Out of Tune: Demystifying Noise-Effects on Quantum Fourier Models (2506.09527v1)

Abstract: The field of variational quantum algorithms, in particular quantum machine learning (QML), produced numerous theoretical and empirical insights in recent years. As variational quantum circuits (VQCs) can be represented by Fourier series that contain an exponentially large spectrum in the number of input features, hope for quantum advantage remains. Nevertheless, properties of quantum Fourier models (QFMs) are not yet fully understood, in particular how they could potentially outperform classical alternatives. Viewing VQCs with Fourier lenses opens up possibilities to analyse which classes of functions can be tackled by variational algorithms such as QML, while also illuminating and quantifying remaining constraints and challenges. Considering that noise and imperfections remain dominant factors in the development trajectory from noisy intermediate-scale to fault-tolerant quantum computers, the aim of this work is to shed light on key properties of QFMs when exposed to noise. In particular, we systematically analyse the effect of noise on the Fourier spectrum, expressibility and entangling capability of QFMs by conducting large-scale numerical simulations of quantum systems. This may help to better utilise hardware resources, and guide the construction of tailored error correction schemes. We find that decoherent noise exerts a uniform deleterious effect on all the tested ans\"atze, manifesting in the vanishing of Fourier coefficients, expressibility and entangling capability. We note however, that the detrimental influence of noise is less pronounced in some ans\"atze than in others, suggesting that these might possess greater resilience to noise.