Fig. 1. Different modeling approaches for calculating and predicting the property of glass, including physics based approaches^[48], experience based approaches^[81], and approaches combing physics and experience^[87]

Fig. 2. Physics based approaches. (a) Broken ergodicity theory of glassy state^[45]; (b) CRN model of SiO₂ glass and first-principle calculation^[46]; (c) genetic but topologically ordered and topologically disordered networks made of rigid triangles^[47]; (d) phase diagrams of glasses structure^[48-49]

Fig. 3. Experience-based approaches^[60,81]

Fig. 4. Approaches combining physics and experience. (a) Phase separation of Tb³⁺ doped fluoroaluminosilicate glass using MD simulation^[87]; (b) structural and electronic properties of aluminoborosilicate glass using AIMD^[92]; (c) dissolution kinetics of silicate glasses by topology-informed machine learning^[93]; (d) ML prediction model by SHAP framework^[94]

Fig. 5. Experimental results. (a) Effects of three different topological structure on luminescence properties of Bi doped laser glass^[97]; (b) calculation of properties of Tm³⁺ doped BaO-GeO₂ using phase diagram approach^[100]; (c) statistical modeling of the relationship between luminescence properties and structural units of Yb³⁺-doped phosphate laser glass with different concentrations of GeO₂^[104]; (d) glass selection charts based on ML model^[63]; (e) reconstruction process of CdSe quantum dot-glass interface by AIMD^[105]

Fig. 6. Development of optical glass modeling, including empirical summary, theoretical description, computational prediction^[105] , and data-intensive applications