Expand description
Löwdin and canonical orthogonalization of the overlap matrix.
The generalized eigenvalue problem Fc = Scε is converted to a standard eigenvalue problem by orthogonalizing the basis:
F’ = X†FX, F’c’ = c’ε, c = Xc’
where X = S^{-1/2} is the (symmetric) orthogonalization matrix.
§Löwdin Orthogonalization
S = UΛU† → S^{-1/2} = UΛ^{-1/2}U†
This preserves the character of the original basis functions as much as possible (least-distortion orthogonalization).
Functions§
- back_
transform - Back-transform coefficients from orthogonal to original basis: C = X C’.
- lowdin_
orthogonalization - Compute the symmetric orthogonalization matrix X = S^{-1/2}.
- transform_
to_ orthogonal - Transform a matrix to the orthogonal basis: M’ = X† M X.