A team of physicists at the University of Vienna have created a new quantum device called a quantum memristor, which could combine the worlds of artificial intelligence (AI) and quantum technology. The experiment, which was carried out alongside the National Research Council (CNR) and the Politecnico di Milano in Italy, was realized on an integrated quantum processor that operates on single photons.
The research was published in the journal Nature Photonics.
Discovering the Memristor
Artificial intelligence applications rely on neural networks inspired by the biological structure of the human brain, and they are mathematically trained until they can carry out human tasks like facial recognition, driving, or interpreting images. One of the major research focus areas in these fields, both in academia and industry, is the integration of devices that can perform computations quickly and efficiently.
Back in 2008, there was a big leap forward in the field with the discovery of the memristor, which is a device that changes its resistance depending on a memory of the past current. Following the discovery of the memristor, scientists learned that the behavior of memristors was similar to neural synapses. This has caused the memristor to become a major part of neuromorphic architectures.
Engineering the Quantum Memristor
The group of experimental physicists was led by Prof. Phillip Walther and Dr. Roberto Osellame, and they demonstrated how a device with the same behavior as a memristor could be engineered. This device also acts on quantum states and can encode and transmit quantum information, leading to it being termed a “quantum memristor.” It is extremely difficult to create such a device given the dynamics of a memristor, which often contradict typical quantum behavior.
The team relied on single photons and their ability to propagate simultaneously in a superposition of two or more paths. This is what dramatically helped the team realize such a device.
They carried out an experiment where single photons were propagated along waveguides laser-written on a glass substrate. These single photons were guided on a superposition of several paths, and one of these paths was used to measure the flux of photons moving through the device. The quantity of these photons modulates the transmission on the other output, which achieves the desired behavior similar to a memristor.
The researchers were also able to carry out simulations demonstrating that optical networks with quantum memristors can be used to learn classical and quantum tasks. This is what led the team to believe that the quantum memristor is what’s needed to merge the fields of AI and quantum computing.
Michele Spagnolo is first author of the research.
“Unlocking the full potential of quantum resources within artificial intelligence is one of the greatest challenges of the current research in quantum physics and computer science,” Spagnolo said.