbb{R}^n)$, the continuous dual of the Feichtinger algebra $M^1(\mathbb{R}^n)$. This choice is critical as it restricts the space of solutions to the stabilizer equations to those with uniformly bounded Zak transform coefficients, excluding pathological distributions (e.g., derivatives of Dirac masses) allowed in the full space of tempered distributions $\mathcal{S}'$.
2. Vector-Valued Zak Transform: The authors employ a vector-valued Zak transform to separate the continuous displacement syndrome from the finite-dimensional logical fibre. For a symplectically integral lattice $\Lambda$, the ideal code is identified with a finite logical fibre $H_d$ supported at the origin of the syndrome torus $\mathbb{R}^{2n}/\Lambda^\circ$.
3. Adjoint-Lattice Coefficients: The framework utilizes multi-window Gabor analysis indexed by the symplectic adjoint lattice $\Lambda^\circ$. The authors establish that the logical information is encoded in a finite block of coefficients sampled from the adjoint lattice, corresponding to the quotient group $K_\Lambda = \Lambda^\circ/\Lambda$.
4. Regularization: To address the non-normalizability of ideal states, the authors construct normalizable approximations using lattice-envelope Gabor multipliers. These are formed by weighting the stabilizer orbit sum with a decaying envelope sequence $w \in \ell^2(\Lambda)$.

Key Contributions and Results

• Isometric Logical Reconstruction: The paper proves that the normalized block map, which extracts a finite block of adjoint-lattice Gabor coefficients from an encoded codeword, is an isometry from the logical Hilbert space $H_d$ to $\ell^2(K_\Lambda)$. This establishes a bijection between the logical vector and the sampled coefficient block.
• Explicit Recovery Projection: An explicit orthogonal projection $\Pi_0$ is derived, which maps arbitrary coefficient data to the space of admissible exact logical blocks. This allows for the stable reconstruction of the logical vector from noisy or incomplete coefficient data via a least-squares approach.
• Minimal Sampling: The authors demonstrate that for specific probes (e.g., Gaussian probes), the logical vector can be reconstructed from a minimal number of coefficients ($d = \dim H_d$), which is the theoretical minimum required to span the logical space.
• Normalizable Approximants: The paper constructs normalizable GKP states as Gabor multipliers acting on ideal codewords. It proves that for asymptotically translation-invariant envelopes, these regularized states yield asymptotically isometric logical encodings. Furthermore, it establishes weak-$\ast$ convergence of these regularized states to the ideal codeword in $M^\infty$.
• Syndrome Extraction via Block Correlation: The displacement syndrome is characterized as a phase relation between translated finite coefficient blocks. The authors derive a block-correlation estimator that recovers the syndrome character by correlating these blocks. They provide a deterministic stability estimate for this estimator under additive coefficient perturbations, showing that redundancy in stabilizer directions improves robustness.

Significance and Scope
The paper claims to provide a "concrete time–frequency coefficient model" that connects existing geometric descriptions of GKP codes (involving theta functions and Heisenberg modules) with a practical coefficient-based representation.

• Mathematical Rigor: By working within $M^\infty$, the authors ensure that the bijection between ideal codewords and logical vectors is well-defined and excludes non-physical distributional solutions.
• Unified Framework: The work unifies the description of logical information (encoded in a single fundamental block), syndrome information (encoded in the phase relations between translated blocks), and time–frequency coefficients as different aspects of the same lattice structure.
• Limitations and Scope: The authors explicitly state that their results are mathematical statements conditional on access to complex time–frequency coefficient functionals. The paper does not propose a physical measurement protocol or a statistical noise model for quantum hardware. "Reconstruction" refers strictly to the recovery of logical vectors from coefficient data, and "coefficient noise" is treated as a deterministic additive perturbation. The framework serves as a theoretical foundation for understanding the stability and structure of GKP codes in the coefficient domain, rather than a direct experimental proposal.

In summary, the paper establishes that the logical information of ideal GKP codes is fully and stably captured by a finite block of adjoint-lattice Gabor coefficients, and that this structure persists asymptotically for normalizable approximations, providing a robust mathematical basis for logical reconstruction and syndrome estimation in the time–frequency domain.