The (Quantum) Signal and the Noise: towards the intermediate term of quantum computation
The (Quantum) Signal and the Noise: towards the intermediate term of quantum computation
Can we compute on small quantum processors? In this talk, I explore the extent to which noise presents a barrier to this goal by quickly drowning out the information in a quantum computation. Noise is a tough adversary: we show that a large class of error mitigation algorithms -- proposals to "undo" the effects of quantum noise through mostly classical post-processing – can never scale up. Switching gears, we next explore the effects of non-unital noise, a physically natural (yet analytically difficult) class of noise that includes amplitude-damping and photon loss. We show that it creates effectively shallow circuits, in the process displaying the strongest known bound on average convergence of quantum states under such noise. Concluding with the computational complexity of learning the outputs of small quantum processors, I will set out a program for wrapping these lower bounds into new directions to look for near-term quantum computational advantage.