A method for preparing a specialized quantum resource has reduced its errors about 40-fold on Google's Willow processor. Google Quantum AI reported the result on October 15, describing a hardware demonstration of magic state cultivation.
The name sounds less practical than the problem. Error-corrected quantum computers need more than ordinary protected operations to execute a universal set of computations. Specially prepared states supply a resource for some of those additional operations.
Preparing enough clean versions of these states can consume substantial hardware. Cultivation is a protocol for improving their quality, rather than an application that solves a particular commercial problem.
Google reports a logical fidelity of 99.99 percent for the resulting T-states, one kind of magic state. Fidelity measures how closely the prepared state matches the intended state. The reported 40-fold error reduction is relative to uncultivated states in this experiment.
Those figures should not be read as a 40-fold speed increase for the computer, or 99.99 percent reliability for every program. They concern a specific resource-preparation task and its measurement.
The team also demonstrated transfer between two error-correction approaches, moving states from a distance-three color code to a distance-five surface code. That tests whether the prepared resource can survive integration with another part of the architecture.
Google says larger-scale demonstrations and further engineering remain necessary. The result turns a proposed preparation technique into a hardware experiment. A practical universal quantum computer requires more than this one component.
