Title:Exploring experimental limit of deep quantum signal processing using a trapped-ion simulator

Abstract:Quantum signal processing (QSP), which enables systematic polynomial transformations on quantum data through sequences of qubit rotations, has emerged as a fundamental building block for quantum algorithms and data re-uploading quantum neural networks. While recent experiments have demonstrated the feasibility of shallow QSP circuits, the inherent limitations in scaling QSP to achieve complex transformations on quantum hardware remain an open and critical question. Here we report the first experimental realization of deep QSP circuits in a trapped-ion quantum simulator. By manipulating the qubit encoded in a trapped $^{43}\textrm{Ca}^{+}$ ion, we demonstrate high-precision simulation of some prominent functions used in quantum algorithms and machine learning, with circuit depths ranging from 15 to 360 layers and implementation time significantly longer than coherence time of the qubit. Our results reveal a crucial trade-off between the precision of function simulation and the concomitant accumulation of hardware noise, highlighting the importance of striking a balance between circuit depth and accuracy in practical QSP implementation. This work addresses a key gap in understanding the scalability and limitations of QSP-based algorithms on quantum hardware, providing valuable insights for developing quantum algorithms as well as practically realizing quantum singular value transformation and data re-uploading quantum machine learning models.