arXiv:2610.34257
Minimal ancilla for boosting type-I photonic fusion
A. A. Melkozerov, S. S. Straupe, M. Yu. Saygin
Abstract
Fusion gates enable the creation of large entangled photonic states for linear-optical quantum computing and communication protocols. However, their success probability is limited to $1/2$ without additional photonic resources. Previous schemes for boosting type-I fusion have used four ancillary single photons or a two-qubit Bell state, whose preparation is probabilistic and resource intensive. Here we propose a boosted type-I fusion gate using only two independent ancillary single photons and no entangled ancillary input. It achieves a direct success probability of $5/8$, matching that of the previous four-single-photon proposal while using half as many ancillary photons. The circuit is predominantly passive, with a single active element that applies a phase shift or swaps two modes according to intermediate photodetection outcomes. Unlike the previous schemes, successful outputs occupy fixed modes and require no further conditional output routing. In particular, we show that the gate is more robust to heralded ancillary-photon loss than previously proposed boosted gates and reduces the average number of single photons required to generate a six-qubit ring graph state by more than $4.6\%$ compared with other type-I schemes. It therefore offers a resource-efficient primitive for scalable fusion-based photonic quantum information processing.
AI Summary
The paper asks how few ancillary photons suffice to boost dual-rail type-I photonic fusion above its unassisted success probability of $1/2$. It presents a circuit using two independent ancillary single photons that achieves $P_s=5/8$, matching the direct success probability of a previous four-single-photon construction without an entangled ancillary input. The circuit is passive except for one feed-forward element implementing an identity, phase correction, or two-mode swap. Successful outputs occupy fixed modes, without subsequent conditional output routing. The result is established by deriving conditional optical output states and classifying photon-number-resolving detection patterns. Two outcome classes contribute $11/32+5/32=1/2$; another contributes $1/8$ after conditional swapping and further detection. The latter outcomes are counted as successes through local-Clifford equivalence for Bell, GHZ, and graph-state inputs. Resource comparisons model independent, heralded ancillary-photon preparation with probability $p_{anc}$, giving an average success probability $1/2+p_{anc}^2/8$. With four sources and $4\rightarrow2$ multiplexing, the scheme outperforms the compared idealized asymptotic boosting procedures for $p_{anc}< approximately\,0.84$. Sequential linear-cluster growth consumes eight ancillary photons per net added qubit. Under ideal, lossless, perfectly resource-efficient multiplexing assumptions, example preparation strategies require approximately $71.5$ photons per three-qubit GHZ state and $605.7$ per six-qubit ring, reductions of $4.3\%$ and $4.7\%$ against the four-photon-boosted comparison. The robustness analysis concerns heralded preparation failures, not general optical loss. Type-I fusion can leave photon loss undetected until later measurements; the resource estimates do not include realistic switching, detection, or propagation losses.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 10 pages, 3 figures