Mounted in a glass case in a San Francisco convention centre, IBM’s new quantum computer could be mistaken for an elaborate chandelier or a steampunk artwork.
Instead of relying on conventional transistors in silicon semiconductor devices to switch electronic signals, it harnesses quantum physics to do so in a much stranger and more powerful way.
Research labs have been working on quantum computers for decades and while companies like D-Wave produce them for research institutions, the Q System One is the first to be made commercially available for any organisation to use.
But you won’t be picking one up for the office any time soon. Quantum computing requires a very carefully controlled environment operating at freezing temperatures to allow for the manipulation of particles. Complex electronics run the system, which is also highly susceptible to vibration and even sound, so is usually tucked away in a custom-built facility.
So IBM is making its quantum computers available via the internet, letting companies and research labs buy time on it to run their experiments. Prototypes of the Q have been available since May 2016 and have been used by over 100,000 researchers around the world, including in New Zealand, to run seven million experiments.
The aim now is to open up the technology to see what people can do with it. For most applications, a conventional supercomputer is still more useful.
“Despite the progress we have made so far, we are just really at the beginning of that full journey,” says Dr. Dario Gil, the Director of IBM Research which employs over 3,000 researchers around the world, many of them working on quantum computing.
“You could say we’ve barely begun to walk.”
How does quantum work?
If classical computers store and transfer data as 1s and 0s, quantum computers store them as qubits, bits of quantum information. The main difference is that quantum computers exploit the phenomenon of superposition, which allows for qubits to be in multiple physical states at the same time.
“Think of a coin, its heads and tails,” explains Gil.
“But when it is spinning, is it heads or tails? You could argue it is in a superposition of head and tails, zeros and ones.”
This physics of quantum lets you exploit superposition to exponentially increase the amount of data that can be processed by a quantum computer. With two qubits you can test four possible outcomes simultaneously. As qubits are added, the testing potential increases exponentially.
Other aspects of quantum physics, such as entanglement and interference are also integral to the system working.
IBM’s Q System One is a 20-qubit machine, which is modest in its processing power. IBM also has a 50-qubit computer and Google has a 72-qubit version. But qubit processing power is just one measure of the potential of a quantum computer. The other element is reliability – minimising the errors the system makes, a much more complex task in the quantum world where everything is held in delicate balance.
“An extraordinarily important measure is the error rate that you have in these qubits. If you go from 50 to 100 qubits but don’t significantly reduce the error rate, it gets you nothing,” says Gil.
The challenge facing quantum engineers then is to increase qubits while reducing errors in the system. IBM’s Q System One has made great strides on that front – the error rate on the system is in the low single digits.
To make it easier to programme quantum computers, IBM has developed Qiskit, open-source software anyone intending to use its quantum computers can access. Its Q Network of research labs, universities and blue chip companies are the main quantum customers currently.
Daimler is using quantum computing to model materials for use in their cars while ExxonMobil sees potential in quantum to learn more about the properties of molecules, particularly as the oil giant looks to transition to cleaner energy such as biofuel.
One day, Fonterra may be using quantum computers to explore the make-up of milk proteins or NIWA’s scientists could employ it to model the impacts of climate change.
Nature isn’t classical
Without quantum, says Gil, we’ll struggle to use computers to model the real world, which is too complex and variable for conventional systems to handle. The brilliant physicist Richard Feynman said it best when he gave a lecture at an IBM event in 1981:
“Nature isn't classical, dammit, and if you want to make a simulation of nature, you'd better make it quantum mechanical, and by golly it's a wonderful problem, because it doesn't look so easy.”
Simulating nature is the real answer to some of our biggest medical and environmental challenges, so the race is on to come up with powerful and reliable systems to handle it. For the average user though, the quantum world will still be accessed through the plain old screen, keyboard and mouse. Ones and zeros will go in and ones and zeros will come out.
“What’s really special is what happens in between,” says Gil.
“It turns out that with classical computers, no matter how powerful they are, in the end we are bound to what are known as easy problems, things that don’t blow up exponentially in the number of variables we have to calculate.”
Quantum is of most interest currently to the materials industry, chemistry, medicine and environmental modellers. But it may also pose a threat to conventional security systems as its vast processing power is leveraged to break the encryption systems that keep our digital activity private.
For IBM, a 107-year-old company with a long legacy of innovation, quantum could give it the edge as it faces tough competition in its core business of enterprise IT systems and cloud computing.
This week in San Francisco, IBM revealed it would make its IBM Watson artificial intelligence engine available to customers using computing services from the likes of rivals Amazon and Microsoft.
It is also in the process of buying Red Hat, which produces the version of Linux that companies around the world use to run their systems. At US$33 billion, it is one of the biggest software acquisitions in history.
The future then for ‘Big Blue’ is about opening up its artificial intelligence and supercomputing power to a larger audience of potential customers who may primarily shop elsewhere. As it delves into the world of qubits, it will be looking to make its multi-billion quantum computing bet pay off.
Peter Griffin visited Think 2019 as a guest of IBM.