Learning and Generating Mixed States Prepared by Shallow Channel Circuits
Researchers have developed a method for learning and generating mixed quantum states within a specific phase of matter, known as the trivial phase. This phase is characterized by the existence of a shallow preparation channel circuit that preserves local reversibility. The new algorithm can efficiently learn these states from measurement data alone, outputting a shallow circuit that approximates the unknown state. The approach has implications for quantum generative models and also inspires an efficient classical algorithm for diffusion models. AI