 This paper proposes a physics-informed information criterion, PIC, to measure the accuracy and simplicity of discovered partial differential equations, PDEs. The PIC was tested on seven canonical PDEs from various physical scenarios, demonstrating its robustness to noisy and sparse data. It was then used to discover unrevealed governing equations from microscopic simulation data in an actual physical scene. The discovered macro-scale PDE-satisfied underlying symmetries and was both accurate and parsimonious, making it easier to understand and simulate the physical process. This shows that the PIC can be applied to discover unrevealed governing equations in broader physical scenes. This article was authored by Hao Siu, Jun Qingxing and Dong Xiaozhuang.