摘要: The intricate amorphous structure and elusive local environment of rare-earth (RE) doped laser glasses render existing physical or machine-learning models inadequate for guiding the design of next-generation laser glasses, thereby maintaining iterative experimentation as the primary method for development. Here, the microenvironment surrounding RE ions in a multicomponent laser glass is treated as a statistical ensemble derived from its neighboring glassy compounds (NGCs). The NGCs model employs statistical ensemble averaging over the NGCs to provide a rigorous mathematical description of the key local structural and luminescent behaviors of RE-doped laser glasses. Validation through molecular dynamics simulations and experimental data for quaternary germanate glass system demonstrates the model's excellent predictive capabilities, allowing it to establish the composition-structure relationship and populate the composition-property space. Moreover, the model enables the creation of multi-luminescence property charts, facilitating the de novo design of chemically complex laser glasses for targeted applications by efficiently screening the compositions which simultaneously meet several performance constraints. This work offers a robust framework for studying the luminescent behaviors of glass and paves the way for new explorations in laser glass technology.