Quantum Annealing performing worse than Simulated Annealing
Hello,
I am researching an image processing task using quantum annealing. Currently, I am running experiments with 128x128 pixel images with grayscale pixel intensities within the range [0-7] instead of the standard [0-255] to simplify the problem. In the problem, I have set up a Random Markov Field to find the optimal pixel values according to some criterion.
I observed that using the CQM that I have set up, simulated annealing gives me solutions with lower energy than quantum annealing. This surprised me because I believed quantum annealing performs at least as well as simulated annealing. Is this not true then? Is there any proof or research regarding this?
In practice, what should one do when quantum annealing results are poorer than simulations? Are there any changes to the problem formulation known to increase/decrease the performance of quantum annealing?
Thanks in advance!
Comments
Hello,
Thank you for reaching out to us.
Some problems will perform better on a purely classical system, depending on the nature of the problem.
Please take a look at this paper for a discussion of quantum annealing vs classical methods:
https://arxiv.org/abs/2305.00883
Please sign in to leave a comment.