The self-consistent field (SCF) technique has been the workhorse of computational chemistry research. As the size and complexity of computational system increases, any new algorithmic development that offers additOional speed-up is always welcome, even if it is as simple as an alternative method that works well for certain chemical systems. The direct inversion in the iterative subspace (DIIS) method is a widely used technique that speeds up the SCF convergence by minimizing an error function associated with each SCF step. In this work, we introduce a different DIIS method that uses quasi-Newton steps as error vectors. Mathematical formalisms necessary for computation of the error vector are presented. Approximate Hessian is used to avoid extra computational cost and storage requirement. A select set of molecules are used to test the performance of the algorithm.
展开▼
