Computers have made it easier to store, process, and transmit information. Quantum computing promises to do the same but with an exponentially large volume of information and the ability to perform calculations impossible for classical computers, which would allow things like designing personalized medicines and finding the optimal route for an autonomous car.
Google Quantum AI, Alphabet’s division dedicated to research in quantum computing, has designed an error code that allows information to be stored more securely in the cubits. Physical cubits have too many errors, but by putting many of them together, the information is encoded more securely in the group, creating what is known as a “logical cubit,” something like a fault-tolerant “supercubit.” The system is based on putting many physical cubits together so that redundancy causes error correction to act faster than the errors themselves are added.
The Google experiment used a superconducting quantum processor with 72 cubits and was tested with two different “surface codes”: one called the distance 5 logic cubit (in 49 physical cubits) and smaller ones called the distance 3 logic cubit (in 17 physical cubits). It was shown that the larger surface code (distance 5) enables better logic cubit performance than the smaller surface code.
However, the experiment has limitations: they were only able to store the information, not operate on it. Although bug fixes occur, the operations demonstrated by Google are not sufficient to perform a complete computation. The “topological” property of the code, its peculiar geometrical structure, and other special properties guarantee the conservation of quantum information, which means that if any cubit is destroyed, the information will not be corrupted because the particle that interacted with the cubit does not have access to it.