IBM Roundtable: Accelerating the Journey to Quantum App Development

By Larry Greenemeier

Perhaps more than any technology before it, quantum computing will create a deep disparity between first movers and fast followers. That was the assessment a panel of academics, entrepreneurs and quantum computing experts at the July 9 virtual roundtable, “The Future of Quantum Software Development.”

The group discussed the need for intuitive application development tools, when quantum advantage might arrive and what it will look like when it does.

Computing will change more in the next 10 years than it has in the last 100 years, said panelist William "whurley" Hurley, CEO and founder of Strangeworks. “Quantum is the thing I’m placing my bet on being the exotic computing technology that brings about that change,” added whurley, whose company offers a hardware-agnostic, software inclusive, collaborative quantum development environment. “If you want to be involved in it you should be preparing now.”

A roundtable on the Future of Quantum Software Development included, clockwise from top left,
Jeffrey Hammond, William "whurley" Hurley, Prineha Narang and IBM Quantum's Blake Johnson

For quantum computing to succeed at the enterprise level, application development tools must have a low barrier to entry, the panelists agreed. With its open-source Qiskit quantum software development kit, IBM is building tools that appeal to programmers as well as quantum computing domain experts. Catering to programmers’ existing strengths is crucial because “we don’t want developers to think they have to learn everything about quantum computing just to get started,” said panelist Dr. Blake Johnson, Control Systems Delivery Lead, IBM Quantum.

IBM’s new Qiskit Optimization Module, which debuted on July 9, is an effort to introduce quantum computing to enterprises in a context they already understand. The module helps developers write programs by first describing the problem they want to solve—in this case some sort of optimization problem—without worrying about a quantum computer’s underlying complexity. “They can write down a description of some of the constraints of the optimization problem and can choose different solvers—both quantum and classical,” Johnson said.

Simplifying the quantum programming process is crucial, because many companies are evaluating whether now is the right time to invest the time and effort required to better understand how they can benefit from these systems. “With the new module, based on the solver type, programmers send their workloads to the IBM Cloud and then we distribute it to the appropriate quantum or classical hardware, and deliver the result back,” Johnson said. “This is the first time you can use a gate-based quantum computer, where you program it just at the level of the problem interface, as opposed to at the circuit interface.”

Whurley put companies’ quantum developer dilemma this way: “If a quantum computer is a million times faster at a task that has a minor impact on return on investment, and a classical system costs a million times less and will take longer but still get the job done, enterprises will still default to that classical system.” The adoption of quantum computing relies on identifying where such devices provide businesses with an advantage—the quantum advantage—and helping programmers easily navigate the application development process. “If we don’t [do that], we won’t have the adoption we need to reach any kind of critical impact,” whurley added.

The Road to Quantum Advantage

Top of mind for the panelists and moderator Jeffrey Hammond, Forrester’s Vice President, Principal Analyst Serving CIO Professionals, was when quantum advantage will be achieved, what that will look like and the role that software development will play. Quantum advantage will be the point at which a quantum computer can perform a computation significantly faster than a classical computer.

The consensus among panelists was that, although we are years away from seeing an undisputable example of quantum advantage, it will happen well within the next decade. Certain disciplines such as material science and financial optimization are most likely to achieve quantum advantage first because the problems in those areas play to quantum computing’s strengths.

Based on research being produced by IBM Research and elsewhere, “we know that near-term devices will show us a quantum advantage,” said panelist Dr. Prineha Narang, CTO and co-founder, Aliro Technologies, which makes software that facilitates user access to present day quantum computers. “There are problems people have been trying to solve for decades—'holy grail’ problems in condensed matter and in chemistry—that are now actually within reach.”

There are a number of factors to consider when calculating the trajectory to quantum advantage. For starters, “researchers don’t know how powerful a quantum computer has to be to achieve quantum advantage,” Johnson said. Still, Johnson added he would be surprised if we have to wait until fault-tolerant quantum computers are developed before researchers can achieve some form of quantum advantage over classical computers, whether that advantage is measured in time, cost or some other metric.

IBM’s commitment to achieving quantum advantage includes doubling quantum volume (QV) every year. QV is a hardware-agnostic metric that IBM defined to take into account the number of qubits, connectivity, as well as gate and measurement errors. IBM now hosts eight quantum computing systems that cross the QV32 performance threshold available to IBM Q Network organizations.

“The most critical thing that’s keeping us from reaching quantum advantage is the call to the hardware,” Johnson said. As such, it is incumbent upon quantum computer makers to create development tools that do not require programmers to know the intricacies of how these systems operate before those programmers can start writing meaningful code that pushes the hardware’s capabilities in search of solutions to real-world problems.

“People are already doing applications research with [quantum computing] devices that exist today and are finding they maybe can’t yet solve a problem better than on a classical resource, but they can solve problems,” Johnson said. “And they’re starting to figure out what the limitations are, how to squeeze out the most utility from the devices today and what they’ll need to be ready for the devices that will exist tomorrow.”

Universal Quantum Computers

Panelists also discussed whether different types of quantum computers will be optimized to solve different problems. “I suspect that there will be certain problems that will work just fine on a variety of hardware and some that will be more specialized to different types of hardware—that’s just associated with the physics underlying that hardware,” said Narang, who is also Assistant Professor at the John A. Paulson School of Engineering and Applied Sciences at Harvard University. “This will be especially important as we try to map problems from condensed matter and chemistry onto some of these devices. We’ll see that not everything is ideal or even possible for all kinds of problems.”

When fault tolerance on quantum computers is achieved, developers will be able to execute operations on error-corrected qubits, Johnson said. Although error correction comes with a big cost in terms of overhead—such as the control system needed to correct errors as a system is executing—most error correction also comes with an enormous advantage. “The virtual connections between error connected qubits are often all-to-all—I can move qubits around on this logical lattice of qubits for almost no additional cost above what I do for error correction,” Johnson added. “When we get to that regime, we’ll become a lot less focused on the need to specialize the hardware for this application or that application.”

Until then, programmers have to be judicious about the system resources their quantum programs use. “I have to count every instruction,” Johnson said. “That forces the programmer to fit into very narrow boxes in order to get something to work. But as these systems become more capable and the abstraction layers get to the point where I don’t worry about their cost, that will change the world a lot.”

→ Read more about the new Qiskit Optimization Module on the IBM Research Blog.