AI Software Tool Helping Quantum Computers to ‘Self-Tune’ for Improved Performance
With the overriding goal of reducing hardware errors and instability caused by environment noise, quantum computing software startup Q-CTRL has unveiled an AI-based toolset billed as enabling quantum computers to “self-tune” for noise and error suppression.
The tools use AI agents to execute algorithms with fewer errors, ultimately boosting the performance of quantum computers for potential enterprise users. The tools are accessible via emerging quantum cloud services.
Los Angeles-based Q-CTRL designs firmware for quantum computers and other quantum devices, employing a form of quantum control to reduce quantum computing errors. One approach involves “robust control” to redefine quantum logic operations used to compose quantum computing algorithms.
“In effect, we rewrite the machine language for the system to yield lower errors,” Michael Biercuk, Q-CTRL’s founder and CEO, noted in an email exchange.
Early demonstrations using IBM’s quantum cloud services platform showed the AI agent forged new quantum logic gates. The autonomously designed gates exhibited lower error rates than the best of those designed by IBM hardware engineers, translating into what the startup claims are more robust algorithms.
“Our AI agent obviates the need for either a mathematical model or human intervention,” Biercuk asserted. “The agent automates the process of learning whatever it decides is relevant to deliver higher-performing gates.”
Q-CTRL said Wednesday (Feb. 10) it is making the AI-based tool available as a new feature on its flagship software platform. Boulder Opal is a Python-based toolkit for developers and R&D teams using quantum control in their hardware or theoretical research.
The startup likens its AI agent to software abstraction in conventional computing whereby programmers write algorithms without getting bogged down in hardware details. As a result, Biercuk said, the AI agent “will accelerate the development of quantum computer hardware and applications” by enabling self-tuning machines delivering quantum logic with fewer errors.
Q-CTRL scientists previous demonstrated the ability to create quantum logic gates for individual qubits. They outperformed standard logic gates running on an IBM quantum computer by a factor of 10. In the lab, the AI agent autonomously identified new multiqubit logic gates that yielded a two-fold reduction in errors when compared with default gates, the startup claimed.
The AI agent also illustrates the growing infrastructure ecosystem coalescing around quantum computing development. Along with IBM (NYSE: IBM), Amazon Web Services (NASDAQ: AMZN) is promoting its own managed quantum computing service. Launched this past August, Amazon Braket provides development tools, simulators and access to different types of quantum hardware, “each with a different physical limitation,” said Simone Severini, director of quantum computing at AWS.
Braket makes it “possible to compare different technologies side by side, and to switch between them by changing only a line of code,” Severini noted in a blog post this week. “It’s not just about access. It’s about envisaging how quantum computing will one day fit into a cloud-based IT infrastructure, working together with other computational resources.”