What makes a quantum computer good?

Eleven years ago, I was just starting to write my doctoral dissertation in theoretical physics and, to be honest, I had never thought about quantum computers or written about them at all. Meanwhile, New scientist The staff worked hard to create the world's first “Quantum Computer Buyer's Guide(we were always one step ahead). Looking through it, you understand what a different time it was – John Martinis at the University of California, Santa Barbara, has been recognized for his work on an array of just nine qubits, and just last week he was presented with an award Nobel Prize in Physics. Meanwhile, quantum computers made from neutral atomswhich have taken the battlefield by storm in recent years are not even mentioned. This got me thinking: What would a buyer's guide for a quantum computer look like today?

Currently, about 80 companies around the world produce quantum computing equipment. Because I talk about quantum computing, I've had the opportunity to watch the development of this field as an industry up close—and hear a lot of sales pitches. If you think choosing between an iPhone and an Android phone is difficult, try getting on the press lists of dozens of quantum computing startups.

Of course, a lot of the hype comes from the marketing, but some of the difficulty in comparing these devices and approaches stems from the fact that there is currently no consensus on how best to build a quantum computer. For example, you can choose qubits made from superconducting circuits, extremely cold ions, light, or some other options. How do you weigh the differences between these machines if they have fundamentally different parts? This helps focus on the performance of each quantum computer.

This is a marked shift from the early days of quantum computing, when the champions of these new devices were determined by the number of qubits—the most basic building blocks of quantum information processing—the machine had. Currently, several research groups broke the 1000 qubit barrier and the path to more and more qubits seems to be becoming clearer every day. Researchers are currently working on how to use standard manufacturing technologies, such as making silicon qubits And even using AI make your quantum computers bigger and more powerful.

In an ideal world, more qubits would always mean more computing power, as this would allow a quantum computer to solve more complex problems. In our real world, ensuring that every new qubit you add doesn't degrade the performance of the ones you already have has proven to be a huge technical challenge. So, it's not just about the number of qubits you have, but also how well they can store information and how well they can “talk” to each other without degrading the quality of that information. A quantum computer could have millions of qubits and be essentially useless if those qubits are prone to glitches that introduce errors into the calculations.

This glitch—or noise—can be quantified using metrics such as “gate precision,” which measures how accurately you can make a qubit or pair of qubits do something, and “coherence time,” which measures how long a qubit can remain in a useful quantum state. But these measures bring us back to the nitty-gritty details of quantum computing hardware. The annoying thing is that even if these numbers are good, you'll still have to worry about how difficult it will be to input data into your quantum computer and start calculating, and whether you'll have problems trying to read the final result.

Part of the quantum computing industry's significant growth has come from the emergence of companies specializing in qubit control and other parts of quantum computers that deal with the complex interface between the quantum guts of these devices and their very non-quantum users. A proper 2025 quantum computer buyer's guide will need to include all of these additions. You will have to choose qubits, as well as a system for managing the qubits and some mechanism for correcting the errors of those qubits. I had the opportunity to talk to researchers who are even developing operating system for quantum computers, so this could also be part of your shopping list in a few years.

If I had to make a wish list for the near term, I would bet on a machine that can perform at least a million operations (roughly speaking, a million-step quantum computing program) with a very low error rate and as much error correction built-in as possible. John Preskill at Caltech they call it “megacop” car. He told me last year that he believed such a machine would be powerful enough to be fault-tolerant and error-proof, as well as make scientifically significant discoveries. But we're not there yet. The quantum computers we have today perform tens of thousands of operations and have only demonstrated error correction for relatively small tasks.

In some ways, today's quantum computers are experiencing adolescence, maturing and becoming useful, but still experiencing growing pains. For this reason, the question I ask quantum computer sellers most often in my inbox is: “What can this machine actually do?”

Here we will have to not only compare different types of quantum computers, but also compare them with traditional analogues. Quantum hardware is expensive and difficult to create, so when will it really become the only viable option to solve the problem?

One way to answer this question is to try to identify calculations that ordinary computers could not perform unless they had an unlimited amount of time. Colloquially known as “quantum supremacy,” it keeps mathematicians and complexity theorists up at night, and steals the sleep of quantum engineers. Examples of quantum supremacy do exist, but they are unpleasant. To be meaningful, they have to be practical – you have to be able to build a machine that can do them – and they have to be provable so that you can be sure that a clever mathematician won't be able to get a regular computer to do them after all.

In 1994, physicist Peter Shore developed a quantum computing algorithm for factoring large numbers that can be used to easily break the most common encryption methods currently used by, for example, the world's banks. A large enough quantum computer that corrects its own errors could practically run this algorithm, but mathematicians have still not been able to rigorously prove that classical computers will never be able to factor large numbers as efficiently. The most famous claims about quantum supremacy also fall into this category – and some of them were eventually defeated on classic cars. The quantum supremacy demonstrations that still exist also do not seem useful yet and are primarily intended to demonstrate the quantumness of the computer that completed them.

On the opposite side of the spectrum are problems in the mathematical field of “query complexity”, where the superiority of the quantum approach has been rigorously proven, but there are no associated algorithms that would be practical to implement or do anything uniquely useful. Recent experiment also introduced the idea of ​​”quantum information supremacy”, where a quantum computer solved a problem using fewer qubits than the number of bits needed to solve the same problem on a classical computer. Here, the resource that a quantum computer needed less was not time, but rather the number of physical building blocks. This may seem promising as it implies that a quantum computer can do something without having to make it huge first, but I wouldn't advise you to buy one for one simple reason – the task in question again had no obvious application in the real world.

Of course, there are real-world problems that seem like good solutions for quantum computer algorithms, such as determining the properties of molecules that are important in agriculture and medicine, or solving logistics problems such as flight planning. But I have to say “it seems” because the truth is that researchers don't have all the details yet.

For example, in recent study about the possible use of quantum computing in genomics, Aurora Maurizio at the San Raffaele Scientific Institute in Italy and Guglielmo Mazzola from the University of Zurich in Switzerland wrote that traditional computing methods are so good that “quantum computing may only provide speedups in the near future for a certain subset of fairly complex problems.” The idea behind their research is that while at first glance combinatorial problems in genomics seem like an area where a quantum computer could speed things up, a closer look reveals that their use must be very targeted and careful.

The truth is that for many problems not specifically designed to prove quantum supremacy, even when quantum computers can overcome noise and all other technical problems and execute algorithms faster than classical computers, there is a spectrum of what “faster” means. Because this doesn't always mean exponential speedup, the time savings that a quantum computer can bring don't always completely balance out the hardware costs. For example, a computer scientist Love GroverThe search algorithm, which is the second best-known quantum computing algorithm after Shor's, only provides a quadratic speedup—it reduces computation time by the square root rather than exponentially. Ultimately, how much faster is fast enough to justify the move to a quantum computer can be decided by each individual quantum computer buyer.

And I know, I know, it's a nasty phrase to put in a so-called buyer's guide, but if I've learned anything about quantum computers from talking to experts, it's that we don't know much more about what quantum computers can do than we know for sure. Quantum computers are an expensive and complex technology of the future, and we barely understand how they could add value to our lives beyond just adding value to the shareholders of some companies. As unsatisfactory as this is, I think it's an indication of how different and new quantum computers really are; how truly advanced computing technologies they are.

But if you're reading this because you have a decent amount of pocket money to spare on the biggest, most reliable quantum computer you can find, please get one and let your local quantum algorithm nerds tinker with it. In a few years they could probably give you much better advice.