By Morgan Adkins
Computers have become an integral part of our society today. From the daily routine of the workplace to the technology behind the car, everywhere we go and everything we do seems to involve a computer. And, as our dependency on technology continues to expand, our need for more innovation only grows. While our computing power continues to increase according to Moore’s Law, doubling approximately every 2 years, this trend cannot continue without our computer technology reaching the atomic scale. This is where quantum mechanics comes onto the scene.
Imagine a computer that can teach your phone to recognize any object it sees, or one that can instantly find the ideal route for thousands of planes to avoid hazardous weather, or one that can search through millions of social media posts to identify a potential terrorist. Classical computers can’t crunch that big data without being given large amounts of time. But scientists have long theorized that a computer that could harness the principles of quantum mechanics would be able to perform these kinds of calculations quickly and efficiently, and even solve problems that would take years for a normal computer to churn through.
Mark Saffman, a professor of physics at the UW-Madison, is researching this intersection between computing and the quantum realm. Whereas digital computers require data to be encoded into binary digits (bits), each of which is always in one of two definite states (0 or 1), quantum computers use quantum bits (qubits) for data storage, which can be a superposition of these states. This means the qubit can be 0 and 1 at the same time. With this fundamental difference, by the time you get to few hundred qubits, the difference in computing power is enormous between parallel computations in a quantum computer compared to its classical counterpart.
While the concept of quantum computing is not a new notion, this is an exciting time for the technology. There has recently been a paradigm shift in the field as large companies are beginning to invest in quantum computing, including powerhouses like Microsoft, Google, Lockheed, and others. Saffman suggests that these big companies are investing because they are major players in computing, and they want to stay abreast of what might be an important development. So just how close are we to having a quantum computer? “It’s all speculative at this point,” Saffman says. “You can ask top researchers when there will be a practical quantum computer, and it all depends what you mean by practical.”
This is, in large part, because quantum computers are very specialized. They aren’t going to replace our classical computers; it’s simply not going to happen. Simple tasks like using Word or checking email aren’t going to be enhanced by using a quantum computer. However, there are certain problems we face today that would be much faster to solve using such a machine. “One of the applications of quantum computing that looks not that far away is something called quantum chemistry,” Saffman says. This involves looking at molecules containing some number of atoms, understanding their internal structure, the energy levels, how they interact with each other. “This is something that chemists do all the time,” Saffman explains. “They calculate things about molecules, but there’s a limit to how large of a molecule they can analyze. The [classical] computers just don’t have enough oomph. They need quantum computing to do that.” Detailed studies have been conducted about the possibility of using a quantum computer to analyze molecules. If the quantum computer had just a few hundred bits, the prediction is that it could begin to solve problems that chemists cannot solve with the technology available today.
The issue here, though, is fidelity: how reliable do those bits have to be? Saffman clarifies, “The problem is that it’s not just having a couple hundred bits, but those have to be really well performing bits that are corrected against errors.” In a classical computer, there is hardware that corrects for various faults. There is always a correction going on in the background. According to Saffman, it will also be necessary that quantum computers contain this extra correction hardware. However, the error correction in the quantum computer is actually much tougher than in a classical computer. Roughly speaking, in the classical computer, the addition of one or two bits per byte lets you check and correct errors. “It’s sort of a 10 percent or 20 percent overhead in a sense,” Saffman points out. In a quantum computer, though, the overhead, or additional qubits needed for error correction, is more like a factor of 10 or a factor of 100. In order to have one really well-performing, reliable, non-error quantum bit or logical quantum bit, you might need 10 or 100 physical qubits. Thus, in order to solve these quantum chemistry problems that require approximately 100 bits, over 1000 qubits might be necessary in order for there to be the 100 logical quantum bits that are needed. “But, if researchers can get to that size,” Saffman says, “then they could have something really quite revolutionary in computing.”
In an effort to create a quantum computer of this magnitude, researchers are developing the machines using a handful of different approaches. The leading approaches use superconductors, semi-conductors, diamond, neutral atoms, charged particles, or photons as the platform of the computer. Each approach has its own pros and cons. The superconductors function with 99.9 percent reliability, as do the charged particles known as “trapped ions”, but it’s not clear at this point in development how to scale up the machines. The semiconducting approach isn’t working extremely well in terms of reliability or scaling, either. Using light — photons — also has a great deal of scaling challenges. The difficulty with scaling in these approaches is due to, in simplified terms, the fact that the interactions can’t be turned on and off. The ions have charges and they are always interacting, and because of that, it makes it difficult to scale up since the unwanted interactions would only increase. The superconductors, as mentioned above, also have other unresolved issues. In the superconducting approach, researchers have to send wires down to control each one of the qubits, and wires are problematic because while every wire sends the necessary electrical signals up and down, each wire also produces heat. This is a problem because the superconductors only operate in an environment merely 10 milliKelvin above absolute zero. There is also cross talk, or interactions, in the wiring that creates additional problems.
Professor Saffman is researching the neutral atom approach to quantum computing. “What’s special about the neutral atoms, which is a positive feature,” he says, “is that we can turn the interactions on and off.” This allows for the neutral atom approach to be scaled relatively easily, but the fidelity of the qubits is still an issue. Qubits are extremely tricky to manipulate, since the slightest disturbance causes them to fall out of their quantum state (or “decohere”). This decoherence is the Achilles heel of the neutral atom approach and quantum computing in general. “In terms of the number of quantum bits that we can actually measure and control, I actually think that here, in my lab, we have the largest that anyone has done. We have a 2D array of 49 qubit sites, so we have up to 49 qubits. And that’s more than anyone else has demonstrated,” Saffman says. The downside is that those qubits aren’t working that well. Right now, researchers at UW-Madison can entangle them, they can perform data operations, but their entanglement is only about 75 percent correct according to Saffman. Entanglement is an extremely strong correlation that exists between quantum particles that allows them to be inextricably linked in perfect unison, even if separated by great distances. This is what allows for the simultaneous calculation in the quantum computer. Some approaches have more control over this interaction, others have more fragile qubits; it’s all about finding a balance between coherence and control.
So what exactly is the next step in quantum computing? Saffman says, “A bit further along, we’ll see a combination of both industrial investment and government-funded research for the next generation of devices.” If you look at computing today, a mature and well established industry, companies are investing in how to make the next generation of computers perform better in various ways, while at the same time other government programs are looking at high performance computing or computing with fast communication rates to solve particular problems; there’s a mix of investment occurring. Saffman believes that will be the case with quantum computing, too. Today, though we are entering the phase of company investment, we are very far from having general-purpose quantum computers available commercially. Because the development of actual quantum computers is still in its infancy, for the foreseeable future, it’ll be primarily government funding, and at some point will transition over to being more of an even mixture of investments.
Here at UW-Madison, research in this field is continuing to bring about innovations. Professor Saffman’s team has a design for a 121-site qubit array that he expects will be put in place later this year. However, their main focus is getting the 75 percent fidelity up to 90 percent or higher. “Fidelity,” he says, “that’s really our main focus. There are some technical improvements we are about to put on our setup. We have ideas of what is causing the issues; now we need to go through the list, check them, and try to fix them.”
Quantum computing has shown sizeable growth nationally and internationally, but also specifically here at UW-Madison. “Fifteen years ago,” Saffman recounts, “there was basically no activity in quantum computing.” Now there are 5 different faculty members involved and more than 20 students working in different quantum computing-related research. UW-Madison is rather unique in that we have 3 different experimental approaches being studied here: neutral atoms, semi-conducting qubits or quantum dots, and super-conducting qubits. There is external funding of more than $5 million per year coming into the physics department. Quantum computing is not only a huge part of this university, but is also growing across the globe, and it’s only expected to get bigger — or is it smaller?