getting there

This commit is contained in:
Pierre-Francois Loos 2019-06-24 22:52:06 +02:00
parent 4dcf1e69c2
commit 5514984601

View File

@ -255,7 +255,7 @@ An important property of the present correction is
\label{eq:limitfunc} \label{eq:limitfunc}
\lim_{\Bas \to \CBS} \DbE{}{\Bas}[\n{0}{\Bas},\n{k}{\Bas}] = 0. \lim_{\Bas \to \CBS} \DbE{}{\Bas}[\n{0}{\Bas},\n{k}{\Bas}] = 0.
\end{equation} \end{equation}
In other words, the correction vanishes in the CBS limit, hence guaranteeing an unaltered CBS limit. \cite{LooPraSceTouGin-JPCL-19} In other words, the correction vanishes in the CBS limit, hence guaranteeing an unaltered limit. \cite{LooPraSceTouGin-JPCL-19}
%In the following, we will drop the state index $k$ and focus on the quantity $\bE{}{\Bas}[\n{}{}]$. %In the following, we will drop the state index $k$ and focus on the quantity $\bE{}{\Bas}[\n{}{}]$.
%Mention that the present approach is a one-shot procedure, i.e.~there is no self-consistent procedure involved. %Mention that the present approach is a one-shot procedure, i.e.~there is no self-consistent procedure involved.
@ -311,7 +311,6 @@ An important feature of $\W{}{\Bas}(\br{1},\br{2})$ is that it tends to the regu
\end{equation} \end{equation}
ensuring that $\bE{}{\Bas}[\n{}{}]$ vanishes when $\Bas$ is complete. ensuring that $\bE{}{\Bas}[\n{}{}]$ vanishes when $\Bas$ is complete.
We refer the interested readers to Refs.~\onlinecite{GinPraFerAssSavTou-JCP-18,LooPraSceTouGin-JPCL-19} for additional details. We refer the interested readers to Refs.~\onlinecite{GinPraFerAssSavTou-JCP-18,LooPraSceTouGin-JPCL-19} for additional details.
\titou{From hereon, we drop the dependency in $\Bas$.}
%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Short-range correlation functionals} \subsection{Short-range correlation functionals}
@ -326,16 +325,18 @@ The local-density approximation (LDA) of the ECMD complementary functional is de
where $\zeta = (\n{\uparrow}{} - \n{\downarrow}{})/\n{}{}$ is the spin polarization and $\be{\text{c,md}}{\sr,\LDA}(\n{}{},\zeta,\rsmu{}{})$ is the ECMD short-range correlation energy per electron of the uniform electron gas (UEG) \cite{LooGil-WIRES-16} parameterized in Ref.~\citenum{PazMorGorBac-PRB-06}. where $\zeta = (\n{\uparrow}{} - \n{\downarrow}{})/\n{}{}$ is the spin polarization and $\be{\text{c,md}}{\sr,\LDA}(\n{}{},\zeta,\rsmu{}{})$ is the ECMD short-range correlation energy per electron of the uniform electron gas (UEG) \cite{LooGil-WIRES-16} parameterized in Ref.~\citenum{PazMorGorBac-PRB-06}.
The $\be{\text{c,md}}{\sr,\LDA}$ used in \eqref{eq:def_lda_tot} presents two main defects: i) at small $\mu$, it overestimates the correlation energy, and ii) UEG-based quantities are hardly transferable when the system becomes strongly correlated or multi-configurational. The $\be{\text{c,md}}{\sr,\LDA}$ used in \eqref{eq:def_lda_tot} presents two main defects: i) at small $\mu$, it overestimates the correlation energy, and ii) UEG-based quantities are hardly transferable when the system becomes strongly correlated or multi-configurational.
An attempt to solve these problems has been proposed by some of the authors in the context of RS-DFT. \cite{FerGinTou-JCP-18} An attempt to solve these problems was suggested by some of the authors in the context of RS-DFT. \cite{FerGinTou-JCP-18}
They proposed to interpolate between the usual Perdew-Burke-Ernzerhof (PBE) correlation functional \cite{PerBurErn-PRL-96} $\e{\text{c}}{\PBE}(\n{}{},s,\zeta)$ at $\mu = 0$ and the exact large-$\mu$ behavior known from RS-DFT\cite{TouColSav-PRA-04, GorSav-PRA-06, PazMorGorBac-PRB-06}. They proposed to interpolate between the usual Perdew-Burke-Ernzerhof (PBE) correlation functional \cite{PerBurErn-PRL-96} $\e{\text{c}}{\PBE}(\n{}{},s,\zeta)$ at $\mu = 0$ and the exact large-$\mu$ behavior. \cite{TouColSav-PRA-04, GorSav-PRA-06, PazMorGorBac-PRB-06}
\manu{In the context of RS-DFT, the large-$\mu$ behavior corresponds to an extremely short-range interaction, and in that regime the ECMD energy only depends on the \textit{exact} on-top pair density $\n{2}{}(\br{},\br{})$ In the context of RS-DFT, the large-$\mu$ behavior corresponds to an extremely short-range interaction.
In this regime, the ECMD energy
\begin{align} \begin{align}
\label{eq:exact_large_mu} \label{eq:exact_large_mu}
\bE{\text{c,md}}{\sr}[\n{2}{},\rsmu{}{}] \propto \frac{1}{\mu^3} \int \dbr{} \n{2}{}(\br{},\br{}) + O(\frac{1}{\mu^4}), \bE{\text{c,md}}{\sr}[\n{2}{},\rsmu{}{}] \propto \frac{1}{\mu^3} \int \dbr{} \n{2}{}(\br{}) + \order*{\mu^{-4}}
\end{align} \end{align}
where $\n{2}{}(\br{},\br{})$ is obtained from the exact ground state wave function $\Psi$ developed in an complete basis set. only depends on the \textit{exact} on-top pair density $\n{2}{}(\br{}) \equiv \n{2}{}(\br{},\br{})$ which is obtained from the \textit{exact} ground state wave function $\Psi$ belonging to the Hilbert space spanned by the complete basis set.
Obviously, an exact quantity such as $\n{2}{}(\br{},\br{})$ is out of reach in practical calculations and must be approximated by a function generally referred here as $\tn{2}{}(\br{},\br{})$.
For a given choice of $\tn{2}{}(\br{},\br{})$, the authors in Ref~\cite{FerGinTou-JCP-18} proposed to use the following functional form for the correlation energy density in order to interpolate between the PBE at $\mu = 0$ and the equation \eqref{eq:exact_large_mu}: Obviously, an exact quantity such as $\n{2}{}(\br{})$ is out of reach in practical calculations and must be approximated by a function referred here as $\tn{2}{}(\br{})$.
For a given $\tn{2}{}(\br{})$, some of the authors proposed the following functional form in order to interpolate between $\e{\text{c}}{\PBE}(\n{}{},s,\zeta)$ at $\mu = 0$ and Eq.~\eqref{eq:exact_large_mu} as $\mu \to \infty$: \cite{FerGinTou-JCP-18}
\begin{subequations} \begin{subequations}
\begin{gather} \begin{gather}
\label{eq:epsilon_cmdpbe} \label{eq:epsilon_cmdpbe}
@ -345,49 +346,48 @@ For a given choice of $\tn{2}{}(\br{},\br{})$, the authors in Ref~\cite{FerGinTo
\beta^\PBE(\n{}{},\tn{2}{},s,\zeta) = \frac{3}{2\sqrt{\pi} (1 - \sqrt{2})} \frac{\e{\text{c}}{\PBE}(\n{}{},s,\zeta)}{\tn{2}{}/\n{}{}}. \beta^\PBE(\n{}{},\tn{2}{},s,\zeta) = \frac{3}{2\sqrt{\pi} (1 - \sqrt{2})} \frac{\e{\text{c}}{\PBE}(\n{}{},s,\zeta)}{\tn{2}{}/\n{}{}}.
\end{gather} \end{gather}
\end{subequations} \end{subequations}
As illustrated in the context of RS-DFT in Ref~\cite{FerGinTou-JCP-18}, such a functional form is able to treat both the weak correlation regime (thanks to the use of the $\e{\text{c}}{\PBE}$) and the strong correlation regime (thanks to good approximations of the on-top pair density $\tn{2}{}$).\\ As illustrated in the context of RS-DFT, \cite{FerGinTou-JCP-18} such a functional form is able to treat both weakly and strongly correlated systems thanks to the explicit inclusion of $\e{\text{c}}{\PBE}$ and $\tn{2}{}$, respectively.
Therefore, in the context of the basis set correction we use the explicit form of equations ~\eqref{eq:epsilon_cmdpbe}~and~\eqref{eq:beta_cmdpbe} with $\rsmu{}{\manu{\Bas}}$ and introduce the general form of the PBE-based complementary functional: }
\begin{multline}
\label{eq:def_pbe_tot}
\bE{\PBE}{\manu{\Bas}}[\n{}{},\tn{2}{},\rsmu{}{\manu{\Bas}}] =
\int \n{}{}(\br{})
\\
\times \be{\text{c,md}}{\sr,\PBE}\qty(\n{}{}(\br{}),\tn{2}{}(\br{}),s(\br{}),\zeta(\br{}),\rsmu{}{\manu{\Bas}}(\br{})) \dbr{},
\end{multline}
\manu{which depends on the approximation used for $\tn{2}{}$.}
\manu{In Ref.~\onlinecite{LooPraSceTouGin-JPCL-19}, some of the authors introduced a version of the PBE-based functional, here-referred as \titou{PBE-UEG}:} Therefore, in the present context, we consider the explicit form of Eqs.~\eqref{eq:epsilon_cmdpbe} and \eqref{eq:beta_cmdpbe} with $\rsmu{}{\Bas}$ and introduce the general form of the PBE-based complementary functional:
\begin{multline} \begin{multline}
\label{eq:def_pbe_tot} \label{eq:def_pbe_tot}
\bE{\titou{\PBE\text{-}\UEG}}{\manu{\Bas}}[\n{}{},\n{2}{\UEG},\rsmu{}{\manu{\Bas}}] = \bE{\PBE}{\Bas}[\n{}{},\tn{2}{},\rsmu{}{\Bas}] =
\int \n{}{}(\br{}) \int \n{}{}(\br{})
\\ \\
\times \be{\text{c,md}}{\sr,\PBE}\qty(\n{}{}(\br{}),\n{2}{\UEG}(\br{}),s(\br{}),\zeta(\br{}),\rsmu{}{\manu{\Bas}}(\br{})) \dbr{}, \times \be{\text{c,md}}{\sr,\PBE}\qty(\n{}{}(\br{}),\tn{2}{}(\br{}),s(\br{}),\zeta(\br{}),\rsmu{}{\Bas}(\br{})) \dbr{},
\end{multline} \end{multline}
in which the on-top pair density was approximated by its UEG version, \manu{\textit{i.e.} $\tn{2}{}(\br{}) = \n{2}{\UEG}(\br{})$, with} which depends on the approximation level of $\tn{2}{}$.
In Ref.~\onlinecite{LooPraSceTouGin-JPCL-19}, some of the authors introduced a version of the PBE-based functional, here-referred as \titou{PBE-UEG}:
\begin{equation}
\label{eq:def_pbe_tot}
\bE{\titou{\PBE\text{-}\UEG}}{\Bas}
\equiv
\bE{\PBE}{\Bas}[\n{}{},\n{2}{\UEG},\rsmu{}{\Bas}],
\end{equation}
in which the on-top pair density was approximated by its UEG version, i.e., $\tn{2}{}(\br{}) = \n{2}{\UEG}(\br{})$, with
\begin{equation} \begin{equation}
\n{2}{\UEG}(\br{}) = n(\br{})^2 (1-\zeta(\br{})^2) g_0(n(\br{})), \n{2}{\UEG}(\br{}) = n(\br{})^2 (1-\zeta(\br{})^2) g_0(n(\br{})),
\end{equation} \end{equation}
and where $g_0(n)$ is the UEG on-top pair distribution function [see Eq.~(46) of Ref.~\citenum{GorSav-PRA-06}]. and where $g_0(n)$ is the UEG on-top pair distribution function [see Eq.~(46) of Ref.~\citenum{GorSav-PRA-06}].
\manu{As illustrated in Ref~\cite{LooPraSceTouGin-JPCL-19} for weakly correlated systems, this PBE-based functional has clearly shown to improve energetics over the pure UEG-based functional $\bE{\LDA}{\Bas}$ (see \eqref{eq:def_lda_tot}) thanks to the inclusion of the PBE functional.} As illustrated in Ref.~\onlinecite{LooPraSceTouGin-JPCL-19} for weakly correlated systems, this PBE-based functional has clearly shown to improve energetics over the pure UEG-based functional $\bE{\LDA}{\Bas}$ (see \eqref{eq:def_lda_tot}) thanks to the inclusion of the PBE functional.
However, the underlying UEG on-top pair density might not be suited for the treatment of excited states and/or strongly correlated systems. However, the underlying UEG on-top pair density might not be suited for the treatment of excited states and/or strongly correlated systems.
\manu{Also, in the context of the basis set correction, we have to compute the on-top pair density $\n{2}{\manu{\Bas}}(\br{})$ to compute $\rsmu{}{\manu{\Bas}}(\br{})$ (see equations~\eqref{eq:def_mu} and \eqref{eq:def_weebasis}), and therefore we can use $\n{2}{\manu{\Bas}}(\br{})$ to have an approximation of the exact on-top pair density. Besides, in the context of the present basis set correction, we have to compute $\n{2}{\Bas}(\br{})$ to obtain $\rsmu{}{\Bas}(\br{})$ [see Eqs.~\eqref{eq:def_mu} and \eqref{eq:def_weebasis}].
More precisely, we propose an approximation of the on-top pair density $\ttn{2}{\manu{\Bas}}(\br{})$ defined as Therefore we can use $\n{2}{\Bas}(\br{})$ to have an approximation of the exact on-top pair density.}
More precisely, we propose an approximation of the on-top pair density $\ttn{2}{\Bas}(\br{})$ defined as
\begin{equation} \begin{equation}
\ttn{2}{\manu{\Bas}}(\br{}) = \n{2}{\manu{\Bas}}(\br{}) \qty( 1 + \frac{2}{\sqrt{\pi}\rsmu{}{\manu{\Bas}}(\br{})})^{-1} \ttn{2}{\Bas}(\br{}) = \n{2}{\Bas}(\br{}) \qty( 1 + \frac{2}{\sqrt{\pi}\rsmu{}{\Bas}(\br{})})^{-1}
\end{equation} \end{equation}
which is inspired by the extrapolation at large $\mu$ of the exact on-top pair density proposed by Gori-Giorgi and Savin\cite{GorSav-PRA-06} in the context of RS-DFT. which is inspired by the large-$\mu$ extrapolation of the exact on-top pair density proposed by Gori-Giorgi and Savin \cite{GorSav-PRA-06} in the context of RS-DFT.
}
Therefore, we propose here the "PBE-ontop" (PBEot) functional which uses the on-top pair density computed in the basis set $\Bas$ as ingredient: Therefore, we propose here the "PBE-ontop" (PBEot) functional which uses the on-top pair density computed in the basis set $\Bas$ as ingredient
\begin{multline} \begin{equation}
\label{eq:def_pbe_tot} \label{eq:def_pbe_tot}
\bE{\PBEot}{\manu{\Bas}}[\n{}{},\ttn{2}{\manu{\Bas}},\rsmu{}{\manu{\Bas}}] = \bE{\PBEot}{\Bas}
\int \n{}{}(\br{}) \equiv
\\ \bE{\PBE}{\Bas}[\n{}{},\ttn{2}{\Bas},\rsmu{}{\Bas}].
\times \be{\text{c,md}}{\sr,\PBE}\qty(\n{}{}(\br{}),\ttn{2}{\manu{\Bas}}(\br{}),s(\br{}),\zeta(\br{}),\rsmu{}{\manu{\Bas}}(\br{})) \dbr{}, \end{equation}
\end{multline}
%Such formula relies on the association between $\n{2}{\Bas}(\br{ },\br{ })$, which is the on-top pair density computed in the basis set $\Bas$ in a given point, and the local value of the range-separation parameter $\rsmu{}{\Bas}$ at the same point.
The sole distinction between \titou{PBE-UEG} and PBEot is the level of approximation of the exact on-top pair density. The sole distinction between \titou{PBE-UEG} and PBEot is the level of approximation of the exact on-top pair density.