String of ions may out-compute best quantum computers

New ion-based quantum computer debuts by computing the ground state of water.

Making a qubit is easy. Controlling how they communicate, however...
Making a qubit is easy. Controlling how they communicate, however…
We will update this story with links to the preprint when it goes live on later today.


Usually, I reflexively delete press releases. This one was no different, but as the message vanished, the subject line registered—“IonQ… quantum computing.” It took a second, but I realized that the name might mean something I had never expected: a commercial, ion-based, quantum computer. A quick visit to the trash confirmed my suspicion.

After some negotiation, I was in receipt of a super-secret paper demonstrating the capabilities of IonQ’s new computer.

Engineering ions is not simple

Making an ion-based quantum computer seems like a bad idea. Think about the engineering required to commercialize the computer. You need to have qubits (quantum bits) that are interconnected so that they can perform logic operations, and those qubits need to preserve their quantum-ness. 

Pretty much all commercial quantum computing efforts focus on using superconducting rings of current as the qubit. These circuits can take advantage of all the engineering tools available for printed circuit board technology. The control circuits and readouts are all electronic—they send in and receive microwave signals via lines on a printed circuit board. The qubits are interconnected by lines that couple them together. In other words, the engineering is comparatively easy.

In research labs all around the world, however, there are small-scale quantum computers based on strings of ions (an ion is an atom with an electron removed). The ions float in a near-perfect vacuum, trapped by electric fields. Each ion needs to be addressed by two laser beams. The interconnection between qubits occurs via the natural motion of the qubits: they all vibrate together. 

Ion qubits and their logic operations outperform their superconducting brethren by a huge margin. But engineering this type of computer at a commercial scale has been an entirely new challenge.

There’s water in your computer

A paper released by researchers uses the IonQ computer to calculate the ground state energy of a water molecule. The calculation itself is something almost any modern computer can do. What makes the calculation stand out is the number of operations required to complete the operation. In choosing water, the researchers have shown ion-based quantum computing at its best. 

Let’s put this in perspective. To model a water molecule, the researchers use a standard approach, where it’s assumed that the electrons in a water molecule take on a bit of the character of electrons that are in oxygen and a bit of the character of electrons that are in hydrogen. The trick to the calculation is to determine the balance of the admixtures and their energies.

The calculation that does this is a repeated approximation, where additional correction terms make the result (hopefully) more accurate as the calculation proceeds. This allows you to, in a sense, choose your accuracy based on when you stop doing math. As you might expect, each correction term requires more computational resources—for a quantum computer with only a few qubits, that’s a challenge.

But if you have nice, stable, long-lived qubits that all talk to each other with a high degree of reliability, then you have a bit of wiggle room. That’s one of the key points of this paper: the ion computer allows you to play some clever tricks that reduce the total number of qubits needed in exchange for increasing the number of operations required. That only works if you have time to perform all the operations, and time is something that quantum systems don’t always provide in abundance.

Zooming in

According to the press release, IonQ’s computer has either 160 or 79 qubits (depending on whether the computer is storing or operating on quantum information). Looking at the circuit diagram (the program, essentially), I estimate that the calculation required some 30 qubits. That in itself is quite a jump from the typical ion quantum computer, which has about ten ions.

That, however, is the least of it. The calculation requires many sequential gate operations, and unlike digital logic operations, quantum logic operations are not exact. The error in each operation accumulates over many operations, which will leave a calculation in tatters. Ion computers, however, have very high precision in their operations. The researchers reported that they were able to perform 50 consecutive operations while still retaining the qubit in the correct state about three quarters of the time.  

The interconnectedness of the ion computer also played an important role in the calculation. The researchers were able to directly entangle arbitrary pairs of qubits during the calculation. In other quantum computers, geometry does not allow all the qubits to be interconnected. As a result, any computation requires information to be swapped back and forth between qubits. Since the information fades away after a certain number of operations, each additional step to move information around reduces the amount of useful computation that can be done.

The end result is an answer that is very close to the results obtained from standard calculations performed on classical computers. And it’s surely not the end. Hopefully, IonQ release more details about the computer soon—you can be sure we’ll cover it when they do. Even in the absence of more technical details, I’m pretty sure we will see a steady stream of results from users.