CALYPSO is a method for predicting the energetically stable or metastable crystal structures of materials based on chemical composition and external conditions. The method uses particle swarm optimization (PSO) and other techniques such as symmetry constraints, bond characterization matrix, and penalty functions to achieve global structural minimization. These techniques are critical for efficiently finding the global minimum energy structure. The CALYPSO code has been tested on various systems, showing high efficiency and success rate in predicting crystal structures. The method involves generating random structures with symmetry constraints, local optimization, elimination of similar structures using a bond characterization matrix, and generating new structures via PSO. Symmetry constraints reduce the search space and improve efficiency. The bond characterization matrix helps eliminate similar structures, enhancing search efficiency. The penalty function rejects high-energy structures, accelerating convergence to the global minimum. Structural diversity is maintained by including random structures in each generation. The PSO algorithm is used to generate new structures, with parameters such as population size, inertia weight, and velocity adjusted to optimize performance. The method has been applied to predict various crystal structures, including high-pressure phases of lithium and bismuth telluride, as well as other compounds. The results demonstrate the effectiveness of the CALYPSO method in predicting crystal structures with high accuracy and efficiency. The code is available for non-profit organizations and includes Fortran source code, installation instructions, and examples. The method has been validated through extensive testing and has shown promising results in predicting crystal structures across a wide range of materials. The success of the method is attributed to the integration of PSO, symmetry constraints, bond characterization matrix, and penalty functions, which together enable efficient and accurate prediction of crystal structures.CALYPSO is a method for predicting the energetically stable or metastable crystal structures of materials based on chemical composition and external conditions. The method uses particle swarm optimization (PSO) and other techniques such as symmetry constraints, bond characterization matrix, and penalty functions to achieve global structural minimization. These techniques are critical for efficiently finding the global minimum energy structure. The CALYPSO code has been tested on various systems, showing high efficiency and success rate in predicting crystal structures. The method involves generating random structures with symmetry constraints, local optimization, elimination of similar structures using a bond characterization matrix, and generating new structures via PSO. Symmetry constraints reduce the search space and improve efficiency. The bond characterization matrix helps eliminate similar structures, enhancing search efficiency. The penalty function rejects high-energy structures, accelerating convergence to the global minimum. Structural diversity is maintained by including random structures in each generation. The PSO algorithm is used to generate new structures, with parameters such as population size, inertia weight, and velocity adjusted to optimize performance. The method has been applied to predict various crystal structures, including high-pressure phases of lithium and bismuth telluride, as well as other compounds. The results demonstrate the effectiveness of the CALYPSO method in predicting crystal structures with high accuracy and efficiency. The code is available for non-profit organizations and includes Fortran source code, installation instructions, and examples. The method has been validated through extensive testing and has shown promising results in predicting crystal structures across a wide range of materials. The success of the method is attributed to the integration of PSO, symmetry constraints, bond characterization matrix, and penalty functions, which together enable efficient and accurate prediction of crystal structures.