next up previous
Next: Skyrme HFB+VAPNP procedure: practical Up: Variation after particle-number projection Previous: The Lipkin-Nogami method

The Skyrme HFB+VAPNP method

Following the VAPNP procedure of Sec. 3.1, one can develop the Skyrme HFB+VAPNP equations by introducing the gauge-angle-dependent transition density matrices:

$\displaystyle \rho ({\bm r}\sigma ,{\bm r^{\prime }}\sigma^{\prime },\phi )$ $\displaystyle =$ $\displaystyle \sum_{nn^{\prime }}\rho_{nn^{\prime }}(\phi )~\psi_{n^{\prime
}}^{\ast }({\bm r^{\prime }},\sigma^{\prime })\psi_{n}({\bm
r},\sigma ),$ (50)
$\displaystyle \tilde{\rho}({\bm r}\sigma ,{\bm r^{\prime }}\sigma^{\prime },\phi
)$ $\displaystyle =$ $\displaystyle \sum_{nn^{\prime }}\tilde{\rho}_{nn^{\prime }}(\phi )
~\psi_{n^{\prime }}^{\ast }({\bm r^{\prime }},\sigma^{\prime })
\psi_{n}({\bm r},\sigma ).$ (51)

In the above equation, the density matrix $ \rho_{nn^{\prime }}(\phi )$ is given by Eq. (36) while

$\displaystyle \tilde{\rho}(\phi ) =e^{-i\phi }C(\phi )\tilde{\rho}.$ (52)

The associated gauge-angle-dependent local densities $ \rho ({\bm
r},\phi )$, $ \tau ({\bm r},\phi )$, $ {\mathsf J}_{ij}({\bm r },\phi )$, $ \tilde{\rho}({\bm r},\phi )$, $ \tilde{\tau}({\bm r},\phi )$, and $ \tilde{\mathsf J}_{ij}({\bm r},\phi )$ are defined by Eqs. (18) in terms of the density matrices (50) and (51). Using the Wick theorem for matrix elements [2], one can show that the gauge-angle-dependent transition energy density $ {\cal H}({\bm r},\phi )$ can be obtained from the intrinsic energy density $ {\cal
H}({\bm r})$ simply by substituting particle (pairing) local densities with their gauge-angle-dependent counterparts (e.g., $ \rho ({\bm r})$ $ \rightarrow$ $ \rho ({\bm
r},\phi )$).

In the case of Skyrme functionals, the HFB+VAPNP energy (26) can be expressed through an integral

$\displaystyle E^{N}[\rho ,\tilde{\rho}]=\int d\phi ~y(\phi )~E(\phi ),$ (53)

where the transition energy reads:

$\displaystyle E(\phi )=\frac{\langle \Phi \vert He^{i\phi \hat{N}}\vert\Phi \ra...
...ert e^{i\phi \hat{N}}\vert\Phi \rangle }=\int d{\bm r}~{\cal H}({\bm r},\phi ).$ (54)

The projected energy (53) is a functional $ E^{N}[\rho ,\tilde{\rho}]$ of the matrix elements of intrinsic (i.e., $ \phi$=0) matrices $ \rho $ and $ \tilde{\rho}$.

In order to compute the derivatives of $ E^{N}(\rho ,\tilde{\rho})$ with respect to $ \rho $ and $ \tilde{\rho}$, one should take first the derivatives of $ E^{N}[\rho ,\tilde{\rho}]$ with respect to $ \rho (\phi )$ and $ \tilde{\rho} (\phi )$, and then the derivatives of $ \rho (\phi )$ and $ \tilde{\rho} (\phi )$ with respect to the intrinsic densities $ \rho $ and $ \tilde{\rho}$. For example,

$\displaystyle \frac{\partial E^{N}[\rho ,\tilde{\rho}]}{\partial \rho
_{nn^{\prime }}}$ $\displaystyle =$ $\displaystyle \int d\phi ~y(\phi )~\left[ {\frac{1}{y(\phi
)}}\frac{\partial y(\phi )}{
\partial \rho_{nn^{\prime }}}~E(\phi )\right.$  
  $\displaystyle +$ $\displaystyle \sum_{\alpha \beta }\frac{\partial E(\phi )}{\partial
\rho_{\alph...
...hi )}\frac{\partial \rho_{\alpha \beta
}(\phi )}{\partial \rho
_{nn^{\prime }}}$  
  $\displaystyle +$ $\displaystyle \sum_{\alpha \beta }\frac{\partial E(\phi )}{\partial
\tilde{\rho...
...ial
\tilde{\rho}_{\alpha \beta }(\phi )}{
\partial \tilde{\rho}_{nn^{\prime }}}$  
  $\displaystyle +$ $\displaystyle \left. \sum_{\alpha \beta }\frac{\partial E(\phi )}{\partial
\til...
...{\rho}_{\alpha \beta }^{\ast }(-\phi )}{\partial \rho
_{nn^{\prime }}}\right] .$ (55)

With the use of the identity:

$\displaystyle \delta_{mm^{\prime }}-2ie^{-i\phi }\sin \phi ~\rho_{mm^{\prime }}(\phi )=e^{-2i\phi }C_{mm^{\prime }}(\phi ),$ (56)

the partial derivatives in Eq. (55) can easily be calculated:
$\displaystyle \frac{\partial y(\phi )}{\partial \rho_{nn^{\prime }}}$ $\displaystyle =$ $\displaystyle y(\phi
)~Y_{nn^{\prime }}(\phi ),$ (57)
$\displaystyle \frac{\partial \rho_{mm^{\prime }}(\phi )}{\partial \rho
_{nn^{\prime }}}$ $\displaystyle =$ $\displaystyle \delta_{m^{\prime }n^{\prime }}C_{mn}(\phi )$  
  $\displaystyle -$ $\displaystyle 2ie^{-i\phi }\sin(\phi )\rho_{n^{\prime }m^{\prime }}(\phi
)C_{mn}(\phi ),$ (58)
$\displaystyle \frac{\partial \tilde{\rho}_{mm^{\prime }}(\phi )}{\partial \rho
_{nn^{\prime }}}$ $\displaystyle =$ $\displaystyle -2ie^{-i\phi }\sin(\phi )\tilde{\rho}_{n^{\prime
}m^{\prime }}(\phi )C_{mn}(\phi ),$ (59)
$\displaystyle \frac{\partial \tilde{\rho}_{mm^{\prime }}(\phi )}{\partial
\tilde{\rho}_{nn^{\prime }}}$ $\displaystyle =$ $\displaystyle e^{-i\phi }C_{mn}(\phi )\delta
_{m^{\prime }n^{\prime }}$  
  $\displaystyle +$ $\displaystyle e^{-i\phi }C_{m\bar{n}}(\phi )\delta_{\bar{n}m^{\prime
}}s_{\bar{n}^{\prime }}s_{\bar{n}}^{\ast },$ (60)

where $ \bar{n}$ and $ s_{n}$ ( $ s_{n}s_{n}^{\ast }=1$, $ s_{\bar{n}}=-s_{n}$) are defined using the time-reversal operator $ \hat{T}$, as

$\displaystyle \hat{T}\psi_{n}({\bm r},\sigma )=s_{n}\psi_{\bar{n}}({\bm r},\sigma ).$ (61)

By inserting Eqs. (57)-(60) in Eq. (55), the latter reads

    $\displaystyle \frac{\partial E^{N}[\rho ,\tilde{\rho}]}{\partial \rho } =\int d\phi
~y(\phi )~Y(\phi )~E(\phi )$  
    $\displaystyle +\int d\phi ~y(\phi )~e^{-2i\phi }~C(\phi )h(\phi )C(\phi )$ (62)
    $\displaystyle -\left[ \int d\phi ~y(\phi )~2ie^{-i\phi }\sin(\phi
)~\tilde{\rho}(\phi )
\tilde{h}(\phi )C(\phi ) + {\rm h.c.}\right],$  

where
    $\displaystyle h_{nn^{\prime }}(\phi ) = \frac{\partial E(\phi )}{\partial \rho_{n^{\prime }n} (\phi )}$  
    $\displaystyle = \sum_{\sigma \sigma^{\prime }}\int
d^{3}{\bm r} ~\psi_{n}^{\ast...
... r,}\sigma
,\sigma^{\prime },\phi )\psi_{n^{\prime }}({\bm r}\sigma^{\prime
}),$ (63)
    $\displaystyle \tilde{h}_{nn^{\prime }}(\phi ) =
\frac{\partial E(\phi )}{\partial \tilde{\rho}_{n^{\prime }n}(\phi )}$  
    $\displaystyle = \sum_{\sigma \sigma^{\prime }}\int
d^{3} {\bm r}~\psi_{n}^{\ast...
... r},\sigma ,\sigma^{\prime },\phi )\psi_{n^{\prime }}({\bm
r}\sigma^{\prime }).$ (64)

The derivative of $ E^{N}(\rho ,\tilde{\rho})$ with respect to $ \tilde{\rho}$ can be computed in a similar manner. The $ \phi$-dependent fields $ h({\bm r},\sigma ,\sigma
^{\prime },\phi )$ and $ \tilde{h}({\bm r},\sigma ,\sigma
^{\prime },\phi )$ are obtained by substituting the local particle and pairing densities in the intrinsic fields $ h({\bm r},\sigma ,\sigma^{\prime })$ and $ \tilde{h}({\bm r} ,\sigma ,\sigma^{\prime })$ with their gauge-angle-dependent counterparts.

The Skyrme HFB+VAPNP equations can finally be written in the form

$\displaystyle \left( \begin{array}{cc} h^{N} & \tilde{h}^{N} \\ \tilde{h}^{N} &...
...( \begin{array}{c} \varphi^{N}_{1,k} \\ \varphi^{N}_{2,k} \end{array} \right) ,$ (65)

with particle-hole and particle-particle Hamiltonians
    $\displaystyle h^{N} =\int d\phi y(\phi )Y(\phi )E(\phi )$  
    $\displaystyle +\int d\phi y(\phi )e^{-2i\phi }~C(\phi )h(\phi )C(\phi )$ (66)
    $\displaystyle -\left[ \int d\phi y(\phi )2ie^{-i\phi }\sin(\phi
)\tilde{\rho}(\phi )
\tilde{h}(\phi )C(\phi ) + {\rm h.c.}\right],$  
$\displaystyle \tilde{h}^{N}$ $\displaystyle =$ $\displaystyle \int d\phi y(\phi )e^{-i\phi }\left[
\tilde{h}(\phi )C(\phi )+(...)^{T}\right].$ (67)

Finally, solutions of the HFB+VAPNP equations (65) allow for calculating the intrinsic density matrices as,
$\displaystyle \rho_{nn^{\prime }} = \sum_{E_k>0} \varphi^N_{2,nk} \varphi^{N*}_{2,n^{\prime }k} ,$     (68)
$\displaystyle \tilde{\rho}_{nn^{\prime }} = -\sum_{E_k>0} \varphi^N_{2,nk} \varphi^{N*}_{1,n^{\prime }k} .$     (69)

Let us re-emphasize that the densities and fields that enter the Skyrme HFB+VAPNP equations are immediate generalizations of the analogous quantities that appear in the standard Skyrme HFB formalism. Of course, due to the presence of $ C(\phi)$ and integrations over the gauge angle, the Skyrme HFB+VAPNP calculations are appreciably more involved.


next up previous
Next: Skyrme HFB+VAPNP procedure: practical Up: Variation after particle-number projection Previous: The Lipkin-Nogami method
Jacek Dobaczewski 2006-10-13