On prospects and limitations of variational quantum algorithms

Variational quantum algorithms have become the de-facto model for the current era of noisy quantum computers. These algorithms utilize a short depth parametrized quantum circuit, which fits within the coherence time of a given device. The parameters of a variational circuit are tuned iteratively by a classical co-processor in an attempt to optimize a certain cost function, which ensures preparation of a desired quantum state. Variational algorithms indeed are partially resilient to some limitations of current noisy devices. Still, their full potential and limitations remain an active research area. In this talk I will cover multiple aspects of variational algorithms, which allow us to quantify their limitations and propose strategies to assist their implementation. The talk will introduce and discuss the following ideas: the structure of optimal parameters and onset of training saturation in the quantum approximate optimization algorithm, the effect of parameter miscalibration on variational quantum eigensolvers, a hardware inspired zero noise extrapolation protocol, and a hardware native implementation of variational algorithms.