next up previous contents
Next: 7.4 Physical interpretation Up: 7. Computational implementation Previous: 7.2 Energy gradients   Contents


7.3 Penalty functional and electron number

The penalty functional $P[\rho]$ is defined by

$\displaystyle P[\rho]$ $\textstyle =$ $\displaystyle \int {\mathrm d}{\bf r}~ \left(\rho^2(1-\rho)^2\right)({\bf r},{\bf r})$  
  $\textstyle =$ $\displaystyle \int {\mathrm d}{\bf r}_1 {\mathrm d}{\bf r}_2 {\mathrm d}{\bf r}...
... \right]
\left[ \delta({\bf r}_4,{\bf r}_1) - \rho({\bf r}_4,{\bf r}_1) \right]$  
  $\textstyle =$ $\displaystyle K^{ij} S_{jk} K^{kl} S_{lm} ( \delta^m_p - K^{mn} S_{np} )
( \delta^p_i - K^{pq} S_{qi} ) .$ (7.37)

The derivative with respect to the density-kernel is then
\begin{displaymath}
\frac{\partial P[\rho]}{\partial K^{\alpha \beta}}=
2 S_{\be...
...^{kl} S_{lm} )
( \delta^m_{\alpha} - 2 K^{mn} S_{n \alpha} ) .
\end{displaymath} (7.38)

The penalty functional depends implicitly upon the support functions though the overlap matrix:
\begin{displaymath}
\frac{\delta S_{ij}}{\delta \phi_{\alpha}({\bf r})}=
\delta_i^{\alpha} \phi_j({\bf r}) + \delta_j^{\alpha} \phi_i({\bf r})
\end{displaymath} (7.39)

so that
\begin{displaymath}
\frac{\delta P[\rho]}{\delta \phi_{\alpha}({\bf r})}=
4 K^{\...
...lta_m^{\beta} - 2 S_{mn} K^{n \beta} ) \phi_{\beta}({\bf r}) .
\end{displaymath} (7.40)

For the sake of completeness, we now describe the expressions for the electron number and its derivatives.
\begin{displaymath}
N = 2 \int {\mathrm d}{\bf r}~ \rho({\bf r},{\bf r}) = 2 K^{ij} S_{ji}
\end{displaymath} (7.41)


\begin{displaymath}
\frac{\partial N}{\partial K^{\alpha \beta}}=
2 S_{\beta \alpha}
\end{displaymath} (7.42)


\begin{displaymath}
\frac{\delta N}{\delta \phi_{\alpha}({\bf r})}=
2 K^{\alpha \beta} \phi_{\beta}({\bf r})
\end{displaymath} (7.43)


next up previous contents
Next: 7.4 Physical interpretation Up: 7. Computational implementation Previous: 7.2 Energy gradients   Contents
Peter Haynes