Making it big: The promise of quantum computers

Innovations in quantum computing may have the power to change the world. But how close are we really to a quantum revolution?

How close are we to technologies like quantum computing becoming ubiquitous?    D-Wave Systems Inc./Wikimedia Commons  (CC BY 3.0)

How close are we to technologies like quantum computing becoming ubiquitous? D-Wave Systems Inc./Wikimedia Commons (CC BY 3.0)

Listening to our less circumspect pop-science pundits, you could be forgiven for thinking that a quantum revolution is imminent. Quantum computers have the potential to break common cryptography techniques, search huge datasets, and simulate quantum systems in a fraction of the time it takes their classical counterparts. Quantum key distribution would allow parties to guarantee – quantum mechanically – that their messages haven’t been intercepted. Innovations in quantum optics, biophysics and inertial sensing promise similarly exciting applications.

While toy implementations of such systems abound in physics departments and R&D labs all over the world, there are formidable barriers to upscaling prototypes so that they can work their magic in the macroscopic realm. With the exception of D-Wave (a company that makes and sells a controversial style of quantum computer), we have yet to see any next-generation quantum technology go commercial. Putting the hype to one side, how close are we to getting quantum computers?

A good measure of progress is investment, particularly by technologists and military-industrial heavyweights. One such heavyweight is the US Air Force; their Scientific Advisory Board (SAB) is currently assessing the suitability of quantum, cyber and unmanned technologies for use in Air Force operations. Although the report on quantum systems won't be released until December, comments at a recent press conference indicated that results were mixed, with the SAB seeking to downplay the hype surrounding quantum technology. That said, they are likely to recommend further research into quantum algorithms to help cope with the exabytes of data generated by Air Force surveillance and intelligence programs. Quantum communications were less favourably appraised, with existing classical protocols said to offer similar levels of security.

A quantum machine developed by the Martinis Lab, designed to place a mechanical resonator in a superposition.   Martinis Lab/Wikimedia Commons  (CC BY 3.0)

A quantum machine developed by the Martinis Lab, designed to place a mechanical resonator in a superposition. Martinis Lab/Wikimedia Commons (CC BY 3.0)

The US Air Force is a conservative organisation by nature, with a limited window for R&D turnaround and a focus on specific applications. Google is a rather different beast though, and in recent years the Internet giant has been actively involved in pushing the quantum computing envelope. In 2013, they partnered with NASA and the Universities Space Research Program to form the Quantum Artificial Intelligence Lab. A year later, they launched a hardware initiative with John Martinis’ lab at the University of California, Santa Barbara, focused on building high-fidelity arrays of quantum bits (qubits).

One of the problems they are tackling is decoherence, also known as quantum noise. Classical computer hardware is extremely reliable, with internal errors often said to be less likely than faults due to cosmic rays. Qubits, on the other hand, exist in a fragile superposition of classical bits that needs to be carefully shielded from the environment, since interactions lead to the state of a qubit changing. In practice, total isolation is impossible, so robust methods for error-correction are a big focus of research. One approach is quantum error-correcting codes, which are close in spirit to classical error-correcting codes, like checksums or Hamming codes. A major difference is that a qubit cannot be directly measured without changing its state, so get around this, measurement qubits, or meas, are attached to the qubits carrying real data. The meas are partly correlated with the data qubit, but can be sacrificially probed while leaving the data alone. With enough meas, the probability of detecting errors becomes high.

Earlier this year, the Martinis lab made headlines (and the prestigious journal Nature) by creating an error-correcting superconducting array of 9 qubits, using a paradigm called surface codes. Dr. Austin Fowler, a senior researcher in the Martinis lab and an expert on surface codes, explained more.

“Imagine a checkerboard with each square a qubit. Black qubits check for errors in their neighbouring four white qubits. The simplicity of doing this allows the code to tolerate a high error rate and simultaneously makes it highly suitable for implementation using microfabrication techniques,” Dr. Fowler said. Microfabrication is the process of creating small, high fidelity structures, and while current techniques are up to the task of a 9-qubit linear array, scalability is an open question. In fact, according to Dr. Fowler, the major challenge facing quantum computing is achieving precision control in a scalable way. “In our case that means cleaner materials and smarter control software in the short term,” Dr. Fowler explained.

A surface code can be thought of as a checkerboard, with black squares representing the measurement qubits that look for errors.   Eduardo Fonseca Arraes/Flickr  (CC BY-NC-ND 2.0)

A surface code can be thought of as a checkerboard, with black squares representing the measurement qubits that look for errors. Eduardo Fonseca Arraes/Flickr (CC BY-NC-ND 2.0)

One way to make control software smarter is to think about its topology. Imagine tracing the history of a surface code as it moves through time. As the computation runs, information moves around the array, leaving tracks intertwined like knots in space-time. This way of viewing things turns out to be useful. “Computation involves turning off and on regions of the array of qubits. If you visualise the off regions in 3D space-time, they look like complicated branching knots. I have students working on the compression of these 3D structures, which I think is a fascinating and beautiful problem,” Dr. Fowler said. Work on this problem isn’t limited to research students; meQuanics, an online game developed by Simon Devitt, Fowler, and colleagues, lets users manipulate these braided 3D structures and solve real design problems.

Collaborative ventures like NASA and Google's Quantum AI Lab are a sign of optimism, but no one really knows what the time frame for solving these problems is, or if they can be solved at all. “Building a quantum computer is an intellectual and physics-stretching challenge for humanity,” said Dr. Fowler. “It is still very much a case of research to see whether we can do it at all, not a steady march to a commercial product with a predictable timeline.” Given uncertainty about that timeline, are quantum computers and their cognates oversold as a short to medium-term prospect? “We are not personally in the business of overselling,” Fowler noted. “We just want to see if we can build a quantum computer. We believe that the potential applications are very hard to predict, but the idea of a fundamentally different computational device is thoroughly worth investigating.”

To defence agencies looking for immediate operational leverage, quantum systems may seem overhyped and underdeveloped. But the gap between theory and existing technology is only half the story. Efforts to create quantum devices are already leading to new physics, new hardware, and new formal techniques for reasoning about systems and controlling their behaviour. It’s a noble, ambitious goal, and investing capital and personnel is likely to result in new directions and unpredictable benefits – and who knows, we may get that quantum revolution after all. Eventually.