Exponential-SWAP gate between bosonic motional modes of trapped ions

AI summary

Overview

Research area: Quantum physics — trapped-ion quantum information processing, hybrid qubit–bosonic (continuous-variable) architectures.

Technical level: Intermediate. The core idea is explainable in plain terms, but full appreciation of the calibration and tomography methods requires familiarity with trapped-ion spin-motion control.

Scope: An experimental demonstration of an exponential-SWAP gate between two motional modes of two trapped ¹⁷¹Yb⁺ ions, characterized by selected slices of the joint characteristic function.

What This Paper Is About

Bosonic modes of trapped ions are attractive carriers of quantum information, but useful multimode processing requires operations that can coherently exchange and entangle their states in a tunable way. The exponential-SWAP (eSWAP) gate does exactly this: it superposes doing nothing (identity) with exchanging the two modes (SWAP), with the balance set by a rotation angle θ on an auxiliary spin. The authors build this gate between two motional modes of trapped ions for the first time and test whether it behaves as predicted for several distinct input states.

Key Contributions

1. First experimental realization of an eSWAP gate between bosonic motional modes of trapped ions, using three motional modes of two ¹⁷¹Yb⁺ ions in a linear Paul trap.
2. A composite gate construction that combines a spin-independent beam splitter on one computational mode and an auxiliary mode, spin-dependent beam splitters conditioned on the spin basis, and an exponential-parity gate eP(θ) built from two YBS gates around a spin rotation.
3. A scheme that moves unwanted motional state-dependent phases acquired during beam-splitter state exchange onto an empty auxiliary mode, making the resulting operation independent of the input state, with the laser phase Φ allowed to be arbitrary.
4. A characterization protocol based on measuring selected slices of the joint characteristic function rather than full two-mode tomography, which is enough to test the identity–SWAP superposition at a fraction of the measurement cost.

Main Findings

• Gate form verified for low Fock inputs: Preparing (1/√2)|0⟩(|0⟩+|1⟩) and applying eSWAP(−π/2) produced data consistent with the target state (1/2)[|0⟩(|0⟩+|1⟩) + i(|0⟩+|1⟩)|0⟩]. Four measured 40×40 imaginary slices gave an average overlap with the ideal prediction of 𝒪_exp = (63 ± 4)%.
• Simulation reproduces the contrast loss: A QuTiP Lindblad master-equation simulation, using independently measured heating and dephasing rates of all three modes, gave 𝒪_sim = (75 ± 3)%. Dephasing of the auxiliary radial out-of-phase mode, with a measured coherence time of approximately (10 ± 2) ms, was identified as the dominant contribution; both axial modes have coherence times of approximately 100 ms or longer.
• Heating is not the limiting factor: Heating of the in-phase axial mode at (17 ± 1) phonons/s had a negligible effect over the 4.2 ms sequence.
• Phase compensation works beyond the lowest Fock states: For the input |1,2⟩, the real slices matched the target (1/√2)(|1⟩|2⟩ + i|2⟩|1⟩) with 𝒪_exp = (78 ± 3)% versus 𝒪_sim = (81 ± 2)%, confirming the expected relative phase between the |1,2⟩ and |2,1⟩ components and compensation across odd and even occupations.
• Continuous amplitude control demonstrated: For the coherent-state input |−γ, γ⟩ with γ = 0.5, the measured characteristic-function values Im{χ(±2γ, ∓2γ)} followed the predicted ±0.5 sin θ dependence with 2π periodicity, agreeing with simulation. Simulations indicate auxiliary-mode dephasing is negligible for this signal, while heating of the axial in-phase mode reduces the oscillation amplitude to approximately 0.37.
• Identified residual error sources: State-detection errors were estimated to reduce the overlap by approximately 2%; imperfect motional-state initialization is cited as a further unmodeled imperfection.
• Discrepancy remains: The measured overlap is below the simulated one in both benchmark cases, which the authors attribute partly to imperfections not included in the simulations.

Methodology in Plain English

The experiment uses two ytterbium ions held in a linear Paul trap. Their collective motion (motional modes) serves as the bosonic system: two axial modes carry the computational states, and a radial out-of-phase mode acts as an auxiliary that starts empty (in its motional ground state). The spin of one ion is used as the control qubit; it is prepared in a superposition of two hyperfine ground states, and the second ion is parked in a metastable dark state so it does not scatter light.

Pulsed Raman laser beams do the work. One configuration of beams produces a beam splitter that moves excitations between a computational mode and the auxiliary mode regardless of the spin state; another configuration, using bi-chromatic beams driving second-order sidebands, produces a beam splitter that acts only for particular spin states. Sandwiching a spin rotation between two such conditional beam splitters creates an operation that multiplies the state by a parity-dependent phase, controlled by the spin rotation angle. Combining these elements in a symmetric sequence, with the auxiliary mode empty, yields the identity-minus-SWAP superposition.

To check the result, the authors measure pieces of the joint characteristic function of the two modes. Spin-dependent displacements are applied to both modes, and the spin expectation value is read out by fluorescence; choosing the analysis rotation angle selects the real or imaginary part. This gives enough information to test the gate's predicted behavior without the much longer measurement time needed for full two-mode tomography.

Why This Matters

Impact on research: The eSWAP gate is a tunable, coherent exchange between two bosonic modes, and this work shows it is available on a trapped-ion platform alongside superconducting circuits, where an eSWAP was previously realized. It extends the trapped-ion toolbox from state transfer and entanglement generation to a full identity–SWAP superposition with continuously adjustable amplitude, which is a building block for hybrid qubit-bosonic processors and for multimode simulation.

Real-world applications (as framed in the paper):

• Bosonic quantum error correction, which relies on controlled multimode operations.
• Gaussian boson sampling and multimode quantum simulation of spin-boson dynamics.
• Quantum machine-learning primitives, since eSWAP underpins density-matrix exponentiation and routines such as principal component analysis, matrix inversion, and vector-distance estimation.
• Universal quantum computation with arbitrary continuous-variable encodings, where eSWAP combined with SWAP tests has been proposed as a route.

Industry relevance: The work targets hardware-level primitives for hybrid quantum processors, and its error budget is diagnostic — the dominant limitation is auxiliary-mode dephasing and motional heating, both of which are engineering targets for trap design and laser control. The gate's construction is stated to be adaptable to other hybrid platforms, which broadens its relevance beyond trapped ions.

Future Directions

• Reduce dephasing of the auxiliary radial out-of-phase mode, whose approximately (10 ± 2) ms coherence time dominates the observed contrast loss, and reduce motional heating to close the gap between measured and simulated overlaps.
• Push characterization beyond selected slices toward full multimode tomography, which the authors note would require substantially longer measurement times.
• Explore the gate in larger multimode settings, since the construction relies on an auxiliary mode and could in principle be extended to networks of modes.
• Port the gate construction to other hybrid qubit-bosonic platforms, as the authors suggest, and test it within error-correction or machine-learning protocols where eSWAP is a required primitive.

Target Audience

Researchers and students in quantum information science working on trapped-ion control, continuous-variable and hybrid qubit-bosonic architectures, or bosonic error correction. It is also useful for readers who want a concrete experimental example of how parity-conditional operations and characteristic-function measurements are used to verify multimode bosonic gates.

Authors’ abstract

The exponential-SWAP (eSWAP) gate generates a coherent superposition of identity and SWAP operations, with continuously tunable amplitudes. We experimentally implement this gate between two bosonic motional modes of trapped ions using spin-independent and spin-dependent beam splitter interactions. Measurements of selected slices of the joint characteristic function for distinct input states test the gate and demonstrate continuous control of the identity-SWAP superposition. This implementation provides a tunable interaction for hybrid qubit-bosonic quantum information processing.