Optimization has many key applications in airline, space, energy, transport, military, and industry. Optimization softwares have been in high demand recently. In this talk, we introduce OptSuite, a C++ open source framework for optimization. The framework is designed to provide fast scientific computing subroutines for general algorithm development, individual optimization modules for conic programming, integer programming, composite optimization, manifold optimization, etc. Specially tuned solvers are also included for real applications such as LASSO and matrix completion. To further improve the flexibility and efficiency, we define an abstract variable class that enables the use of any user-defined variable datatype in the solvers of OptSuite. Preliminary results in matrix completion and semidefinite programming justify the technical strategy and indicate the effectiveness of our framework. 

Note: this is a joint work with Prof. Zaiwen Wen (BICMR, PKU), and is still in the early stage of development.