Forty years ago, on 6–8 May 1981, a group of physicists and computer scientists got together at MIT’s Endicott House. The event was the Physics of Computation conference. The hottest topic of discussion — the possibility of mimicking nature to design ever more powerful ways of computation.
The possibility of building a quantum computer.
Fast-forward to today. Some of those physicists have gathered at a celebratory event QC40 this week to chat about the future and the present of quantum computing. After all, several companies have been busy building these machines, using different approaches. So how can one tell what prototype quantum computer is in the lead?
Whatever the approach, all quantum computers rely on qubits to work. Short for ‘quantum bits,’ they are the fundamental information units of a quantum computer, analogous to the bits of a classical machine.
But the hype around qubits is misplaced: their number alone doesn’t matter that much. Qubits don’t make a quantum computer powerful. They don’t even make it work. Just like a car with a 1,000-horsepower engine is useless if you can’t corner or brake without crashing — a million qubits won’t bring you an inch closer to building a fully-functional quantum computer.
What matters is the machine’s ability to run complex quantum algorithms, or circuits, that can’t be simulated classically. That’s the essence of a quantum computer of tomorrow.
In today’s electronics, circuits are binary with only two possible states. These circuits complete the electron flow in a computer and make modern electronics work. A circuit of a quantum computer is a set of instructions on quantum data, an ordered sequence of operations played on different qubits. Running a quantum circuit requires the combined effort of the qubits, the hardware and the software.
How good are your qubits?
It’s the ability to execute specific quantum circuits that defines just how powerful a quantum computer really is — not in terms of the number of qubits but in terms of how stable and interconnected they are. That’s what companies should pay attention to when choosing a quantum computer for their specific task. That’s the notion of ‘quantum volume’ — the quality, capacity, and variety of quantum circuits.
At IBM, we aim to double quantum volume every year to eventually achieve quantum advantage — the moment when a quantum computer should outperform a classical machine in a meaningful task. From that point on, quantum computers and classical systems will be working together, leading to a significantly better performance than classical systems alone.
We expect to reach a quantum advantage before the end of this decade.
Ultimately, we aim to keep the so-called ‘frictionless development’ of quantum computing, bringing us to the point when you won’t have to be a quantum expert to use these nature-mimicking machines. In the future, developers should be able to program in their familiar environment, without having to worry about learning the intricacies of quantum gates and circuits.
To get there, three advances should come together first: better hardware, better software and improved algorithms.
Take the hardware, the beautiful gold-plated steampunk-like ‘chandelier.’ Researchers keep improving the cryostat, the control electronics, the amplifiers, all the cabling and the nanofabricated components on the chip that make up the qubits, kept at 15 millikelvin — nearly 200 times colder than outer space — and in total vacuum. Researchers manipulate the states of the qubits with ultra-precise microwave pulses they send into the cryostat through the snaking cables.
As the pulses arrive, electrons flow though oscillators made of metals such as niobium or aluminum that, when cooled below one degree Kelvin, become superconducting. Electrons flowing through them make superconducting qubits act as atoms, obeying the laws of quantum mechanics. When two qubits reach the same resonant frequency, they get entangled. Just like the shape of a musical instrument sets the timbre of the notes it can play, a quantum machine has frequencies determined by its physical properties.
Keeping the ‘noise’ down
Researchers are constantly improving and calibrating the microwave pulses, making them ever more precise. They can be sent from anywhere in the world through the cloud, using the ever-improving software.
At IBM, it’s the Quantum Composer, Quantum Lab and Qiskit, the open-source software development framework. Thanks to the software, developers control the placing of quantum gates — state manipulations — and run the measurements, improving the execution of their quantum programs and applications. All of this is part of quantum circuits.
The cold temperatures and the vacuum shield the qubits from the disturbances of the outside world, the so-called ‘noise’ — anything from physical vibrations of the scientists walking around, to heat, light, magnetic fields or a stray microwave pulse.
Any of those external effects impact the qubits, yanking them out of their quantum states of superposition and entanglement — because only in these states qubits can perform meaningful calculations. Once they are out — once they de-cohere — scientists get computational errors and the quantum computer no longer computes. Researchers are working hard on trying to suppress the noise as much as possible.
At IBM, we have a prototype quantum computer that works with 65 qubits, kept in superposition for just a few fractions of a second before they decohere, or reset. Later this year, we aim to have one with 127 qubits. That’s not enough to reach a quantum advantage, but that’s already very promising.
To maintain superposition for longer, we need to ensure that our qubits are very low noise. Then we’ll be able to correct any remaining errors using classical computers. But this approach of error correction is still theory as we can only apply it once we scale the number of qubits to hundreds or more and lower the error rate in their operations.
When that happens, working together, these low-noise, error-corrected physical qubits will form one so-called logical qubit. And going forwards, we will need hundreds of these logical qubits for a quantum computer of the future to become better than a classical computer in at least one meaningful task. That will be the moment of achieving quantum advantage.
The future of error correction
Hardware will never be error-free, so we just need to make sure that errors that the hardware introduces are of the type that we can correct using error correction codes. And we have to make sure that the fridge — the cryostat — doesn’t collapse. That could happen, if we were to continue adding, by brute force, more and more superconducting circuits to the bottom of the fridge. A cryostat with even one logical qubit made of, say, 500 physical qubits would be a structure of half a ton — and that is simply unfeasible.
To reduce the amount of weight and cost, researchers are looking into different approaches. As microwave signals zip down from room temperature to the quantum processor, they get smaller and smaller, and need to be amplified to be measured. One approach is to make the cables smaller and lighter but amplify the signals more efficiently.
And finally, there are algorithms. They are getting more and more elaborate, aimed at ever more complex problems of the future — from simulating large molecules to predicting financial risk.
Soon, thousands upon thousands of quantum circuits will form quantum libraries, thanks to all the developers, researchers and enthusiasts of the Qiskit community, who keep designing new ones. In the future, you’ll just have to choose the right circuit from an online open access library — call it a ‘quantum app store’ of sorts — and plug it into your application through the cloud. Companies like IBM will be helping organizations around the world to choose the best circuit to solve a specific problem. The quantum magic will happen in the background in the cloud all on its own.
That’s the future. And the roadmap of hardware, software and algorithm advances should get us there in no time.