Quantum CNOT Gate with Actively Synchronized Photon Pairs
Haim Nakav, Tanim Firdoshi, Omri Davidson, Bankim Chandra Das, Ofer Firstenberg · 2025
Read Quantum CNOT Gate with Actively Synchronized Photon Pairs by Haim Nakav, Tanim Firdoshi, Omri Davidson, Bankim Chandra Das, Ofer Firstenberg (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-03960.txt.
Introduction
Haim Nakav and co-authors build a two-photon logic gate whose photon pairs are first parked in a memory so they arrive together, in a 2025 paper. Photons from random sources rarely meet on schedule, which strangles photonic quantum computing. Their memory is a warm atomic vapor that stores one photon until its partner is ready. The gate they realize is the CNOT, the workhorse entangling gate of quantum circuits.
Its truth table scored above eighty-five percent fidelity, and the gate produced all four Bell states with the expected inequality violations. The main drag on fidelity is the dimming of two-photon interference during storage. The team derives an exact relation between that interference visibility and gate fidelity for pure photons, valid for any interference-based photonic gate. The result doubles as a design rule for the whole approach.
Catalog record
| CREATOR | Haim Nakav, Tanim Firdoshi, Omri Davidson, Bankim Chandra Das, Ofer Firstenberg |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-03960
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Quantum CNOT Gate with Actively Synchronized Photon Pairs Haim Nakav, Tanim Firdoshi, Omri Davidson, Bankim Chandra Das, Ofer Firstenberg Abstract Controlling the synchronization of photons from probabilistic quantum sources plays a pivotal role in advancing efficient quantum information processing. We report the realization of a probabilistic entangling gate operating on actively synchronized photon pairs, using a quantum memory based on warm atomic vapor. We achieve a truth-table fidelity exceeding 85% and demonstrate Bell-inequality violation for all four Bell states. The reduction in visibility of the Hong-Ou-Mandel (HOM) interference introduced by the storage process is identified as a primary factor limiting the gate fidelity. We derive an exact quantitative relation between HOM visibility and gate fidelity for pure single photons, which applies to all photonic gates relying on interference, regardless of the physical platform. Single-photon sources based on atomic vapor are prominent platforms for photonic quantum computing and communication, due to their simplicity and scalability Glorieux et al. 2023 . These sources often rely on spontaneous multi-photon processes, generating high-fidelity single photons in well-defined modes Lee et al. 2016 ; Davidson et al. 2021 . Nevertheless, the stochastic nature of these sources inherently limits their ability to generate large photonic states by multiplexing, as the success probability decreases exponentially with the photon number. To this end, integrating quantum memories with photon sources is imperative Nunn et al. 2013 . In particular, atomic-vapor memories can store single photons from spontaneous parametric down-conversion (SPDC) or four-wave mixing (FWM) sources Michelberger et al. 2015 ; Kaczmarek et al. 2018 ; Buser et al. 2022 ; Davidson et al. 2023a . These memories thus enhance the generation of photon pairs, facilitating the efficient operation of two-photon gates. Photonic entangling gates are most commonly implemented using linear optical components within the Knill-Laflamme-Milburn (KLM) framework Knill et al. 2001 . Being probabilistic, these gates rely on post-selection but nevertheless are proven to be sufficient for scalable quantum computation Kiesel et al. 2005 ; Okamoto et al. 2005 ; Kok et al. 2007 . Linear-optics gates are based on two-photon interference – namely, the Hong-Ou-Mandel (HOM) interference effect. Their fidelity is governed by the HOM interference visibility. For a pair of pure single photons in identical spatial modes, the HOM visibility reflects their temporal indistinguishability Kambs and Becher 2018 ; Legero et al. 2003 . Maintaining indistinguishability becomes particularly challenging when interfacing photons from heterogeneous modules, such as a source and a memory. Figure 1: Experimental scheme for a quantum CNOT gate with actively synchronized photons. (a) Atomic level scheme of the photon source, stochastically generating pairs of signal and idler photons in two different channels through spontaneous four-wave mixing. (b) Atomic level scheme of the quantum memory. (c) A signal photon (s1) is stored when detecting the idler photon (i1) and retrieved (s1 ′ ) when detecting the second idler photon (i2). This synchronizes s1 ′ with the second signal photon (s2), which are sent to the CNOT setup. The input qubit states are encoded using polarization wave-plates (not shown). The gate operation relies on two-photon interference on partially-polarizing beam splitters (PPBS). The outgoing qubit states are measured via polarization tomography using single-photon detectors. (d) The temporal shapes of the synchronized photons show that the photons retrieved from the memory (blue) are temporally broader (full-width at half maximum, FWHM =1.5\penalty\ ns) and more symmetric compared to the photons produced by the source (purple, FWHM =0.9\penalty\ ns). In this work, we realize a KLM-based two-photon quantum gate executing a controlled-not (CNOT) operation in the polarization basis of actively synchronized photon pairs. For the gate input, we synchronize two single photons by combining a stochastic single-photon source Davidson et al. 2023b with a quantum memory Davidson et al. 2023c ; Davidson et al. 2023a . The source and memory use an identical atomic level scheme in warm vapor, ensuring mutual compatibility, maximizing memory performance, and preserving high photon indistinguishability. Our fast ladder memory (FLAME) Finkelstein et al. 2018 can store nanosecond-long single photons and has negligible noise (15 \times 10^{-5} noise photons per retrieved photon). The integration of FLAME with the photon source dramatically enhances the rate of synchronized photon pairs arriving at the quantum CNOT gate, which in turn leverages their high indistinguishability for producing high-fidelity Bell states. Our experiment realizes a key step toward scalable photonic quantum computing. Experimental setup.— The first part of the setup is responsible for generating synchronized photon pairs. It includes a dual-channel single-photon source and the FLAME memory, as described in Ref. Davidson et al. 2023a , both utilizing the ladder-type orbital-level scheme 5S_{1/2}\rightarrow 5P_{3/2}\rightarrow 5D_{5/2} of 87 Rb, shown in Fig. 1 (a,b). The source is driven by two pump lasers and stochastically generates signal and idler photons (at 780 nm and 776 nm, respectively) along phase-matched directions in two different channels. In both channels, the detection of an idler photon heralds the generation of a signal photon. The present source generates signal photons at an average rate of 40,000 counts/sec per channel, with heralding efficiency of 18% and 22.7% for channels 1 and 2, respectively, a multi-photon component of g^{(2)}_{\mathrm{h}}(0)=0.0129(1) , and HOM visibility of \mathcal{V}_{\mathrm{source}}=0.89(2) [ SM , Sec. S1]. A signal photon from one channel is sent to FLAME, where it is stored by overlapping it with a drive-field pulse. The present FLAME has an end-to-end efficiency of 25 \% , a 1/e lifetime of nearly 100 ns, and the retrieved photons exhibit g^{(2)}_{\mathrm{h}}(0)=0.028(5) [ SM , Sec. S2]. The retrieval of the stored photon using another drive pulse is synchronized with the generation of a signal photon in the second source channel. The two corresponding idler photons are used for triggering, as illustrated in Fig. 1 c. Figure 1 d presents the temporal profiles of the synchronized photons. Due to the finite bandwidth of FLAME, acting as a spectral filter, the retrieved photons are 65% longer and temporally more symmetric than those emitted from the source. The relative timing between the two photons, indicated in Fig. 1 d, is tuned to maximize their HOM visibility \mathcal{V} , reaching \mathcal{V}=0.79(2) [ SM , Sec. S3]. The synchronized photons are directed via fibers to the second part of the setup, comprising qubit state preparation, the CNOT gate, and measurement. We use the linear polarization states |\mathrm{H}\rangle\equiv|0\rangle and |\mathrm{V}\rangle\equiv|1\rangle as the CNOT’s basis, that is, the target qubit flips between |\mathrm{V}\rangle\leftrightarrow|\mathrm{H}\rangle when the control qubit is in |\mathrm{V}\rangle . The photon arriving from the memory acts as the control, while that arriving directly from the source serves as the target. We prepare the initial qubit states using waveplates. Our CNOT implementation, based on Ref. He et al. 2013 ; Zhai et al. 2022 , is shown in Fig. 1 c. The control and target photons impinge on a partially polarizing beam splitter (PPBS) having 1 and 1/3 transmission for the horizontal and vertical polarizations, respectively. The HOM interference on this PPBS generates non-classical correlations between the two qubits. Additional PPBSs along each path, oriented perpendicular to the first PPBS, balance the output amplitudes for all polarization configurations. Two half-wave plates flanking the PPBSs in the target qubit path are oriented at 22.5^{\circ} relative to the horizontal axis and act as