By Andre Saraiva, UNSW
There is a lot of excitement over the recent progress in ion-trap-based quantum computers. The technical developments demonstrated by Honeywell and IonQ are leading to significant action from the business side (for example, Honeywell announced the launch of a quantum-as-a-service project, while IonQ unveiled a quantum data center). These culminated on two announcements last month: both companies have announced they have the most powerful quantum computer in the world (see here and here). It is a hard task to be the referee between these two contenders, even though they use the same qubit technology.
The first thing that needs clarification is the scoring system. Both companies used the metric proposed by IBM in 2017, called “quantum volume”. Different quantum computers suffer from different noise correlations, qubit crosstalk and have different qubit connectivities. The principle of the quantum volume is to run a benchmark and verify whether the processor achieves complex multiqubit entangled states. Here, both the number of qubits and the quality of the entangling operations are rewarded.
Honeywell announced that they have measured a quantum volume of 128 in their H1 processor, which consists of 10 qubits. IonQ, on the other hand, announced that they have tweaked their qubit control in a way that would allow to reach a quantum volume in excess of 4 million with only 32 qubits. At a glance, it would seem that IonQ has by far the advantage, but two things need to be understood. Firstly, that the number quoted by IonQ is an estimate assuming that their new control scheme will lead to the expected positive impact on the multiqubit overall operation – a final number can only be set by actually running the benchmark, similarly to what Honeywell did. The second and most relevant question, as acknowledge by IonQ’s head scientist Christopher Monroe, is that the quantum volume is not synonymous with a good quantum computer.
Some quantum scientists believe that useful quantum computations will only be performed once an error-corrected processor is available. This means that these small-scale demonstrations of high-quality qubits need to be accompanied by good methods for error detection/correction, as well as a clear pathway for scaling up the number of qubits. Both companies know this is true and have been exploring different pathways for this task. This is perhaps the largest technical difference between the approaches of the two companies.
Ion traps have a limitation related to the number of qubits that can fit the same trap and be addressed individually by the control lasers. A few qubits in the same trap can be addressed by frequency modulation, but the next step for both companies is to connect two traps to ramp up the total number of qubits. Since these traps are very far apart, no direct interaction is possible. So, it is necessary to create a “flying qubit”.
Honeywell is exploring what they call the “Quantum Charge Coupled Device” approach – in reference to the classical CCD technology used in modern cameras. The idea is to reconfigure the trap in a way that one or a few ions can physically move between traps in a superposition state, communicating the quantum information between groups of data qubits. In this case, the ions themselves are the flying qubits. This reconfigurable trap scheme is in its infancy, and significant challenges must be overcome in order to coherently move ions at large enough scales.
IonQ bets on the conversion between ion qubits and photons, which can be routed between adjacent traps through fibers and waveguides. The advantage of this approach is that distance is not an issue – photons can travel long distances coherently (even thousands of kilometers). The downside is that the frequencies of light that carry well in fibers and solid state systems are very specific, and don’t match the typical transition frequencies of the Ytterbium atoms used by IonQ. For that purpose, it would be necessary to introduce a different element (Barium was mentioned recently in a research paper by them) that generates light of the correct wavelength. On top of that, interspecies high-fidelity entangling gates would become necessary. All these challenges are being tackled one-by-one, but the technology is not yet ready to deploy. Ultimately, it is likely that a combination of both of these techniques will be the path forward when many modules are integrated together, as acknowledged by Christopher Monroe in his talk at the IEEE Quantum Week conference on October this year.
So, what’s next?
Before IonQ completely benchmarks its new quantum processor, it will be hard to provide a full comparison with Honeywell’s H1. While the claim of a quantum volume surpassing 4 million may or may not pan out, it is to be expected that it will have an attention-grabbing performance. But in the world of exponential quantum advantage, the jump from a quantum volume of 64 to the speculative number of 4 million can mean just a handful of good quality qubits. The true divider will be the integration of multiple traps, which will probably give the championship belt to one of these two companies.
Dr. Saraiva has worked for over a decade providing theoretical solutions to problems in silicon spin quantum computation, as well as other quantum technologies. He works on MOS Silicon Quantum Dot research and commercially-oriented projects at the University of New South Wales (UNSW).
November 12, 2020