by David Shaw, GQI
IBM has announced exciting new results that provide significant support to those who believe relatively noisy medium-scale quantum devices (NISQ) in the short term could yield useful commercial results. This news will excite the quantum computing community. It also promises distraction for some.
IBM has used their Eagle R3 processor for demonstrate effective execution of a 127Q circuit to a gate depth of 60. Published in Nature, the results show how quantum devices outperform the best classical alternatives currently available. In addition, it is performed in the type of problem (2D cross-plane Ising model time evolution) that will impress many scientists as a guide for future materials science applications. IBM does not claim that this is quantum supremacy or quantum superiority in terms often discussed before. As well as bypassing debates about possible future improvements in conventional algorithms or definitions of exponential quantum advantage, he has chosen to demonstrate impressive practical skills in problems that exhibit real-world applications.
While the quantum community will welcome this new success, GQI believes it is also a potential source of disruption within the sector.
100×100 will set a new table stake
IBM’s results are a strong indicator that it is on track to deliver the 100×100 challenge. This promises, by the end of 2024, the ability to consistently calculate the expected value of a 100Q by 100 gate depth circuit in a 24 hour timeframe. IBM is already at 127×60 and the Eagle R3 hasn’t even benefited from the increased 2Q gate fidelity we’ve come to expect from the move to a tunable coupler architecture (pioneered in the Falcon R10/Egret). we look forward to seeing his Heron kit.
100×100 isn’t a slam dunk for any given application, but it’s definitely going to be a very interesting ‘go beyond classic’ capability. We have experienced an era of quiescence in quantum computing, where many different qubit vendors competed to secure early hardware implementations around the world. Most of this has supported research interests that want to experiment on real devices, but recognize that development of real code may also continue on simulators. The successful delivery of the 100×100 challenge promises to change that. Labs and facilities that can’t offer their developers access to these new capabilities will feel increasingly far from superior.
Algorithm developers need to recalibrate
IBM’s new route to useful NISQ-era applications could also irritate algorithm developers. The 100×100 challenge isn’t just about increasing hardware fidelity. This will require sophisticated error suppression. It also uses aggressive error mitigation. Techniques such as zero-noise extrapolation and probabilistic error cancellation rely on a runtime regime that automatically switches at very high numbers of shots (hence the 24-hour clock time window).
GQI suspects that it won’t be as easy as taking existing VQA concepts and just applying them to new capabilities. Software players who focus on error suppression and error mitigation will do well as others playing catchup. Quality will come out as others try to adapt to the new opportunities presented.
Significant NISQ earnings will mean stress for those who don’t
IBM has made major progress, but has not directly paved its long-term path to fault-tolerant quantum computing (FTQC). The architecture still faces the challenge of demonstrating increased performance at the required thermal and interconnect budgets. However, even competitors focused on their own long-term roadmaps toward fault tolerance will feel the pressure from this announcement. Many like the view that commercial applications in the NISQ era are increasingly impossible. IBM hasn’t proven that this will be achieved, but they have significantly moved the dial back in that direction. Players without the prospect of access to those revenues will inevitably feel more pressure than those who do.
For additional information about this technical demonstration, see our related article which explains it in more detail.
June 14, 2023