Energy evaluation complexity
Hi, I have an idea for two versions of an algorithm for Quantum Computer, one of which uses much less QUBO variables, but the evaluation of solution's energy has a higher complexity. How is this complexity affecting the behaviour and complexity of the whole annealing process? I'll be grateful for any tips in that matter :)
Comments
Hi Marek,
I'm interested if this has to do with your previous question on the JSP as I would be very interested in a formulation that uses fewer variables. As far as I know, the D-Wave JSP demo uses the time-indexed JSP formulation and it is already pruned to use fewer variables.
Could you explain what you mean by "evaluation of solution's energy has a higher complexity"? I think evaluating the energy is just simply plugging the solution variables into the BQM. Being that the BQM is just linear and quadratic terms with coefficients, it should be computationally not complex.
I think one important aspect of the BQM/formulation in terms of the annealing process would be the size of the energy gap between the ground state and the next state (smaller = more difficult). https://docs.dwavesys.com/docs/latest/c_gs_2.html#annealing-in-low-energy-states gives a good visual and description. If you are using dwavebinarycsp.stitch, you can try using the min_classical_gap parameter. Outside of CSPs, I'm not really sure.
I've spent some time making sure about it, but it turns out I misunderstood a procedure of aquiring the final energy.
My concept was to use an operation-order-indexed JSP formulation, but now I can't see it happening. Anyway, thank You for your support, I'll try to dig further ;)
Please sign in to leave a comment.