Jul 18, 2023 2:00 AM

Where quantum computing is already delivering value

While gate model quantum computing holds immense promise for tomorrow, quantum annealing systems are solving complex optimization problems for enterprises today.

sakkmesterke/Shutterstock

You’ve heard that quantum computing could represent a step change from current, classical computing. Well, you don’t have to wait. Annealing quantum computers have already demonstrated that, for certain workloads, they’re magnitudes faster than classical computers alone, and will soon be capable of previously unimaginable calculations. Even better, they work synergistically with classical computers, so quantum-hybrid applications allow for the best of both worlds.

But first, let’s discuss what can be done with quantum computing in the enterprise today. To understand where quantum is delivering value today and why, you need to distinguish between the two leading quantum computing models: gate model and quantum annealing.

By definition, both quantum approaches rely on qubits—that is, bits that possess the quantum trait of superposition, which means they can represent a combination of 1 and 0 rather than just the on-or-off binary state of classical bits. This superposition of states, along with the quantum mechanical phenomenon of entanglement, enables quantum computers to manipulate enormous combinations of states at once. Qubits can be made up of different technologies including superconducting, ion traps, photonics, and more. 

In gate model, the logic gates of classical computers are replaced with quantum gates, which, when properly programmed at a machine level, manipulate qubits to yield computational results. Annealing model quantum computers, by contrast, can be programmed at a much higher level to manipulate qubits in the service of solving real-world optimization problems.

In both cases, a classical computer is required to control the quantum computer. All approaches to quantum computing require significant engineering and careful environmental control of the QPU (quantum processing unit). Fortunately, both gate model and annealing quantum computers make their functionality available through the cloud. Of the two quantum computing systems, quantum annealing is miles ahead in delivering practical value to enterprises.

The quantum annealing edge

Classical computers have become so powerful and versatile, it’s hard to imagine what they can’t do well. But they do have limits. Choosing the best solution among many possible solutions can sometimes be a problem that is too large for today’s classical systems to provide an answer in a realistic time frame. This is where annealing quantum computers shine.

Annealing has proven to be over three million times faster than classical methods for a certain type of quantum simulation. For a class of classical optimization problems called 3D spin glasses, annealing was recently shown to improve solution quality faster than today’s classical computers. So, what does this mean for the real world? One company has used the technology to develop a quantum hybrid application that looked at 67 million different scenarios and provided an answer back in approximately 13 seconds!

The annealing approach derives from quantum physics itself. The underlying principle is simple: If treated gently, physical systems tend to want to remain in their lowest energy configuration. That’s the basis of quantum annealing—a paradigm on top of which you can run optimization workloads that pick the “lowest energy” solution, from the most efficient delivery route to the financial portfolio with the lowest risk.

With quantum annealing today, these real-world optimization problems are solved in hybrid fashion—that is, they combine classical and quantum computing capabilities. This is great for developers. They can simply write applications in Python, for example, and use a quantum software development kit to tap into the power of quantum annealing.

When a developer accesses quantum-classical hybrid solvers through the cloud, they don’t have to address that quantum annealing system directly. Instead they can rely on a front line of classical computing that shunts the appropriate portions of the workload to the annealing quantum computer behind the scenes. Choosing the optimum solution is the job of the annealing quantum computer. And it can do so in much less time and with better results than a classical computer could accomplish alone.

Another annealing advantage is error correction—as in, quantum annealing doesn’t need it. This may sound strange, because all quantum computing is susceptible to noise. When noise occurs on an annealing quantum computer, however, the quantum state can ultimately reemerge and the process of determining the optimal solution can proceed until completion.

Neither the developer nor the user is exposed to such machinations because the classical computer they interact with addresses that complexity. Moreover, even as the hybrid model enables you to extend quantum computing capacity, parts of many of the problems will be best handled by classical computers. The future of quantum computing is hybrid. As quantum systems advance, we’ll continue to see expansion in the types of problems that can be solved.

Annealing quantum systems are available now and will likely always be best for addressing optimization problems. Optimization spans a wide variety of problem sets seen by private and public sectors alike.  

The gate to the quantum future

When you hear people talk about quantum computers one day replacing our classical, binary friends, that prediction originates with the gate model quantum computer. After all, the original concept was to replace conventional bits with qubits and conventional gates with quantum gates. Just write all manner of applications for that new platform and, hello, welcome to a brave new world of computing.

But there are a few problems with that notion, beginning with the heavy burden developers must bear. The software development kits designed for the gate model require developers to learn the equivalent of assembler for QPUs, which involves some very advanced math. The bottom line is that to fully understand the algorithms that can be created for gate model quantum computers, developers must gain a solid working knowledge of quantum physics and learn to speak a whole new computing language, so to speak.

Moreover, the errors in gate model systems make them unable to remain in a quantum state long enough to solve real-world problems. As a result, users of gate model quantum computers today are mainly academics rather than industry. They are using gate model quantum for experimentation in quantum chemistry, differential equations for fluid flow dynamics, and other areas where classical computers tend to hit the wall. In highly competitive research domains such as these, it’s worth the time and money to train or hire specialized quantum developers with an eye toward the future.

But enterprises need to realize that it’s early days for gate model quantum computing, and that gate model systems may never be better than annealing systems in solving optimization problems. Unlike annealing quantum computers, gate model quantum computers require error correction—the biggest single engineering challenge for quantum computing. In the gate model, information is put into a quantum state. If that state collapses, a quantum system should be able to correct the error and roll back to where it left off. But the ability to do that at scale has not yet been achieved.

That’s why, for the time being, some gate model systems have foregone error correction, earning them the designation Noisy Intermediate Scale Quantum (NISQ) computers. There is no evidence to suggest that you can support commercial applications with a NISQ computer. At D-Wave, our best estimate is that gate model quantum computers with reliable error correction are at least seven years away.

A partnership of computing paradigms

The hype about quantum computing replacing classical computing is simply incorrect. Quantum and classical computing will work side by side for the foreseeable future. At the same time, the lament that quantum is stuck in the lab fails to recognize the value that annealing quantum computers are delivering today.

Some say that annealing quantum computers are “limited” to optimization applications. But when you think about it, what endeavor is more urgent across organizations than getting the best possible return on the investment of resources? At D-Wave, we see that today in such areas as financial portfolio management, protein design problems, traffic routing, customer offer allocation, airport or hospital personnel scheduling, missile defense, electrical grid resilience, and space exploration, just to name a few. 

By the end of this decade or the beginning of the next, the error correction and programming difficulties of gate model quantum computing may be addressed, opening up an even wider range of application. But there is no need to wait on applying quantum to the enterprise. More and more enterprises are discovering the value that quantum annealing is already delivering today—not only practical optimization benefit, but also valuable experience in the quantum domain.

Nearly all organizations can achieve benefits from quantum-classical hybrid technology now, backed by today’s quantum annealing systems. At the same time, they will be preparing for our inevitable quantum future.

Murray Thom is VP of quantum business innovation at D-Wave.

New Tech Forum provides a venue to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to newtechforum@infoworld.com.