arXiv:2610.28749
Bias-Preserving Entanglement of Kerr-Cat Qubits with an ATS Coupler
Kerem Yurtseven, Shantanu R. Jha, Apoorva Limaye, Risav Pokhrel, Maral Mahmoudi Kamelabad, Muhammad Waqar Amin
Abstract
Kerr-cat qubits offer a hardware-efficient route to fault-tolerant quantum computing by suppressing bit-flip errors exponentially, while phase-flip errors increase only linearly with the mean photon number. To maintain this protection, entangling operations must be bias-preserving, ensuring that phase-flip errors occurring during the gate do not propagate into bit-flip errors. In this paper, we propose an asymmetrically threaded SQUID (ATS) parametric coupler to implement high-fidelity, bias-preserving entangling gates between Kerr-cat qubits. By engineering a photon-number asymmetry between the control and target modes, we enable fast gate operations that suppress non-adiabatic leakage without complex pulse shaping. Through numerical simulations of the full circuit Hamiltonian, we demonstrate a bias-preserving CX gate achieving 99.69% fidelity in the absence of photon loss for a gate duration of 50 ns. Under a single-photon loss rate of 100 kHz, our protocol yields a noise bias of 9922, exceeding the noise bias of a naive non-bias-preserving CX implementation by approximately a factor of 570. These results support the ATS coupler as a valuable building block for concatenated cat-qubit architectures.
AI Summary
The paper addresses how to implement a bias-preserving CX between Kerr-cat qubits at the circuit level, so photon-loss-induced phase errors do not become logical bit flips. It proposes an asymmetrically threaded SQUID (ATS) coupling two cat resonators. Two flux-pump tones activate the three-wave-mixing and conditional-displacement interactions required by an existing bias-preserving CX Hamiltonian, while Kerr confinement remains active. An effective-model analysis identifies control-mode nonadiabatic excitation governed approximately by $\xi=\alpha^2/(\beta^3KT)$, motivating a larger control cat than target cat. Simulations of the full nonlinear coupler Hamiltonian, supplemented by effective Kerr-cat stabilization terms and compensating drives, use $\alpha^2=2$, $\beta^2=10$, and $K/2\pi=3$ MHz. Logical process tomography gives 99.69% average fidelity without photon loss for a 50 ns gate. At $\kappa_{\rm 1ph}/2\pi=100$ kHz, the phase-to-nonphase Pauli-error-weight ratio is approximately 9922, versus 17.44 for a naive logical CX; fidelity is 77.53%, close to equivalent idling at 77.72%. At 0.75 kHz loss, fidelity is 99.48% and bias approximately 1150. The results are numerical, not experimental. Residual leakage, coherent errors, strong-drive-induced displacements and coupler excitation remain limitations; intrinsic coupler loss is neglected. A benchmark with 1% junction asymmetry and second-harmonic contributions lowers lossless fidelity to 99.38% and bias to approximately 6924. The asymmetric cat sizes make fast operation directional. Frequency crowding, flux crosstalk, integration of the cat-stabilization circuitry, and performance within syndrome-extraction cycles remain to be established.
Error found: The abstract reports noise bias 9922 at a single-photon loss rate of 100 kHz, but the supporting simulations use a Lindblad loss coefficient κ=2π×100 kHz≈628,319 s⁻¹, so that numerical bias is not established at κ=100,000 s⁻¹.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 23 pages, 12 figures