Harnessing Generative AI for Quantum Computing Programming

Researchers in Austria have made a groundbreaking advancement in quantum computing by utilizing generative AI to program a quantum computer, with the goal of overcoming a major bottleneck in the technology.

To prepare a specific quantum state or execute an algorithm on a quantum computer, the precise sequence of quantum gates must be determined for the operations. While this process is relatively straightforward in classical computing, it poses a significant challenge in quantum computing due to the unique characteristics of the quantum world.

The innovative AI method developed at the University of Innsbruck is capable of generating quantum circuits based on user specifications and tailored to the specific features of the quantum hardware on which the circuit will be executed.

“Our new model for programming quantum computers generates quantum circuits based on the textual description of the quantum operation to be carried out,” explained Gorka Muñoz-Gil from the Department of Theoretical Physics at the University of Innsbruck, Austria.

There are various methods for constructing quantum circuits using machine learning, which is why many supercomputer centers are integrating AI infrastructure with quantum computers. However, the training of these machine learning models can be quite challenging due to the need to simulate quantum circuits as the machine learns.

Generative AI models, on the other hand, circumvent such difficulties in training. “This offers a significant advantage,” noted Muñoz-Gil, who collaborated on the development of the method with Hans Briegel and Florian Fürrutter.

“We have demonstrated that denoising diffusion models are not only accurate in their generation but also highly flexible, enabling the generation of circuits with varying numbers of qubits, as well as different types and quantities of quantum gates,” Muñoz-Gil added.

The models created at the University of Innsbruck can also be customized to design circuits that consider the connectivity of the quantum hardware, such as how qubits are interconnected within the quantum computer.

“Since generating new circuits is cost-effective once the model is trained, it can be used to uncover new insights about quantum operations of interest,” Muñoz-Gil emphasized.

This breakthrough represents a significant leap forward in unlocking the full potential of quantum computing. The research has been published in Nature Machine Intelligence and received financial support from the Austrian Science Fund FWF, as well as the European Union, among others.

The paper titled Quantum Circuit Synthesis with Diffusion Models can be accessed in Nature Machine Intelligence at https://doi.org/10.1038/s42256-024-00831-9.

www.uibk.ac.at