24.10.2014 Views

Get my PhD Thesis

Get my PhD Thesis

Get my PhD Thesis

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Part 1<br />

Improving Self-consistent Field Convergence<br />

tot<br />

(1)<br />

∆c = α ⋅ c (1.67)<br />

DSM<br />

and a search is made in ∆ E SCF<br />

to find the step ∆c tot that leads to the largest decrease in energy.<br />

E SCF (α) is found by evaluating the averaged density of Eq. (1.45) for the coefficients (c 0 + ∆c tot ),<br />

purifying it as in Eq. (1.32)–(1.33) and inserting it in the energy expression of Eq. (1.1). Then<br />

DSM<br />

∆ E SCF ( α)<br />

can be found as DSM<br />

∆ E ( α ) = E ( α ) − E ( D ). (1.68)<br />

SCF<br />

SCF SCF 0<br />

Fig. 1.20 and Fig. 1.21 illustrate the search in α, again for iteration seven of the HF and LDA<br />

calculations on the zinc complex in Fig. 1.3. For α = 0, no step is taken and hence no energy<br />

decrease is seen. For the marked choice of α, the optimal step length is obtained.<br />

0<br />

-5<br />

-10<br />

-15<br />

-20<br />

-25<br />

-30<br />

-35<br />

0 4 8 12 16 20<br />

α<br />

Fig. 1.20 Decrease in HF energy as a function of<br />

the step length α.<br />

0<br />

-5<br />

-10<br />

-15<br />

-20<br />

-25<br />

0 4 8 12 16 20<br />

α<br />

Fig. 1.21 Decrease in LDA energy as a function of<br />

the step length α.<br />

1.4.2.5 The Missing Term<br />

In the construction of the TRDSM energy model Eq. (1.55), the term of second order in the<br />

idempotency correction D δ was neglected from Eq. (1.50), since this term required a new Fock<br />

evaluation F(D δ ), which would increase the expenses of the scheme considerably. This section will<br />

be concerned with this neglected term and how a part of it can be described without the evaluation<br />

of a new Fock matrix, leading to an improved energy model for TRDSM at no considerable extra<br />

cost. The actual effect of this improvement to the energy model will then be discussed through a<br />

case study. This section will only be concerned with Hartree-Fock theory and examples, but it might<br />

equally well be done for DFT even though the improvement should be less significant since for<br />

DFT, also terms of order ||D + || 3 are neglected. These are of the same size as the neglected term<br />

quadratic in D δ . In Section 1.5.2 these errors are discussed.<br />

Since the only neglect in the DSM energy model Eq. (1.55) for Hartree-Fock is the term quadratic<br />

in D δ , and since the only term quadratic in the density is TrDG(D), the HF energy for the density D <br />

can be written as<br />

30

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!