Bad Qubits — Learn Quantum by Doing

Multiple choice

Why is SPSA (Simultaneous Perturbation Stochastic Approximation) a default optimizer for hardware VQE under noise?

It computes the exact analytic gradient of every parameter with zero shot overhead
It estimates the whole gradient from just two energy evaluations using a single random perturbation direction, regardless of the number of parameters, and is resilient to shot noise
It requires no cost-function evaluations at all, relying only on the circuit structure
It needs one energy evaluation per parameter, which makes it cheaper than parameter-shift for large circuits