Can the solution energy of the actually-solved (as opposed to the submitted) Hamiltonian be returned?
Integrated control errors such as flux noise, next-nearest-neighbor couplings, etc. result in the problem solved being different than the problem submitted to quantum annealing machines. My understanding is that though this is the case, the returned energy, corresponding to the solution found, is based on the submitted problem. (That is, the solution found is plugged into the submitted Hamiltonian to yield the energy of the solution.)
My intuition is that it is impossible to return the energy of the actually-solved problem, as one would have to know each of the biases and couplings exactly at the moment of solution, these biases and couplings being the sum of those applied and those due to noise. Is this true?
Comments
Although theoretically possible, but very difficult to do with high accuracy, since errors are dependent on time, the programmed Hamiltonian, and the annealing protocol.
This paper might be of interest to you:
https://advances.sciencemag.org/content/4/3/e1700791
Please sign in to leave a comment.