Dr. Joe Fitzsimons, a physicist academic turned startup entrepreneur, believes that developing software using quantum computing is equivalent to programming computers in the 1940s or developing for Arpanet in the 60s. In other words, it’s still early. But his company, Horizon Quantum Computing, is about to launch a software development platform that will, he says, enable developers who are proficient in languages like C and C++ to build quantum computing applications.
As I’ve noted in previous articles, quantum computing is currently very reliant on hardware gains being eked out by the likes of IBM and Google. Indeed, just this week IBM announced a new 433-quantum bit (qubit) processor, Osprey, more than tripling the 127 qubits on last year’s IBM Eagle processor. Despite these impressive gains, we’re still a ways off the 1,000 qubit processor that IBM and others in the industry are targeting. That will be the so-called “quantum advantage” level, when quantum computing will outperform classical computing in certain use cases.
Fitzsimons acknowledges that hardware improvement is still the main priority in quantum computing. “The current state of quantum computing is such that we don’t yet have hardware that is capable of showing a real advantage for real-world problems,” he said. However, he noted that the hardware is improving every year and that his company is doing its bit by building a software platform to take advantage of those gains.
“You need the software tools, you need the development tools,” he insisted. “Even just the kind of intellectual framework in terms of how to think about harnessing quantum effects — effects of superposition, effects of interference — to gain an advantage on real-world problems. And it’s not at all obvious how you would go about doing that, [so] we’re really focused on trying to get to a point where that becomes straightforward.”
An Abstraction Platform
What Horizon Quantum Computing is attempting to do is create a framework that allows programmers to use the classical programming languages they’re familiar with — Fitzsimons mentioned C and C++ — and the platform will compile that code into hardware instructions for the quantum computers. But, according to Fitzsimons, it’s not just about compiling code. Their platform will also help developers optimize their approach to the problem they are trying to solve, in order to take advantage of the potentially massive speed gains on a quantum machine.
“So what we’re trying to do is bridge between a problem and a solution accelerated on quantum hardware,” he explained. “We’re saying, give us a very concrete definition of your problem in the form of conventional code that would solve the problem, if only your computers were fast enough to run it.”
This optimization work involves “looking at how you can pull apart the structure of the program, refactor it, reconfigure it in such a way that you get to components that are amenable to quantum speed,” he added.
The platform, which is a work-in-progress and has not yet been launched, will also allow developers to specify how “low-level” they want to go. There will be layers of abstraction, said Fitzsimons.
“Essentially, we’ve been building up abstraction layers and you can enter it at varying levels of abstraction depending on your familiarity with quantum computing and the level of detail you want to go into — how low a level you want to interact with the hardware .”
Horizon vs. Classiq
Horizon isn’t the only company attempting to create a quantum computing software platform. Earlier this year, I profiled a company called Classiq that is working on a similar solution. Its co-founder and CEO, Nir Minerbi, told me in July that Classiq has created a high-level functional model that can be translated into quantum assembly language. In that system, a developer models the circuit design, via Python or VSCode, and Classiq’s platform turns that into actual circuit code. It can then be run on one of the leading quantum computing processing services, like Qiskit or Amazon Braket.
I asked Fitzsimons how his platform differs from Classiq’s. He replied that what Classiq does is “circuit templating,” which involves “a lot of reused components.” He says his company is “going beyond that.”
“We’re completely refactoring from an entirely classical program,” he said, “figuring out a quantum algorithm from scratch — not using a pre-baked one.”
Horizon’s approach sounds ambitious, but it’s important to note that it hasn’t yet been launched — whereas Classiq is already a production-ready solution. Fitzsimons describes his company as “pre-launch,” adding that it has “some R&D contracts where there is some level of access.” Horizon aims to reveal its IDE [integrated development environment] in a preview early next year.
Provided the launch sometime next year is successful, I asked Fitzsimons what kinds of use cases he hopes the developers who sign up will tackle using quantum computing?
“The most obvious areas are the industries which depend on high-performance computing, or are bottlenecked by compute capacity,” he replied. He noted that other companies in the space have focused on finance, solving optimization problems, or problems in chemistry in the pharma industry as early use cases (and indeed, all of those were mentioned as use cases in my interview with IonQ and GE Research back in July). But he thinks there are other opportunities.
“It’s also the case that there are a large number of other industries for which there are not necessarily currently well-studied algorithms, but where the problems have the right flavor to be amenable to quantum speed-up,” he said. He listed the energy sector, computational geophysics, and aerospace and automotive as areas in which “you would expect to see a large quantum speed-up.”
Problems such as computational fluid dynamics, common in aerospace or car design, are “hard on conventional computers,” said Fitzsimons, adding that it’s “why wind tunnels still exist.” His hope is that quantum computers can tackle some of these problems in the years ahead.
To the Future
It’s too early to tell how effective Horizon’s abstraction platform will be, since it isn’t yet generally available to use. Also, just like computing in the 1940s, there will likely be a lot of trial and error among pioneering companies like Horizon and Classiq before a standard type of SDK for quantum computing arises. That said, 2023 is shaping up to be the turning point for quantum computing software development, with significant gains in both hardware and software likely.