@dabacon but also @nick_c_rubin thanks for engaging, I am enjoying the tweeting over multiple drinks. One argument you cite is the mitigation of systematic errors and using this method as a subroutine. I am struggling to see how this can be ported to the algorithms cited here. https://twitter.com/dabacon/status/1299409675536932865

Systematic errors are heavily state-dependent: the optimal variational angles for one state are not the same for another state. Both in Hamiltonian simulation and VQE measurements the underlying quantum state is unknown and certainly not an HF state.

So I don't see how you can benefit from variational systematic error mitigation for these applications, using gate angles optimization on a HF state as a subroutine. One may be better off just classically compiling the Gaussian gates.

To be clear, I think this is nice work and agree with @jaygambetta: I see this result as a wonderful benchmark of your hardware, but fail to see where it fits as a step towards quantum advantage.