Variational quantum circuits learn control fields for nonlinear Schrödinger dynamics

Variational quantum circuits have found a new role as learning architectures for controlling quantum dynamics, according to a study published in Quantum Information Processing. The work combines a mathematical analysis of the control problem with numerical experiments that compare the circuit approach against established classical parameterizations.

The core of the framework is a parameterized quantum circuit with time reuploading, which generates two distinct control fields. These fields modulate the displacement and confinement terms of the nonlinear Schrödinger equation, allowing the hybrid quantum-classical optimizer to steer the wave dynamics toward a target state. The authors also established theoretical guarantees: they proved global well-posedness and the existence of optimal controls in a weighted energy space, using an energy identity that requires no time differentiability of the controls. This means the control fields can be quite irregular in time and still admit a well-defined optimal solution.

The analysis also characterizes exactly what the circuit can express. The realizable controls turn out to be trigonometric polynomials whose frequencies are drawn from a finite set, and this frequency set grows linearly with the circuit depth. However, the family is not dense in the space of continuous functions, because the readout imposes a uniform bound on the amplitude of the controls. This limitation is an honest constraint on the expressivity of the approach, and the authors report it alongside the positive results rather than claiming universal approximation.

The numerical evidence comes from experiments run over twenty restarts, with all compared methods matched in parameter count, amplitude bounds, budget, and initial occupancy. In these tests, the quadratic channel raised the terminal fidelity from 0.833 to 0.999. When compared with equal-size Fourier, chopped-random-basis, and spline parameterizations, the circuit achieved comparable fidelity while using approximately half the control energy. The study is a computational and theoretical contribution, so the reported fidelities and energy savings are numerical results rather than demonstrations on hardware, and the framework’s performance on actual quantum devices remains to be tested. Still, the combination of provable well-posedness, a transparent description of the expressible control class, and competitive numerical performance makes this a notable step in applying variational quantum circuits to the control of nonlinear wave dynamics.

Subject of Research: A dual-channel variational quantum framework for bilinear optimal control of the nonlinear Schrödinger equation

Article Title: Variational quantum circuit learning for bilinear control of nonlinear Schrödinger dynamics

Article References: Vijayaram, S., & Balasubramaniam, P. (2026). Variational quantum circuit learning for bilinear control of nonlinear Schrödinger dynamics. Quantum Information Processing, 25(11), Article 354. https://doi.org/10.1007/s11128-026-05367-y

Image Credits: AI Generated

DOI: 10.1007/s11128-026-05367-y

Keywords: variational quantum circuits, quantum control, nonlinear Schrödinger equation, bilinear optimal control, hybrid quantum-classical optimization, terminal fidelity, Fourier parameterization, chopped-random-basis, spline parameterization, well-posedness, control energy, quantum computing