Quantum Computers Bust Problem Conventional Computers Can't Solve
Fluorescent lights emerge from the University of Maryland quantum simulator, revealing the results of the experiment.
Credit: University of Maryland

Fifty-odd atoms buzz through a pocket of empty space. Invisible lines of force — quantum magnetism — chain them together. Jiggle one, the others jiggle in sympathy. Ring another like a bell and the others will pick up the song at a different pitch or a slower speed. Every action on any one atom impacts each other atom in the 50. It's a tiny world of unfolding subtlety and complexity.

There are limits in our larger world that make such jiggles tricky to predict. For instance, nothing moves faster than the speed of light and no frozen point gets colder than absolute zero. Here's another limit: Our clunky, classical computers can't predict what will happen in that little world of 50 interacting atoms.

The problem isn't that our computers aren't big enough; if the number were 20 atoms, you could run the simulation on your laptop. But somewhere along the way, as the small world swells to include 50 atoms, the problem of predicting how they'll behave too difficult for your laptop, or any normal computer, to solve. Even the biggest conventional supercomputer humanity will ever build would lose itself forever in a labyrinth of calculations — whatever answer it might eventually spit out might not come until long after the heat death of the universe. [The 18 Biggest Unsolved Mysteries in Physics]

And yet, the problem has just been solved.

Twice, actually.

Two laboratories, one at Harvard and one at the University of Maryland (UMD), built machines that can simulate quantum magnetism at this scale.

Their results, published as twin papers Nov. 29 in the journal Nature, demonstrate capabilities of two special quantum computers that leap far beyond what any conventional or quantum computer previously built has been able to accomplish.

Referring to the machine in his laboratory, Mikhail Lukin, one of the leaders of the Harvard team, told Live Science that "It's basically a quantum simulator."

That means the computer is built for a specific task: to study the evolution of quantum systems. It won't be breaking encryption codes on the world's banks, finding the highest mountain in a mountain range or pulling off any of the other tasks for which general quantum computers are suited.

Instead, the Harvard and UMD machines are really good at solving a particular kind of problem: If a complicated quantum system starts in one state, how will it move and evolve?

It's a narrow question, but in solving it, the researchers are developing technologies and making new discoveries in physics that will allow for even more complicated computers, which will pull off even more impressive tasks.

Maryland's and Harvard's quantum simulators are similar in a lot of ways. They solve the same sorts of problems. They use individual atoms as qubits — the fundamental units of quantum computers. They involve expensive lasers and vacuum chambers. But they're not the same.

At Maryland, the qubits are ions — electrically charged atoms — of the silvery-white metal ytterbium. The researchers trapped 53 of them in place, using small electrodes that created magnetic fields in a vacuum that was far emptier even than outer space. Then, they struck them with lasers in a way that caused them to cool way down, until they were nearly still. [Elementary, My Dear: 8 Elements You Never Heard Of]

The UMD qubits stored their information deep inside the atom as "spin states" — special quantum-mechanical features of small particles.

"The thing about quantum bits is that they hold all their information as long as they're isolated," Christopher Monroe, who led the Maryland team, told Live Science.

But if researchers let those qubits shake around too much, or crash into air particles or even measure the spin state the qubit holds, all that data gets lost. (Under the mind-bending rules that govern the quantum world, measuring or even observing a subatomic particle alters it.)

Those magnetic fields pin the atoms in place without touching them, allowing them to remain mostly undisturbed.

Once Monroe and his team had the ions where they wanted them, they pushed on them, again using lasers. That push had a quirky effect, though.

"We apply a force to the atom that pushes the atom [different ways], depending on the spin state of the qubit."

But because the state of the qubit is unknown, the strange laws of quantum mechanics cause the atom to move in both directions at the same time. The tiny particle smears itself across space, turning into a fairly large quantum magnet that interacts with all its siblings in the electrode trap.

Once all the ions have spread and transformed in this strange way, they interact with one another very quickly. The researchers observe the results, and the simulation is complete.

Harvard's simulator doesn't work with ions or electrodes.

"What we have is about 100 individual, tightly focused laser beams focused on a vacuum cell," Lukin said. "Inside the cell is a very thin vapor of rubidium atoms."

As if they're fine optical tweezers, those lasers pluck individual atoms out of the vapor and trap them in place. And they allow the Harvard team to finely program their device, arranging the atoms into exactly the setup they want to test, before they begin their simulation. [Beyond Higgs: 5 Elusive Particles That May Lurk in the Universe]

Once all of the atoms are set in space, and the whole system cools to near-absolute zero, the machine again strikes the atoms with lasers. These lasers don't move or cool the atoms, though. Instead, they cause them to grow excited — and enter something called a Rydberg state.

In a Rydberg state, the atoms don't get smeared between two points. Instead, they swell.

Every atom has electrons orbiting around it, but usually those electrons stay confined to tight orbits. In a Rydberg state, the electrons swing wider and wider, farther and farther away from the core of the atoms — until they cross paths with the other atoms in the computer simulation. All these wildly excited atoms suddenly find themselves sharing the same space, and — just like in the Maryland machine — interact with one another as quantum magnets that the researchers can observe.

A 50-qubit quantum simulator is interesting, but it isn't yet incredibly useful. Monroe said the next step for his lab is to go bigger, to create arrays of 50-plus-qubit quantum simulators networked together to simulate even more complex quantum events.

He also said that his team's and Harvard's atomic qubits offer a roadmap for other groups trying to build quantum machines.

"The great thing about atomic qubits is that they are perfect," he said.

Unlike more complicated, bigger "solid-state" qubits printed on chips in labs at Google and IBM, an atomic qubit will hold on to its information as long as it's undisturbed.

The challenge for researchers like Monroe and Lukin is to build lasers and vacuum chambers that are precise enough that they won't disturb their growing arrays of qubits.

Originally published on Live Science.