Quantum State Generation with Structure-Preserving Diffusion Model

arXiv:2404.06336v1 Announce Type: cross
Abstract: This article considers the generative modeling of the states of quantum systems, and an approach based on denoising diffusion model is proposed. The key contribution is an algorithmic innovation that respects the physical nature of quantum states. More precisely, the commonly used density matrix representation of mixed-state has to be complex-valued Hermitian, positive semi-definite, and trace one. Generic diffusion models, or other generative methods, may not be able to generate data that strictly satisfy these …

abstract article arxiv cs.lg denoising diffusion diffusion model generative generative modeling innovation key matrix mixed modeling nature quant-ph quantum representation state stat.ml systems the key type