To unequivocally distinguish the genuine quantumness from classicality, a widely adopted approach appeals to the negativity within a join quasi-distribution representation as a compelling evidence for the nonclassical essence. However, to construct a joint quasi-distribution with negativity from experimental data typically proves to be highly cumbersome. Here we propose a computational approach utilizing a deep generative model integrated with color mapping to construct the bivariate joint quasi-distribution functions by processing three marginals. We first apply our model to predict the Wigner functions subject to thermal noises. Our model successfully predicts the Wigner functions with a prominent accuracy by processing three marginals of probability distributions. We also tackle a challenging problem of the canonical Hamiltonian ensemble representation (CHER), which is developed for characterizing the dynamical process nonclassicality. Furthermore, we also design optimal synthetic datasets to train the model for overcoming the ground-truth deficiency of the CHER problem. While trained with synthetic data, the physics-informed optimization enables our model to capture the detrimental effect of the thermal fluctuations on nonclassicality. Our approach also provides a significant reduction of the experimental efforts of constructing the Wigner functions of quantum states.
Comment: 12 pages, 5 figures