A recent study has sparked debate over which type of quantum computer is better at solving complex optimization problems: gate-model quantum computers like IBM’s or analog quantum annealers like D-Wave Quantum Annealers. Researchers compared the performance of these two types of quantum computers on a specific problem known as a planar spin glass. They found that D-Wave’s Advantage2 prototype, which uses analog quantum annealing, outperformed IBM’s gate-model quantum computer regarding solution quality and speed.
The study’s authors, criticized previous claims that gate-model quantum computers were superior because they didn’t take into account the cost of producing a sample. They also tested the effect of doubling the energy scale on D-Wave’s processor, which led to significantly better performance.
Companies involved in this work include D-Wave Systems and IBM, two leading players in the quantum computing industry. The study sheds light on the strengths and weaknesses of different approaches to quantum computing, an area that has the potential to revolutionize fields like medicine, finance, and climate modeling.
The authors are comparing the performance of two quantum computing modalities: gate-model quantum processors (like IBM’s) and analog quantum annealing processors (like D-Wave’s). They’re focusing on optimization problems, specifically spin-glass models, which are notoriously difficult to solve.
Here are the key takeaways:
- Classical postprocessing is crucial for good samples with Q-CTRL: The authors show that classical postprocessing significantly improves the quality of samples obtained from Q-CTRL’s hybrid variational algorithm running on IBM quantum processors. In contrast, postprocessing has a much smaller impact on the quality of samples from D-Wave’s analog quantum annealing processors.
Analog quantum annealing outperforms digitized QA: When comparing the performance of D-Wave’s Advantage system and Advantage2 prototype with IBM’s Torino system on a 133-qubit heavy-hex spin-glass optimization problem, the authors find that analog quantum annealing achieves lower residual energies, especially with the Advantage2 prototype. This is due to its higher energy scale and lower noise.
Doubling the energy scale improves performance: By modifying the input construction of Pelofske et al. and replacing the third gadget with a spin-reversal transformation of the first gadget, the authors demonstrate that doubling the QA energy scale significantly reduces the need for postprocessing and improves performance.
Digitized QA is not competitive with analog QA: The comparison on a planar spin-glass instance shows that digitized quantum annealing running on IBM quantum processors does not match the solution quality achieved by analog quantum annealing running on D-Wave processors.
The authors conclude that their experiments demonstrate the superiority of analog quantum annealing over gate-model quantum processors for optimization problems, at least in this specific context. They also highlight the importance of considering the cost of producing a sample and accounting for postprocessing when evaluating the performance of different quantum computing modalities.
External Link: Click Here For More
