Unit Fidelity Entangling Gates Achieved Via Continuous Dynamical Decoupling and Optimal Control

Key Takeaways

• Continuous dynamical decoupling suppresses low‑frequency noise.
• Variational geodesic optimisation yields near‑unit gate fidelity.
• CZ and CX gates achieved with realistic control fields.
• Method avoids neural networks, simplifies pulse design.
• Framework extensible to higher‑dimensional SU(9) control.

Summary

Researchers at the São Carlos Institute of Physics have combined continuous dynamical decoupling (CDD) with variational minimal‑energy optimal control to create two‑qubit entangling gates with virtually unit fidelity. The unified scheme actively suppresses low‑frequency flux noise, calibration drift, and spurious couplings, stabilising the effective Hamiltonian for tunable CZ and CX operations. By employing a variational geodesic optimisation that solves a nonlinear Schrödinger equation, the team eliminates the need for complex pulse‑shaping or neural‑network methods. Simulations on superconducting transmons show robust, high‑fidelity performance using experimentally realistic control fields.

Pulse Analysis

The quest for high‑fidelity quantum gates remains the linchpin of superconducting quantum computing. Conventional approaches typically layer error‑mitigation techniques—dynamic decoupling, calibration routines, and pulse‑shaping—each addressing a separate noise source. By integrating continuous dynamical decoupling with a variational minimal‑energy control framework, the new method creates a stable effective Hamiltonian that neutralises low‑frequency flux fluctuations and static Hamiltonian errors before the gate operation even begins. This pre‑emptive noise suppression reduces the error budget dramatically, allowing the subsequent control stage to focus solely on driving the desired entangling interaction.

At the heart of the technique lies a geodesic optimisation on the SU(4) manifold. Researchers define a single parametric matrix Λ(0) that governs the time‑dependent control fields via a nonlinear Schrödinger equation, then propagate a Lagrange‑multiplier matrix Γ(t) backward to solve a boundary‑value problem. This second‑variation principle sidesteps the computational overhead of neural‑network‑based pulse synthesis while delivering smooth, low‑energy single‑qubit pulses. Simulations on transmon qubits—operating at 5 GHz with an engineered ZZ coupling of 82.5 MHz—demonstrate CZ and CX gates with fidelity approaching 99.999%, even under realistic crosstalk and coherent noise conditions.

The broader impact is twofold. First, achieving virtually unit fidelity with a single optimisation loop simplifies hardware calibration pipelines, shortening the time from prototype to production‑grade quantum processors. Second, the variational framework is inherently extensible; the authors plan to expand from SU(4) to SU(9) to incorporate leakage suppression for faster, higher‑dimensional gates. As the industry pushes beyond the noisy intermediate‑scale quantum era, such unified, noise‑resilient control strategies are poised to become foundational tools for building scalable, fault‑tolerant quantum computers.