Prof. Marcello Restelli of Politecnico di Milano’s Dipartimento di Elettronica, Informazione e Bioingegneria is among the authors of a study published on Nature Communications which describes a quantum compiler capable of programming, thanks to artificial intelligence, an algorithm on any quantum computer based on logic gates. The study is the result of the collaboration with Enrico Prati of the CNR - Istituto di Fotonica e Nanotecnologie and Matteo Paris of Università Statale di Milano and stems from a research conducted by Lorenzo Moro, PhD student at Politecnico di Milano.

Similarly to conventional computers, where the bits are subjected to calculations through logic gates, in quantum computers, capable of solving computing problems much more quickly, it is necessary to use quantum logic gates, which however must be programmed by a sort of operative system that knows which operations can be carried out. Quantum computer hardware is complex and offers only some fundamental logic gates, while an algorithm may require operations based on logic gates that are not among those available. There is a theory – the Solovay-Kitaev theorem – which states that there is always a way to build a quantum logic gate starting from those actually implementable on a specific hardware, but which does not explain how to do it.

The challenge was taken up by training an artificial intelligence to build, in a few milliseconds, a sequence for each quantum logic gate starting from the available operations. The research has also been patented since the model developed by the Italian researchers surpasses a similar patent registered by Google.

While the latter requires new training for each logic gate, the compiler developed by Moro, Paris, Prati and Restelli, thanks to deep learning, is able to build all quantum logic gates with a unique training, after which the solution for any logic gate can be immediately recalled.