The future of quantum RAM? Scientists watch two time crystals interact
Here's an opening line pulled straight from the pages of a science fiction novel: a team of quantum physicists has finally succeeded in creating a system composed of two time crystals.
In 2016, researchers created the first lab-grown time crystals. Now, the team of quantum physicists has upped the ante by joining the time crystals together according to a recent study published in the peer-reviewed journal Nature Communications.
Crucially, the team even observed the interactions of time crystals.
But while the breakthrough could have huge implications for quantum computing — industry experts tell IE that it's too soon to say concretely how they'll unfold.
The reality check of time crystals
Time crystals don't let you see the future or travel back in time, sadly. But they're remarkable in a different way: Ten years ago, a Nobel Prize-winning physicist theorized that it should be possible — under precise conditions — for some kinds of matter to perpetually oscillate through a series of repeated motions. The remarkable thing about these supposed "time crystals" was that they didn't require any additional energy. According to the research, the laws of physics in the quantum realm should, in theory, allow for these miniature perpetual motion machines.
Interesting Engineering sat down with a pair of experts from the quantum computing industry to discuss the practical applications of this new development. Mark Mattingley-Scott is an executive at Quantum Brilliance, and Florian Preis is a quantum physicist and developer at the same firm.
This interview has been edited for length and clarity.
Interesting Engineering: What stands out to you about the new paper?
Mark Mattingley-Scott: It's the use of the time crystals. It sounds very science fiction-y. When you make qubits, you can draw a Josephson junction transistor on a piece of silicon, take it down to absolute zero, and put it in a vacuum. Or, you can take a nitrogen-vacancy in diamond, or you can trap an ion. These are all things you'll be able to see with a good enough microscope.
The interesting thing with this paper is that the qubits are being made in basically a system in which you're replacing the spatial dimensions with time. It's an interesting technique, an interesting technology. Is it ready for the market? Or when will it be? Very hard to say. I think it's still very much in its very, very early stages.
IE: From an industry perspective, are there any likely advantages or disadvantages of this new approach?
Mattingley-Scott: I can talk about a disadvantage, which is that you're still going to need a lot of infrastructure. To make these qubits, you probably need to get down to very low temperatures.
Florian Preis: I want to add to the problems of this approach, at least in the way it is realized using the magnets in helium 3. Apart from the infrastructure, the question for me is also how to scale up. How would you get so you [scale up from] one qubit at the moment, which seems to be in the millimeter size range. Can you scale up by doing something else in terms of cubic topology and connectivity, but only maybe a linear chain that this is the setup at the moment? These are the questions that arise for me.
IE: Do you see any advantages?
Mattingley-Scott: It's very difficult to say. Maybe the coherence time.
Preis: Coherence time is the big advantage here. Time crystals have a long coherence time, which is one of the reasons they might be considered in the future as a quantum RAM — a device for quantum memory — at some point. I've seen a few mentions of this in some articles.
IE: Could you talk a little more about coherence time? What does it mean, and why is this option more attractive?
Mattingley-Scott: With qubits, you need to put some simple quantum mechanical system into a state where you can do stuff to it, where you can do operations to it. That may be gate operations. In the case of an analog quantum computer, you're doing manipulations of the quantum mechanical system.
But sooner or later, that quantum mechanical system decays or descends basically into noise. That's the point at which it's useless: you've lost coherence. Anything you do with this qubit, or that qubit, or that quantum mechanical system — any computation you want to do — you have to do it within the coherence time.
The longer the coherence time, the better. The other parameter is, how long it takes you to do an individual operation. It's quite important to understand the relationship between the two but in general, more coherence — longer coherence time — means you can do more.
Preis: It's basically the limit of how many operations you can do.
IE: What magnitude of operations are we talking about here?
Florian: I didn't find anything about gate times in the article. That is, how long one operation would take. Again, we are just talking about a single-qubit system here. Two qubit gates usually have a different timescale than the one qubit gate times. And this is not stated. So we cannot say if — comparing the coherence time — there is an advantage there.
IE: Let's turn to the state of the industry that time crystal-enabled quantum computers might enable in the future. What's the state of the industry right now?
Mattingley-Scott: The quantum computers we have — it doesn't really matter what the underlying technology is — are they're giving us a few 10s or a couple of 100s of qubits. The things we can do with them are still so limited that they are really not comparable to in performance to a classical computer of about the same size. The number of qubits definitely is nowhere near that computing power, what you get if you took the same volume of space and you filled it with classical computers.
IE: And yet, a lot of people are extremely excited about quantum computing. How will we start seeing these machines used in the real world?
Mattingley-Scott: It's captured in this concept we call quantum utility, which is the recognition that quantum computers for the foreseeable future — in the next 20 or 30 years — will be used as accelerators in much the same way that a GPU is an accelerator today. It's something you add to a CPU. We call them QPUs, just to be consistent. QPUs will be deployed to make algorithms faster, but they won't completely replace classical computers.
We're close now to having a positive business case where somebody, for some specific problem — in business or in chemicals or pharmaceuticals or finance — can say, "Okay, if I buy one of these things, it's more accurate or faster or better than the classical alternative."
IE: Time crystals are still mostly the domain of researchers, but things are changing all the time in the quantum computing industry. What changes have you seen recently?
Preis: There's some change last year and this year as solid-state technologies are coming, becoming more mature. One of these, of course, is [the use of] diamonds. Another example of solid states is silicone dots. This is quite new, at least in the industrial area. Superconducting has been there for a long time, and trapped ions have, too. But what I can see there is that, at least in terms of fidelity, there doesn't seem to be much progress anymore, at least for superconducting qubits. And for the trapped ions, they are also maxing out what they can do at the moment in one linear trip. It will be interesting how they then bridge this gap in the future.
In terms of application, there is definitely a turn towards parallelism. On the software side, you see more workflow orchestration frameworks becoming more mature. It's all about how to orchestrate many of these QPUs because — for these noisy, intermediate scale quantum systems in particular — it will be very important as they run on so-called variational quantum algorithms to leverage parallelism to get to reasonable run times. That's what will bring quantum advantage or utility, as we call it.
Experts say this year's monsoon rains in Pakistan were the worst ever, flooding one-third of the South Asian country, ruining crops on hundreds of thousands of acres, and uprooting at least 33 million people of its 220 million population.