next up previous contents
Next: 5 Relativistic Effective Core Potentials Up: 4 Two-Component Methods Previous: 4.2 Perturbative Relativistic Treatments

4.3 The Douglas-Kroll Transformation

Another method of decoupling the large and small components of the DHF wavefunctions is given by the Douglas-Kroll method[22,23]. First presented by Douglas and Kroll in 1974, this method begins with a first order Foldy-Wouthoysen transformation performed in momentum space. The momentum space Hamiltonian is given by

where the action of the external potential, , on a momentum space state vector, , is given by the expression

The unitary operator used in the first-order transformation is constructed using those free-particle momentum-space eigensolutions of the Dirac Hamiltonian associated with the positive energy eigenvalues

The resultant unitary transformation operator is given by

When this unitary operator is applied to the single particle Hamiltonian, the result is given by

with and given by

Use of the definitions

provides more compact expressions for the even and odd operators

From this point, Douglas and Kroll proposed the use of a second unitary transformation, of the form

where is anti-Hermitian, and is selected such that, in the resulting transformed Hamiltonian

the term is exactly canceled by the term. This condition is fufilled if obeys the equation

Because the operator is linear in , an integral operator, must be expressed in terms of its kernel

Using the functional form of , a more explicit expression for the anti-Hermitian operator, may be obtained

This transformation removes all coupling of the upper and lower halves of the transformed wavefunction to first order in . Douglas and Kroll suggested that further separation, to arbitrary orders in the external potential were possible through the repeated application of the n order transformation

where is, again, an anti-Hermetian operator linear in the external potential. Separation to second order in the external potential represents a practical stopping point, since the higher order transformations become extremely complicated. Furthermore, higher order contributions should only be important for the proper treatment of the lowest lying electrons, and so should not substantially affect the description of the valence electronic structure. The large component block of the single-particle Hamiltonian which has been decoupled to second order in is given by

The resultant single-particle Hamiltonian may be employed used in conjunction with the coulomb inter-electron repulsion operator to define a many-electron hamiltonian:

This Hamiltonian is frequently known as the no-pair Douglas-Kroll Hamiltonian, and it has been extensively discussed and tested by Hess and coworkers[24,23,25]. Molecular wavefunctions may be obtained with the no-pair DK Hamiltonian in much the same fashion that solutions are obtained in the non-relativistic case, and investigations of systems containing heavy atoms by means of DK based electronic structure methods have given very encouraging comparison to fully relativistic results.[4]



next up previous contents
Next: 5 Relativistic Effective Core Potentials Up: 4 Two-Component Methods Previous: 4.2 Perturbative Relativistic Treatments



This document is copyright 1996,
Thu May 29 08:12:02 EDT 1997