A Novel Valence-Bond-Based Automatic Diabatization Method by Compression.

A novel valence-bond-based automatic diabatization method by compression, called valence-bond-based compression approach for dibatization (VBCAD), is presented in this Letter. It is a "black-box" type method that provides an automatic diabatization from a classical valence bond (VB) perspective. In VBCAD, a model space projection is performed by an eigenvalue decomposition algorithm followed by dimensional reduction based on a sequence of Householder transformations. Our diabaticity criterion is implemented in a way that maximizes the diversity of VB structure weights between different diabatic states. Owing to the rigorous Householder transformations employed in this entire procedure, the invariance of the target eigensubspace is preserved. This is illustrated on two prototypical examples.