done with parquet notes for now
This commit is contained in:
parent
a3a4c31b23
commit
0eef510f49
|
|
@ -47,7 +47,12 @@
|
|||
\newcommand{\si}{\sigma}
|
||||
\newcommand{\la}{\lambda}
|
||||
\newcommand{\ii}{\mathrm{i}}
|
||||
|
||||
\newcommand{\up}{\uparrow}
|
||||
\newcommand{\dw}{\downarrow}
|
||||
\newcommand{\upup}{\up\up}
|
||||
\newcommand{\updw}{\up\dw}
|
||||
\newcommand{\dwup}{\dw\up}
|
||||
\newcommand{\dwdw}{\dw\downarrow}
|
||||
% addresses
|
||||
\newcommand{\LCPQ}{Laboratoire de Chimie et Physique Quantiques (UMR 5626), Universit\'e de Toulouse, CNRS, UPS, France}
|
||||
|
||||
|
|
@ -145,8 +150,8 @@ with
|
|||
\end{gather}
|
||||
One must be very careful not to double-count diagrams in the pp sector.
|
||||
|
||||
The complete ($\Gamma$ and $\Gamma_P$) and irreducible vertices ($\Gamma^\pp$, $\Gamma^\ph$, and $\Bar{\Gamma}^\ph$) are usually spin adapted.
|
||||
For example, $\Gamma^\ph$ is written as a sum of its density ($\Gamma^\ph_\text{d}$) and magnetic ($\Gamma^\ph_\text{m}$) components, while $\Gamma^\pp$ is decomposed in its singlet ($\Gamma^\ph_\text{s}$) and triplet ($\Gamma^\ph_\text{t}$) components.
|
||||
The complete ($\Gamma$ and $\Gamma_P$) and irreducible vertices ($\Gamma^\pp$, $\Gamma^\ph$, and $\Bar{\Gamma}^\ph$) are usually spin adapted (see below).
|
||||
For example, $\Gamma^\ph$ is written as a sum of its density ($\Gamma^\ph_\text{d}$) and magnetic ($\Gamma^\ph_\text{m}$) components, while $\Gamma^\pp$ is decomposed in its singlet ($\Gamma^\pp_\text{s}$) and triplet ($\Gamma^\pp_\text{t}$) components.
|
||||
Similar expressions can be found for $\Bar{\Gamma}^\ph$ (crossing relation), and the approximated forms of $\Gamma$ and $\Gamma_P$ (but usually violate crossing conditions).
|
||||
|
||||
|
||||
|
|
@ -230,7 +235,7 @@ Now, we focus on the self-energy which can be written as a non-scattering and a
|
|||
\begin{equation}
|
||||
\Sigma(11') = \Sigma_1(11') + \Sigma_2(11')
|
||||
\end{equation}
|
||||
where, after various manipulations, on gets
|
||||
where, after various manipulations, one gets
|
||||
\begin{equation}
|
||||
\begin{split}
|
||||
\Sigma_2(11')
|
||||
|
|
@ -254,9 +259,157 @@ where, after various manipulations, on gets
|
|||
Thanks to the spin adaptation, we have the following picture: the electrons interacting through the exchange of four varieties of composite ``bosons'' made of electrons and holes which themselves strongly interact through the exchange of other bosons: density, magnetic, singlet-pair, and triplet-pair fluctuations.
|
||||
Thanks to the crossing symmetry of the complete vertices, the parquet equations automatically build in the nonlinear coupling between dressed electron and boson excitations necessary for full consistency.
|
||||
|
||||
We now explicitly treat the spin degrees of freedom starting from the ph vertex functions that we decompose in density and magnetic components.
|
||||
We hence remove the spin variable from the composite variable $1 \equiv (\br_1,t_1)$.
|
||||
Defining the density and magnetic operators
|
||||
\begin{gather}
|
||||
d(12) = \frac{1}{\sqrt{2}} \qty[ \cre{1\up}\ani{2\up} + \cre{1\dw}\ani{2\dw} ]
|
||||
\\
|
||||
m(12) = \frac{1}{\sqrt{2}} \qty[ \cre{1\up}\ani{2\up} - \cre{1\dw}\ani{2\dw} ]
|
||||
\end{gather}
|
||||
we have
|
||||
\begin{gather}
|
||||
\Gamma_\text{m}^{+1}(12;34) = \Gamma_{\updw,\updw}(12;34)
|
||||
\\
|
||||
\Gamma_\text{m}^{-1}(12;34) = \Gamma_{\dwup,\dwup}(12;34)
|
||||
\end{gather}
|
||||
and
|
||||
\begin{gather}
|
||||
\Gamma_\text{d}(12;34) = \Gamma_{\upup,\upup}(12;34) + \Gamma_{\upup,\dwdw}(12;34)
|
||||
\\
|
||||
\Gamma_\text{m}^{0}(12;34) = \Gamma_{\upup,\upup}(12;34) - \Gamma_{\upup,\dwdw}(12;34)
|
||||
\end{gather}
|
||||
Hence, $\Gamma_\text{m} \equiv \Gamma_\text{m}^{0} = \Gamma_\text{m}^{\pm1}$.
|
||||
|
||||
Now, let us decompose the pp vertex functions into their singlet and triplet components in very much the same way.
|
||||
Again, the 3 triplet components
|
||||
\begin{gather}
|
||||
\Gamma_\text{t}^{+1}(12;34) = \Gamma_{\upup,\upup}(12;34)
|
||||
\\
|
||||
\Gamma_\text{t}^{-1}(12;34) = \Gamma_{\dwdw,\dwdw}(12;34)
|
||||
\\
|
||||
\Gamma_\text{t}^{0}(12;34) = \Gamma_{\updw,\updw}(12;34) - \Gamma_{\updw,\dwup}(12;34)
|
||||
\end{gather}
|
||||
are equal and abbreviated as $\Gamma_\text{t}$ while the singlet component reads
|
||||
\begin{equation}
|
||||
\Gamma_\text{s}^{0}(12;34) = \Gamma_{\updw,\updw}(12;34) + \Gamma_{\updw,\dwup}(12;34)
|
||||
\end{equation}
|
||||
We recover the fact that the singlet component is symmetric under the exchange of space-time labels, while the triplets are antisymmetric, i.e.
|
||||
\begin{gather}
|
||||
\Gamma_\text{s}(12;34) = + \Gamma_\text{s}(12;43)
|
||||
\\
|
||||
\Gamma_\text{t}(12;34) = - \Gamma_\text{t}(12;43)
|
||||
\end{gather}
|
||||
|
||||
We are now ready to write down the parquet equations in a spin-adapted basis.
|
||||
Defining
|
||||
\begin{gather}
|
||||
\Phi_\text{d/m}(12;34) = \qty[ \Gamma_\text{d/m}^\ph (\bI - G^\ph \Gamma_\text{d/m}^\ph)^{-1} G^\ph \Gamma_\text{d/m}^\ph ](12;34)
|
||||
\\
|
||||
\Psi_\text{s/t}(12;34) = \qty[ \Gamma_\text{s/t}^\pp (\bI - G^\pp \Gamma_\text{s/t}^\pp)^{-1} G^\pp \Gamma_\text{s/t}^\pp ](12;34)
|
||||
\end{gather}
|
||||
we get
|
||||
\begin{align}
|
||||
\begin{split}
|
||||
\Gamma_\text{d}^\ph(12;34)
|
||||
= \Lambda_\text{d}^\irr(12;34)
|
||||
& - \frac{1}{2} \Phi_\text{d}(42;31)
|
||||
- \frac{3}{2} \Phi_\text{m}(42;31)
|
||||
\\
|
||||
& + \frac{1}{2} \Psi_\text{s}(41;32)
|
||||
+ \frac{3}{2} \Psi_\text{t}(41;32)
|
||||
\end{split}
|
||||
\\
|
||||
\begin{split}
|
||||
\Gamma_\text{m}^\ph(12;34)
|
||||
= \Lambda_\text{m}^\irr(12;34)
|
||||
& - \frac{1}{2} \Phi_\text{d}(42;31)
|
||||
+ \frac{1}{2} \Phi_\text{m}(42;31)
|
||||
\\
|
||||
& - \frac{1}{2} \Psi_\text{s}(41;32)
|
||||
+ \frac{1}{2} \Psi_\text{t}(41;32)
|
||||
\end{split}
|
||||
\\
|
||||
\begin{split}
|
||||
\Gamma_\text{s}^\pp(12;34)
|
||||
= \Lambda_\text{s}^\irr(12;34)
|
||||
& + \frac{1}{2} \Phi_\text{d}(24;31)
|
||||
- \frac{3}{2} \Phi_\text{m}(24;31)
|
||||
\\
|
||||
& + \frac{1}{2} \Psi_\text{s}(14;32)
|
||||
- \frac{3}{2} \Psi_\text{t}(14;32)
|
||||
\end{split}
|
||||
\\
|
||||
\begin{split}
|
||||
\Gamma_\text{t}^\pp(12;34)
|
||||
= \Lambda_\text{t}^\irr(12;34)
|
||||
& + \frac{1}{2} \Phi_\text{d}(24;31)
|
||||
+ \frac{1}{2} \Phi_\text{m}(24;31)
|
||||
\\
|
||||
& - \frac{1}{2} \Psi_\text{s}(14;32)
|
||||
- \frac{1}{2} \Psi_\text{t}(14;32)
|
||||
\end{split}
|
||||
\end{align}
|
||||
Check out this nice symmetry!
|
||||
Now that we have the parquet equations written in a convenient spin-adapted form, the last step is to write down the self-energy via the Dyson-Schwinger equation.
|
||||
Doing so displays the four varieties of composite bosons mentioned previously:
|
||||
\begin{equation}
|
||||
\label{eq:Sig2_sa}
|
||||
\begin{split}
|
||||
\Sigma_2(11')
|
||||
= \frac{1}{2} \Bigg\{
|
||||
& - \frac{1}{2} G(76) \qty[
|
||||
\frac{1}{2} \Lambda_\text{d}^\irr G^\ph v_\text{d} + \frac{3}{2} \Lambda_\text{m}^\irr G^\ph v_\text{m} ](17;1'6)
|
||||
\\
|
||||
& + G(67) \qty[ \frac{1}{2} \Lambda_\text{s}^\irr G^\pp v_\text{s} + \frac{3}{2} \Lambda_\text{t}^\irr G^\pp v_\text{t} ](17;1'6)
|
||||
\Bigg\}
|
||||
\\
|
||||
& - G(76) \qty[ \frac{1}{2} \Phi_\text{d} G^\ph v_\text{d} + \frac{3}{2} \Phi_\text{m} G^\ph v_\text{m} ](17;1'6)
|
||||
\\
|
||||
& + G(67) \qty[ \frac{1}{2} \Psi_\text{s} G^\pp v_\text{s} + \frac{3}{2} \Psi_\text{t} G^\pp v_\text{t} ](17;1'6)
|
||||
\end{split}
|
||||
\end{equation}
|
||||
Finally, the crossing relations become
|
||||
\begin{gather}
|
||||
\Gamma_\text{m}(12;34) = - \frac{1}{2} \Gamma_\text{d}(42;31) + \frac{1}{2} \Gamma_\text{m}(42;31)
|
||||
\\
|
||||
\Gamma_\text{m}(12;34) = - \frac{1}{2} \Gamma_\text{s}(14;32) - \frac{1}{2} \Gamma_\text{t}(14;32)
|
||||
\end{gather}
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
\section{Fluctuation-exchange approximation}
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
The FLEX approximation is a variant of the Baym-Kadanoff approximation and then focuses on the self-energy itself.
|
||||
This lack of two-body self-consistency limits quantitative accuracy of FLEX compared to parquet.
|
||||
In FLEX, one sets $\Lambda^\irr = \Gamma^\ph = \Gamma^\pp = v$.
|
||||
Equation \eqref{eq:Sig2_sa} then becomes
|
||||
\begin{equation}
|
||||
\label{eq:Sig2_sa}
|
||||
\begin{split}
|
||||
\Sigma_2(11')
|
||||
= \frac{1}{2} \Bigg\{
|
||||
& - \frac{1}{2} G(76) \qty[
|
||||
\frac{1}{2} v_\text{d} G^\ph v_\text{d} + \frac{3}{2} v_\text{m} G^\ph v_\text{m} ](17;1'6)
|
||||
\\
|
||||
& + G(67) \qty[ \frac{1}{2} v_\text{s} G^\pp v_\text{s} + \frac{3}{2} v_\text{t} G^\pp v_\text{t} ](17;1'6)
|
||||
\Bigg\}
|
||||
\\
|
||||
& - G(76) \Big[
|
||||
\frac{1}{2} v_\text{d} (\bI - G^\ph v_\text{d})^{-1} (G^\ph v_\text{d})^2
|
||||
\\
|
||||
& + \frac{3}{2} v_\text{m} (\bI - G^\ph v_\text{m})^{-1} (G^\ph v_\text{m})^2
|
||||
\Big](17;1'6)
|
||||
\\
|
||||
& + G(67) \Big[
|
||||
\frac{1}{2} v_\text{s} (\bI - G^\pp)^{-1} (G^\pp v_\text{s})^2
|
||||
\\
|
||||
& + \frac{3}{2} v_\text{t} (\bI - G^\pp)^{-1} (G^\pp v_\text{t})^2
|
||||
\Big](17;1'6)
|
||||
\end{split}
|
||||
\end{equation}
|
||||
Because FLEX is a parquet with non-self-consistent two-body vertices, their expressions are a bit different and two flavours exist.
|
||||
The first flavour includes the single-fluctuation-exchange diagrams while the second flavor includes also the so-called Aslamazov-Larkin diagrams.
|
||||
When the vertex functions are spin-diagonalized omitting the Aslamazov-Larkin contributions, we recover the parquet equations where one sets $\Lambda^\irr = \Gamma^\ph = \Gamma^\pp = v$.
|
||||
It has been shown that the Aslamazov-Larkin diagrams deterioetas the accuray of the two-body vertices obtained via FLEX.
|
||||
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
\acknowledgements{
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user