Boffins make noise about D-Wave chip: it seems quantum
Thermal 'knob' turns up the heat
Researchers from University College, London, and the University of Southern California, have weighed into the ongoing “is it quantum?” D-Wave debate with an interesting approach, testing the device under a variety of noise conditions.
As their paper at Arxiv explains, the thermal environment of a D-Wave chip isn't directly accessible: the machine operates as a “black box”, in that respect. However an energy model is part of how problems are coded for the computer as a whole – and that gave the researchers, led by USC's Daniel Lidar, a “knob” they were able to adjust in their tests.
The control knob the researchers accessed is that the behaviour of the D-Wave device has “a controllable overall energy scale, acting as an effective (inverse temperature) 'noise control knob.'” Reducing the energy scale “amounts to increasing thermal excitations” during the computation.
Why would this matter? The D-Wave chip is chilled to 20 millikelvin to prevent thermal noise from overwhelming the quantum effects the company says are the basis of its computations. Therefore, the UCL / UC researchers reasoned, it should be possible from the input-output behaviour of the device to predict the degree to which the chip's “quantumness” varied at different energy scales.
That, they say, is exactly what they observed. As they write:
“At the largest energy scale available, the annealing process appears to be dominated by coherent quantum effects, and thermal fluctuations are negligible. As the energys cale is decreased, thermal excitations become more relevant, and for a sufficiently small energy scale, the system behaves more like a classical annealer based on incoherent Ising spins.”
For this research, Lidar's group tested groups of 40 qubits against three classical models, and one quantum model: “The classical models are all found to disagree with the data, while the master equation agrees with the experiment without fine-tuning, and predicts mixed state entanglement at intermediate evolution times”.
Is this the end of the debate? Of course not: it's not even a final proof that D-Wave is quantum, inside the black box.
However, Scott Aaronson – let's call him “semi-retired D-Wave sceptic-in-chief – told El Reg this experiment does represent another addition to our knowledge of what's going on.
“I think the two sides are slowly converging on a real physical understanding of the current D-Wave devices – in particular, under what circumstances the devices can produce 'signatures' of various kinds of quantum behaviour and under what circumstances those signatures go away,” Aaronson told The Register in an e-mail.
He added that evidence for quantum behaviour still doesn't demonstrate that D-Wave is “faster” than classical computing even on its home turf. While “clear evidence of global quantum behaviour” is a prerequisite of ultimately achieving a quantum speed-up in computing, that doesn't yet guarantee that the speed up will ever be achieved.
“You can have global quantum behaviour without a quantum speedup, but you can't have a quantum speedup without global quantum behaviour,” Aaronson said, also noting that observing quantum-like behaviour in special instances doesn't predict the scaling behaviour of the D-Wave device. ®