A new preprint proposes generating optical GKP states at 85% fidelity with 3.76 dB of squeezing, framing itself as a simpler alternative to cat state schemes.
A theoretical preprint available on arXiv outlines a route to optical GKP states, the hardware building block several photonic quantum-computing roadmaps depend on, by sending a single-photon-added squeezed vacuum (SPASV) through a 50:50 beam splitter and watching for a specific photon-number click on the other output port. The authors report a peak fidelity of 85% to a finite-energy GKP target at 3.76 dB of squeezing, and they position the result as a simpler and more experimentally accessible alternative to building the same states out of squeezed odd Schrödinger cat states (preprint: "Theoretical Study for Generating Optical GKP State via a Single-Photon-Added Squeezed Vacuum").
The scheme is conditional. A squeezed vacuum source feeds a photon-addition stage, producing the SPASV; that mode is then interfered with vacuum on a beam splitter, and a heralding detector on one port fires only when a particular photon-number pattern shows up. When the herald succeeds, the remaining optical mode collapses toward a finite-energy approximation of an ideal GKP grid state. The probability of that success is bounded, and the 85% figure sits at the peak of a curve that trades fidelity against heralding rate and against how much squeezing the source has to deliver in the first place.
That trade-off is the actual point. The authors compare their peak to the peak fidelities reported in earlier theoretical work on squeezed optical odd Schrödinger cat states and argue the SPA-squeezed-vacuum route reaches a higher ceiling for a given squeezing budget. That comparison is a model-internal claim, not an independent benchmark. The two numbers come from different theoretical analyses with different resource assumptions, and neither has been demonstrated in the laboratory at the reported fidelity. The honest read is that 85% at 3.76 dB is a coordinate on a design map, useful for locating one proposal against another, not a verdict that one ancestor state has won.
For a photonic hardware team, picking an ancestor state commits the group to a specific shopping list. The SPA-squeezed-vacuum route wants a high-purity continuous-wave squeezed vacuum source, a heralded photon-addition stage (typically a low-reflectivity tap and a single-photon detector), a low-loss 50:50 beam splitter, and a number-resolving detector for heralding. The squeezed-odd-cat route asks for a different non-Gaussian operation: a photon-subtraction stage on a squeezed vacuum, engineered to produce a specific odd-parity superposition. Each resource carries its own loss budget, and each heralding scheme has its own success rate, and those budgets compound before either path ever approaches the fidelities a fault-tolerance argument would actually want.
The 85% number itself sits well below what most fault-tolerance analyses require of a GKP ancilla. Photonic fault tolerance generally needs GKP states whose finite-energy envelopes approach the ideal grid closely enough that the residual error sits well under the relevant code threshold after realistic GKP-to-GKP gates, with realistic overhead. 85% theoretical fidelity is the kind of starting point a group uses to argue that a future experiment is worth running, not a result that lets a compiler close a resource estimate today. The paper's own framing is consistent with that: it presents the protocol as a step toward fault-tolerant photonic quantum computing, not as a demonstration of it.
What would move the proposal from map to milestone? A first experiment would have to assemble the SPASV source and the heralding apparatus, characterize the prepared state's Wigner function against the finite-energy GKP target, and report a measured fidelity alongside a measured heralding rate at a stated squeezing level. The natural next variable to vary is squeezing itself. The paper's curve shows fidelity climbing with the squeezing parameter r, and the experimental question is how much of that climb survives realistic loss, mode mismatch, and detector inefficiency. Until that experiment exists, the SPA-squeezed-vacuum proposal is best read as a concrete, quantitative entry in the design-fork conversation about which non-Gaussian resource optical GKP generation should anchor on, alongside the Schrödinger-cat line it claims to improve on.
The full theoretical scheme, the squeezing-dependent fidelity curve, and the resource assumptions that produce the 85% figure are in the preprint itself. The arXiv listing was confirmed via direct fetch; the title, abstract, and authors are consistent with the draft's characterization.