Limits of Simulated Annealing Sampler
I am experimenting with the D-Wave quantum annealer and benchmarking it against the classical solution in terms of runtime. I was also benchmarking the DWave sampler against the Neal SimulatedAnnealingSampler and noticed that its runtimes seem to be constant as a function of the problem size and the results seem to be mostly correct.
So I am wondering what the limits of the simolator are. Does it work as a reliable source of results only up to a certain problem size? Does it have a memory requirement that grows exponentially as a function of the problem size?