arXiv:2609.34803
General Photon Subtraction from a Gaussian Perspective
Niklas Budinger, Ulrik L. Andersen, Peter van Loock
Abstract
A near-deterministic source of non-Gaussianity is the missing piece to reach universal and fault-tolerant quantum computing with continuous-variable optics. Due to a lack of strong non-linearities, the probabilistic photon-subtracted Gaussian states remain the most promising contender. In combination with Gaussian measurements and single-mode feed-forward operations, they have been shown to provide sufficient non-Gaussianity at high rates. However, these existing breeding protocols lack the necessary loss tolerance to be experimentally feasible. In this work, we introduce a mathematical formalism based on the application of a Gaussian blur and filter on the measured Fock Wigner function to describe general photon-subtracted Gaussian states. Within our representation, the initial squeezing, Gaussian measurements, and photon loss can all be understood as contributions to the total blurring and interchanged accordingly. We find that this intuitive approach can be used to find improvements to cat, cubic phase, and GKP state generation by compromising between success rates, feed-forward compatibility, and loss tolerance. In the case of multiple photon subtractions, this is achieved by introducing a hybrid setup variant bridging the gap between protocols based on breeding and post-selection. Finally, we establish a general lower bound on the resources needed to generate a given target state near-deterministically by introducing the expected Wigner logarithmic negativity as an additive monotone of non-Gaussianity.
AI Summary
The paper studies how photon-number-conditioned Gaussian optical circuits can generate non-Gaussian resource states while balancing success probability, Gaussian feed-forward compatibility, and photon loss. It rewrites the conditional Wigner function exactly as a Gaussian convolution of the detected Fock-state Wigner functions, followed by an affine coordinate transformation and Gaussian filtering. The convolution covariance is a Schur complement of the initial Gaussian covariance matrix, allowing finite squeezing and loss to be analysed within the same representation. Performance is measured using expected Wigner logarithmic negativity, $\mathrm{xWLN}=\sum_{\vec n}P_{\vec n}\mathrm{WLN}(\rho_{\vec n})$. Additivity and Gaussian monotonicity give the resource-accounting bound $N\geq\mathrm{WLN}(\rho_{\rm target})/\mathrm{xWLN}$ for deterministic conversion from identical independent sources. Numerical single-detector scans identify Fock-like outputs as maximising this quantity, although their outcome-dependent negativity rings prevent useful Gaussian alignment. At 15 dB squeezing and 5% loss, modified cat-like and cubic-phase-like configurations increase xWLN by 21% and 226%, respectively, relative to the standard configurations considered. For multiple detectors, a tunable hybrid construction interpolates between compact post-selection circuits and breeding protocols. In a two-detector GKP-like example, selected-pattern xWLN increases from 0.041 for the compact setup to 0.052 for a hybrid setup, versus 0.034 at full added squeezing. The analysis combines Gaussian covariance identities with numerical Wigner integration. Limitations include heuristic feed-forward-compatibility measures rather than target-state fidelity guarantees, imperfect hybrid feed-forward, residual post-selection, and computationally prohibitive multimode optimisation. Extension of the single-detector maximum to a universal per-detector resource bound remains conjectural.
Error found: The universal optimal resource bound per detector or squeezing budget is not established, because the single-detector maximum is assessed numerically and the multi-detector extension remains conjectural. The stated minimum-copy ratio needs a success-probability qualification: exact preparation with probability p requires N·xWLN ≥ p·WLN(target), not necessarily N·xWLN ≥ WLN(target). The added vacuum projection has success probability sech²(ξ₀/2), rather than the stated sech(ξ₀/2), although its effect on the reported numerical results cannot be determined from the text.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 20 pages, 21 figures