Generative models such as diffusion models are one of the most important developments in the field of machine learning (ML). Programs such as Stable Diffusion and Dall.e have recently revolutionized the field of image generation. These models are able to generate high-quality images based on a text description. "Our new model for programming quantum computers does the same, but instead of generating images, it generates quantum circuits based on a description of the quantum operation to be performed," explains Gorka Muņoz-Gil from the Institute of Theoretical Physics at the University of Innsbruck.
In order to generate a certain quantum state or execute an algorithm on a quantum computer, you need the appropriate sequence of quantum gates to perform such operations. While this is relatively simple in classical computing, it is a major challenge in quantum computing due to the peculiarities of the quantum world. Recently, many scientists have proposed methods for the development of quantum circuits that rely on machine learning. However, training these ML models is often very difficult as the quantum circuits need to be simulated while the machine is learning. With diffusion models, such problems are avoided due to the way they are trained. "This is a huge advantage," says Gorka Muņoz-Gil, who developed the new method together with Wittgenstein Prize winner Hans J. Briegel and Florian Fürrutter. "In addition, we show that these diffusion models are accurate in their results and also very flexible, allowing circuits to be built with different numbers of qubits and different types and quantities of quantum gates." The models can also be adapted to create circuits that take into account how the quantum hardware is interconnected, i.e. how the qubits in the quantum computer are connected to each other. "As the production of new circuits is very inexpensive once the model has been trained, it can also be used to gain new insights into quantum operations," says Gorka Muņoz-Gil, citing a further potential of the new method.
The method developed at the University of Innsbruck generates quantum circuits based on user input and tailored to the properties of the quantum hardware on which the operations are to be performed. This is an important step towards exploiting the full potential of quantum computers. The work has now been published in the journal Nature Machine Intelligence and was financially supported by the Austrian Science Fund FWF and the European Union, among others.
Publication: Quantum circuit synthesis with diffusion models. Florian Fürrutter, Gorka Muņoz-Gil, and Hans J. Briegel. Nature Machine Intelligence 2024 DOI: 10.1038/s42256’024 -00831-9 [arXiv: 2311.02041 ]
