I’ve been running small test cases on both lately and finally got a basic traveling salesman problem working on D-Wave’s annealer last night. Once I mapped it to the qubits it found a decent route fast, but the results still came across as more of a good guess than an exact answer, especially once the hardware noise showed up. IBM’s gate approach took way longer to build the circuit, yet the steps stayed clean enough that I could tweak them when something drifted. Not sure if this is just how the two solvers behave or if I’m missing something with the embedding on D-Wave. Is there a standard way to check embedding quality on these TSP mappings?