In the world of computing, three months can be a short time. For those watching the progress of quantum computing, waiting for the next 12 weeks will test their patience.
But that’s how long they’ll have to wait until Honeywell proves what it says is a breakthrough in the still-early stage of the race to create a practical quantum computer.
Earlier this week the company used the word “breakthrough” to describe its recent work and promised “to release the world’s most powerful quantum computer within the next three months.”
It’s using trapped-ion (charged atom) technology to make qubits, the building blocks of a quantum computer. Until now, this was generally considered a less advanced approach than the super-conducting technology used by IBM, Google and others.
But Honeywell promised it will shortly reveal the world’s most powerful quantum computer in terms of quantum volume, a measure of quantum capability that goes beyond the number of qubits. “When released, Honeywell’s quantum computer will have a quantum volume of at least 64, twice that of the next alternative in the industry,” the company said.
That presumably is a reference to IBM, which said in January it has achieved a quantum volume of 32 within its 53-qubit quantum computer.
It also said it is on the way to increase its computer’s quantum volume by an order of magnitude each year for the next five years.
Honeywell also said that it will collaborate with JPMorgan Chase, a global financial services firm, to develop quantum algorithms using Honeywell’s computer.
The announcement caught quantum computing experts by surprise, including Michele Mosca, co-founder of the University of Waterloo’s Institute for Quantum Computing. He’s also co-founder of sofwareQ, which makes platform-agnostic quantum software designed to run on any quantum circuit platform or on simulators that run on standard computers for testing solutions.
“To be able to now surpass IBM in terms of quantum volume would be remarkable. But [Honeywell] is a serious company with serious people on their team.
“And it’s not like they said three years or even three quarters. I also respect it in that it’s precise. We can either validate it or not. I don’t see a lot of wiggle room.”
Honeywell said in 2018 it had been working a project using a trapped-ion platform for some time. A Maryland company called IonQ is using the same approach. But Honeywell had been quiet for over a year.
Mosca cautioned that whatever Honeywell pulls the cover off of this summer will still be in essence an experimental computer, unable to do run fabulous applications beyond the power of today’s classical computers, like break current cryptography. But having bigger quantum volume would be an achievement.
It also would mean the race to build a practical quantum computer now solidly has two platforms: Trapped ion and super-conducting.
To give some context — and to be very simplistic — current computing platforms use binary digits (bits) that can be zeros and ones. Quantum computing allows bits to be zeros OR ones because the information is stored and manipulated at the sub-atomic level. But to do that you have to create qubits. Creating qubits involves quantum mechanics, playing with physics, possibly chilling a computer to 460 degrees below zero or using lasers (depending on the approach) and likely a couple of billion dollars. And that may only get a machine that runs in a lab, not one that you can buy from a channel partner.
Practical quantum computers — anywhere from five to 20 years away — could open undreamed of possibilities for pharmaceutical and aerospace companies, as well as nation-states wanting to crack the encoded secrets of other countries.
Last October Google said its latest quantum computer did in just over three minutes a mathematical calculation that supercomputers could not complete in under 10,000 years.
Not that all this is easy. One of the reasons why it will continue to take so long to make a practical quantum computer is there are things like field gate errors and system noise that have to be controlled.
According to Mosca, the trapped ion approach Honeywell is using dates back to the mid-1990s when the technology was used to make atomic clocks. It’s been known for some time the approach could be used to build small computers. Building computers that scale is another thing that has yet to be proved.
Honeywell says its system leverages numerous, individual, charged atoms (ions) to hold quantum information. Electromagnetic fields hold (trap) each ion so it can be manipulated and encoded using laser pulses.
Honeywell says trapped-ion qubits can be uniformly generated with errors better understood compared with alternative qubit technologies that do not directly use atoms.
The trapped ion approach may be relatively easy because nature makes ions, Mosca said, but getting a lot of them on a chip “and not flying off in random directions” as nature prefers is hard.
The super-conducting approach said Mosca, essentially creates synthetic qubits. This approach can lead to problems.
“It’s a huge pain to get one qubit to work,” he said. After that it should be relatively easy to fabricate more on a chip. (Relatively easy if you have a couple hundred million dollars or more to spend.)
Briefly, Mosca says companies trying to build quantum computers are still wrestling with the problem of internal noise interfering with calculations of their qubits. That has to be conquered before quantum computers can scale. Honeywell’s announcement, however, “certainly makes this interesting.”