An algorithm to locate stationary points on QM/MM surfaces

We want to design a second order optimization method applied to QM/MM energy surfaces. We want also to test some important features: The shape of the initial Hessian, the update Hessian formula, the usage of Cartesian coordinates for algorithms where internal coordinates are frequently used, the efficiency of the method to converge at high and low gradient norms. All these aspects are interrelated and will influence altogether to the efficency of the method.

The comparison of performance with other methods such as BFGS and L-BFGS will be a reference test to know the efficiency of our method.

Besides, the implementation task is by itself a tedious work that must be carefully checked before we proceed to apply the method to big size systems.