Quantum computing gets real: It could even shorten your airport connection

Quantum computers could potentially reduce the distances travelers need to walk in airports by helping airlines assign planes to gates more efficiently. REUTERS
Quantum computers could potentially reduce the distances travelers need to walk in airports by helping airlines assign planes to gates more efficiently. REUTERS


Technological advances are letting companies and researchers explore new practical uses for quantum computers, such as optimizing airline gate assignments.

One day, you might avoid a missed flight connection because of the weird ability of very small particles to act as though they are in two places at once.

This bizarre behavior of the subatomic world is what allows so-called quantum computers to perform some calculations far, far faster than their conventional counterparts. It also could soon be helping smooth some problems in our daily lives.

Ordinary computers store information as binary digits, or bits, which can be either zeros or ones. Quantum computers use qubits, or quantum bits, which are much richer objects. Their values can be a complex mixture of zero and one because they rely on this behavior of atoms and smaller particles. Qubits can also coordinate their actions with other qubits instantaneously, no matter how far apart they are—a phenomenon that Albert Einstein called “spooky action at a distance."

Eventually, quantum computers could make it possible to engineer materials at the molecular level and crack many of the defenses used to secure the internet.

The massive, high-quality machines needed to perform these tasks are likely still at least a decade away. But recently, quantum computers from companies including IBM and D-Wave Systems have beaten the world’s most powerful conventional computers at some calculations relevant to physics. While these quantum machines are still small and error-prone, the advances are spurring companies and researchers to pursue more practical applications—such as swiftly calculating how to minimize the distance airline passengers must go to make their connections.

“We are now living in an era where we really have the chance to investigate where should we use a quantum computer," says Karl Jansen, a physicist at the German research center Deutsches Elektronen-Synchrotron, or DESY, who is working on the flight-gate problem with IonQ, a quantum-computing company based in College Park, Md.

D-Wave has used its quantum computer to help clients determine driver schedules for grocery-store deliveries, the routing of cross-country promotional tours and cargo-handling procedures at the Port of Los Angeles.

The optimization problem

These tasks are examples of so-called optimization problems, which are devilishly difficult because of the huge number of options they entail. Others include how to most efficiently pack boxes into containers and balance risk versus reward in financial portfolios.

There are 100,000 ways to assign five planes to 10 gates at an airport. Dial that up to 50 planes and 100 gates, and the number of possibilities balloons to 10 to the hundredth power—far more than the number of atoms in the visible universe. No conceivable conventional computer could keep track of all these possibilities.

But a quantum computer potentially could.

Collections of qubits act much like waves that contain an enormous amount of data. A quantum computer containing just 350 qubits could theoretically keep track of all the possible solutions to the 50-aircraft-to-100-gate flight-assignment problem. (Today’s machines generally have tens or hundreds of qubits.)

Angelo Bassi, a physicist at Italy’s University of Trieste, describes the difference between ordinary and quantum computing as like that between a surfer and a wave when they encounter a rock. The surfer goes to the left or right of the rock, while the wave does both at the same time. Some basic features of the rock can be deduced from the surfer’s path, but much more can be learned from the pattern of ripples that results in the water.

“Waves carry more information than particles," Bassi says.

But qubits are exceptionally hard to work with. Often created with superconducting circuits or trapped ions, qubits are easily destroyed by the slightest disturbance and must typically be cooled to temperatures lower than those of interstellar space. Even then, qubits are far more susceptible to errors than bits, which rely on ordinary electronic circuits.

Quantum computers of the future will need a huge number of qubits—possibly millions—to handle the error problem and still have enough firepower left over for tasks such as simulating the dynamics of atoms and molecules, a 2022 Microsoft study found.

Crossing a threshold

But even today’s comparatively puny devices have crossed a threshold that makes them powerful enough to outdo the world’s most advanced supercomputers in some calculations. That critical point lies somewhere between 50 and 100 qubits, says Travis Humble, director of the Quantum Science Center at Oak Ridge National Laboratory in Oak Ridge, Tenn.

A milestone came in June of last year when IBM published a study in Nature showing its 127-qubit processor’s ability to beat conventional computers at certain calculations related to magnetic materials. Then this March researchers at D-Wave posted a paper, not yet peer-reviewed, saying that its latest machine, applied in a similar situation, can compute quantities in minutes that would take the world’s most powerful supercomputer millions of years.

“Of all the computational supremacy claims so far," of quantum compared with conventional computers, “this one is actually the strongest," says Daniel Lidar, director of the University of Southern California’s Center for Quantum Information Science & Technology.

D-Wave has bet big on optimization applications by developing a special type of quantum computer called an annealer, tailored to solving this type of problem. It contains about 5,000 qubits but is limited to looking for approximate answers quickly rather than performing exact calculations.

In a promising sign that D-Wave’s annealing technology might have an advantage over ordinary computers for practical problems, USC’s Lidar showed earlier this year how it could be used to win a mathematical game akin to optimization, in a paper that is currently being peer-reviewed.

Now, the race is on to figure out what other practical uses the latest generation of quantum computers can be put to.

Jansen, of Germany’s DESY, says he has successfully solved small versions of the flight-gate optimization problem on a trapped-ion quantum computer made by IonQ and has glimpsed early hints that his technique, at large enough qubit counts, might outperform conventional computing methods.

Using a similar approach to Jansen’s, researchers at the Cleveland Clinic say an IBM quantum computer bested a state-of-the-art artificial-intelligence algorithm at predicting the shape of a section of a protein molecule from knowledge of its amino acids—a task that could be useful for detecting and treating certain diseases as the capabilities of quantum computers evolve. Their paper was published in the ACS Journal of Chemical Theory and Computation.

Souping up AI

Since optimization is involved in training machine-learning algorithms, some companies think quantum can make AI applications even smarter.

IonQ has been working with Hyundai on quantum-powered AI to enable self-driving cars to recognize road signs and other objects. Switching to quantum-based AI training in a small machine-learning model doubled its accuracy to 60% from 30%, IonQ says. Once its qubit count grows from the current 36 to 64—planned for next year—the company believes its algorithm will outperform any imaginable non-quantum machine-learning model.

Trial-and-error approaches could certainly uncover new uses for quantum computers, says Scott Aaronson, director of the Quantum Information Center at the University of Texas at Austin. But known theory suggests that quantum speed-ups in optimization and AI will be relatively modest and are unlikely to have commercial impact until quantum computers are much larger and error-corrected, he says.

“It is really, really hard to see how with the current generation of devices you could get a win," Aaronson says. “Something has to happen that would go outside of what we know about any of the current algorithms."

But that is exactly what some of today’s quantum pioneers are hoping for: a breakthrough that springs from experimentation.

It has happened before, says Ricardo Garcia of Moody’s Analytics, who worked with quantum-computing company Rigetti on a project to boost the accuracy of an AI-based recession-forecasting model. One of the most powerful methods used for optimization problems today, the simplex algorithm, was devised in the 1940s, he notes, long before theorists could explain why it worked so well.

“Just because there are no theoretical guarantees today," Garcia says, “that doesn’t mean that there are no opportunities in the near term."

Write to Bob Henderson at bob.henderson@wsj.com

Catch all the Technology News and Updates on Live Mint. Download The Mint News App to get Daily Market Updates & Live Business News.