What’ll it take to solve the quantum computing challenges of the future? Microsoft has an app for that — and now developers around the world can have it, too.
The app is called the Azure Quantum Resource Estimator. It’s a software tool that was originally developed for Microsoft’s internal use. The tool is already guiding the company’s effort to develop full-stack quantum computers, and now it can also help outside developers figure out how much computing power they’ll need to execute a given quantum algorithm in a reasonable amount of time.
That’s a key question, because the guidelines used for classical computing don’t necessarily apply to the quantum frontier. Unlike classical computers, quantum computers take advantage of an environment where a quantum bit — better known as a qubit — can represent a one and a zero at the same time.
Quantum approaches can be far more efficient than the standard binary computing approach for solving particular kinds of problems: optimizing a network, for example, or figuring out how to design a synthetic molecule to perform a specific chemical task.
“We’ll be able to study, for example, how to help remove harmful gases from the atmosphere,” Krysta Svore, distinguished engineer and vice president of quantum software at Microsoft, told GeekWire.
“Ten years ago, we thought it would take a billion years’ run time on a quantum computer,” Svore said. “That’s a really long time to wait. But over the last decade, we’ve been able to bring that down to a month’s run time on a quantum computer … using exactly the resource estimator, this tool, to understand the cost of the algorithm. And we’ve been able to redesign our hardware accordingly as well.”
There’s a bit of a catch: The kind of quantum computer that the resource estimator uses as its baseline doesn’t exist yet. “What we’ve found is that these quantum machines to run problems that we identify as having practical quantum advantage will need at least a million qubits,” Svore said.
Just last week, IBM unveiled its largest quantum processor, which knits together a mere 433 qubits. IBM is aiming to scale up its systems to more than 4,000 qubits by 2025, and D-Wave Systems is planning to roll out a 7,000-qubit annealing quantum computer in the 2023-2024 time frame. Even those machines will fall far short of the power that Svore and her colleagues at Microsoft have in mind.
“Reaching a quantum machine that has upwards of a million physical qubits is measured in years,” Svore acknowledged. But she pointed out that it’ll also take years to gain a deep understanding of the applications for quantum computing. “So we do need to get ready,” she said.
That’s where the resource estimator will come in handy, particularly in cases where developers blend classical and quantum approaches to come up with hybrid methods to solve problems.
“This is a great tool to understand hybrid,” Svore said. “I have classical compute and quantum compute coming together. What is the cost of each? The quantum compute needs to be used when it enables a speed-up over using classical. So you want to compare that classical-plus-classical against classical-plus-quantum. This is a tool to enable that type of study.”
The estimator tells you roughly how much processing time would be required to execute a given algorithm in different computational scenarios, depending on the number of qubits, the type of error correction scheme and other parameters.
Svore said the estimator could show software developers how making tweaks in their quantum algorithms could lead to faster run times. “At Microsoft, we’ve also been using the tool to develop the underlying architecture of the machine, to understand what kind of machine will even enable these algorithms,” she said. The tool supports Microsoft’s view that a topological-based quantum machine “will enable the scaling up that’s required,” Svore said.
The first steps in the process of using the estimator involve setting up an Azure account and creating an Azure Quantum workspace. Then you can follow the procedure outlined in this introduction to quantum resource estimation.
Michal Stechly, a quantum software engineer at Zapata Computing, said In a Microsoft blog posting that the estimator is “easy to use.”
“The integration process was simple, and the results give both a high-level overview helpful for people new to error correction, as well as a detailed breakdown for experts,” Stechly said. “Resource estimation should be a part of the pipeline for anyone working on fault-tolerant quantum algorithms.”
Back in 1964, Nobel-winning physicist Richard Feynman famously declared, “I think I can safely say that nobody understands quantum mechanics.” The fact that there’s now an “easy-to-use” tool for quantum computing just might be an indication of how far things have come since then.
