Curator's Take
AI Commentary
This article shows the first systematic way to program arbitrary anharmonic potentials on a superconducting harmonic oscillator, turning a simple transmon‑coupled cavity into a versatile analog quantum simulator. By leveraging bosonic quantum signal processing to synthesize non‑Gaussian phase gates such as cubic, double‑well and Morse‑type operations, the work bridges a long‑standing gap between continuous‑variable hardware and universal quantum computation. The ability to generate high‑fidelity non‑Gaussian states on existing superconducting platforms opens immediate pathways for simulating molecular vibrations, tunnelling phenomena, and other nonlinear dynamics that were previously out of reach. While the demonstrations are still limited to modest gate depths, the modular control scheme promises scalable extensions as coherence times improve.
— Mark Eatherly
Summary
Continuous-variable quantum systems offer a resource-efficient route to universal quantum information processing and analogue quantum simulation of real-world processes, such as molecular physics and chemical reactions. Realising these applications, however, requires non-Gaussian operations that implement anharmonic potentials, which are challenging to engineer on demand. Here, we demonstrate a systematic framework to implement programmable non-Gaussian phase gates $e^{-iV(\hat{X})}$, corresponding to the impulsive action of a potential $V(\hat{X})$, in a superconducting harmonic oscillator coupled to a transmon qubit. Using modular circuits derived from bosonic quantum signal processing, we realise a range of target anharmonic potentials on a single piece of hardware by varying a set of qubit rotations interleaved with a fixed calibrated control unitary. We first demonstrate a cubic phase gate, a key ingredient for universal quantum information processing. The resulting high-fidelity non-Gaussian states and the potential reconstructed using our pointwise force reconstruction method jointly confirm the cubic nature of the target gate. We then engineer a family of double-well potentials, relevant models of tunnelling and biased transfer processes, and experimentally validate the double-well topology and the tunable asymmetry. Finally, we engineer an approximate Morse gate, a step towards realistic potentials of molecular vibrational systems, and provide a concrete path towards high-quality engineering and reconstruction of the exponential form. Together, these results establish a practical and reconfigurable route towards continuous-variable quantum information processing and anharmonic quantum simulation.