Quantum Computer Does Not Need Coherent Quantum Access for Advantage

Abstract: Demonstrating quantum advantage has been a pressing challenge in the field. A majority of claimed quantum speedups rely on a subroutine in which classical information can be accessed in a coherent quantum manner, which imposes a crucial constraint on the practicability of these quantum algorithms. It has even been shown that without such an access, the quantum computer cannot be (exponentially) stronger than the classical counterpart for many problems. Thus, whether or not a quantum computer can be useful for practical applications has been central to investigation. In this work, we develop a quantum gradient descent algorithm for polynomial optimization, which is a fundamental technique that enjoys a wide range of applications. The algorithm's running time is logarithmical relative to the number of variables and consumes logarithmic numbers of qubits. Our method can work with many classes of polynomials and, most importantly, it removes the requirement for such a coherent quantum access. Thus, it implies great potential for near-term realization and particularly surpasses the belief that quantum speedup is only possible with strong input assumptions. As immediate applications, we show how to translate the capability of performing gradient descent into solving a linear system, performing least-square fitting, building a support vector machine, performing supervised cluster assignment, training neural network, and solving for ground-state/excited-state energy, performing principle component analysis, with end-to-end applications of quantum algorithms. Thus, our result adds another prominent example to some existing literature, provably demonstrating the quantum advantage toward highly practical problems, suggesting a new and promising prospect for the real-world application of a quantum computer.