P. Singkanipa, D.A. Lidar (Feb 15 2024).
Abstract: Variational quantum algorithms (VQAs) hold much promise but face the challenge of exponentially small gradients. Unmitigated, this barren plateau (BP) phenomenon leads to an exponential training overhead for VQAs. Perhaps the most pernicious are noise-induced barren plateaus (NIBPs), a type of unavoidable BP arising from open system effects, which have so far been shown to exist for unital noise channels. Here, we generalize the study of NIBPs to arbitrary, completely positive, trace-preserving maps, establishing the existence of NIBPs in both the unital and non-unital cases and tightening to logarithmic earlier bounds on the circuit depth at which an NIBP appears. We identify the associated phenomenon of noise-induced fixed points (NIFP) of the VQA cost function and prove its existence for both unital and non-unital noise. Along the way, we extend the parameter shift rule of VQAs to the noisy setting. We provide rigorous bounds in terms of the relevant parameters that give rise to NIBPs and NIFPs, along with numerical simulations of the depolarizing and amplitude-damping channels that illustrate our analytical results.