Computing leading eigenvalue of a real symmetric matrix is an important problem in both theory and application. In some applications such as full configuration interaction in quantum many-body calculation, the dimension of the matrix is exponentially large, so traditional algorithms fail due to the huge computational cost. Coordinate descent method is found to be efficient for the large scale eigenvalue computation recently. In this talk, we will discuss Coordinate Descent Full Configuration Interaction (CDFCI), which applies the coordinate descent method to the leading eigenvalue problems in quantum physics and chemistry. This is a joint work with Yingzhou Li and Prof. Jianfeng Lu.

海报