Added RMP critical point discussion... UMP to come
This commit is contained in:
Hugh Burton
parent
6d03501307
commit
d88f50da22
@ -519,6 +519,7 @@ the spin-$\sigma$ electrons as
|
||||
\\
|
||||
\mathcal{A}^{\sigma} & = - \sin(\frac{\theta_\sigma}{2}) \Lsi + \cos(\frac{\theta_\sigma}{2}) \Rsi
|
||||
\end{align}
|
||||
\label{eq:RHF_orbs}
|
||||
\end{subequations}
|
||||
In the weak correlation regime $0 \le U \le 2t$, the angles which minimise the HF energy,
|
||||
\ie, $\pdv*{E_\text{HF}}{\theta_\sigma} = 0$, are
|
||||
@ -1194,6 +1195,28 @@ $\lambda$ values closer to the origin.
|
||||
With these insights, they regrouped the systems into new classes: i) $\alpha$ singularities which have ``large'' imaginary parts,
|
||||
and ii) $\beta$ singularities which have very small imaginary parts.\cite{Goodson_2004,Sergeev_2006}
|
||||
|
||||
%------------------------------------------------------------------%
|
||||
% Figure on the RMP critical point
|
||||
%------------------------------------------------------------------%
|
||||
\begin{figure*}[t]
|
||||
\begin{subfigure}{0.49\textwidth}
|
||||
\includegraphics[height=0.65\textwidth]{rmp_critical_point}
|
||||
\subcaption{\label{subfig:rmp_cp}}
|
||||
\end{subfigure}
|
||||
\begin{subfigure}{0.49\textwidth}
|
||||
\includegraphics[height=0.65\textwidth]{rmp_critical_point_surf}
|
||||
\subcaption{\label{subfig:rmp_cp_surf}}
|
||||
\end{subfigure}
|
||||
\caption{%
|
||||
\hugh{Modelling the RMP critical point using the asymmetric Hubbard dimer.
|
||||
(\subref{subfig:rmp_cp}) Exact critical points with $t=0$ occur on the negative real $\lambda$ axis (dashed).
|
||||
(\subref{subfig:rmp_cp_surf}) Modelling a finite basis using $t=0.1$ yields complex-conjugate EPs close to the
|
||||
real axis, giving a sharp avoided crossing on the real axis (solid).
|
||||
}
|
||||
\label{fig:RMP_cp}}
|
||||
\end{figure*}
|
||||
%------------------------------------------------------------------%
|
||||
|
||||
% RELATIONSHIP TO BASIS SET SIZE
|
||||
The existence of the MP critical point can also explain why the divergence observed by Olsen \etal\ in the \ce{Ne} atom
|
||||
and \ce{HF} molecule occurred when diffuse basis functions were included.\cite{Olsen_1996}
|
||||
@ -1245,100 +1268,92 @@ states which share the symmetry of the ground state,\cite{Goodson_2004} and are
|
||||
\label{sec:critical_point_hubbard}
|
||||
%=======================================
|
||||
|
||||
%------------------------------------------------------------------%
|
||||
% Figure on the RMP critical point
|
||||
%------------------------------------------------------------------%
|
||||
\begin{figure*}[t]
|
||||
\begin{subfigure}{0.49\textwidth}
|
||||
\includegraphics[height=0.65\textwidth]{rmp_critical_point}
|
||||
\subcaption{\label{subfig:rmp_cp}}
|
||||
\end{subfigure}
|
||||
\begin{subfigure}{0.49\textwidth}
|
||||
\includegraphics[height=0.65\textwidth]{rmp_critical_point_surf}
|
||||
\subcaption{\label{subfig:rmp_cp_surf}}
|
||||
\end{subfigure}
|
||||
\caption{%
|
||||
\hugh{RMP critical point in the Hubbard dimer.}
|
||||
\label{fig:RMP_cp}}
|
||||
\end{figure*}
|
||||
%------------------------------------------------------------------%
|
||||
|
||||
\hughDraft{%
|
||||
The simplified site basis of the Hubbard dimer makes explicitly modelling the ionisation continuum
|
||||
impossible.
|
||||
To model the auto-ionisation in the Hubbard dimer, we need to turn one of the sites into a ghost atom that will act as
|
||||
a destination for ionised electrons. We do this by considering an asymmetric Hubbard dimer where the "atomic" site
|
||||
has a negative diagonal term in the one-electron Hamiltonian, representing the nuclear attraction. This term is already
|
||||
encoded in the Initialisation section such that the core Hamiltonian is
|
||||
To model the doubly-occupied atom, we can define our reference HF state as the configuration with $\theta = 0$ and energy
|
||||
% INTRODUCING THE MODEL
|
||||
\hugh{%
|
||||
The simplified site basis of the Hubbard dimer makes explicilty modelling the ionisation continuum impossible.
|
||||
Instead, we can use an asymmetric Hubbard dimer to consider one site as a ``ghost atom'' that acts as a
|
||||
destination for ionised electrons.
|
||||
In this asymmetric model, we introduce a one-electron potential $-\epsilon$ on the left site to
|
||||
represent the attraction between the electrons and the model ``atomic'' nucleus, where we define $\epsilon > 0$.
|
||||
%The exact Hamiltonian [Eq.~\eqref{eq:H_FCI}] then becomes
|
||||
%\begin{equation}
|
||||
%\label{eq:H_FCI_Asymm}
|
||||
%\bH =
|
||||
%\begin{pmatrix}
|
||||
% U-2\epsilon & -t & -t & 0 \\
|
||||
% -t & -\epsilon & 0 & -t \\
|
||||
% -t & 0 & -\epsilon & -t \\
|
||||
% 0 & -t & -t & U \\
|
||||
%\end{pmatrix}.
|
||||
%\end{equation}
|
||||
The reference Slater determinant for a doubly-occupied atom can be represented using the RHF
|
||||
orbitals [Eq.~\eqref{eq:RHF_orbs}] with
|
||||
\begin{equation}
|
||||
E_\text{HF}(0, 0) = \frac{1}{2} (2 U - 4 \epsilon)
|
||||
\theta_{\alpha}^{\text{RHF}} = \theta_{\beta}^{\text{RHF}} = 0.
|
||||
\end{equation}
|
||||
The RMP Hamiltonian then becomes
|
||||
%and energy
|
||||
%\begin{equation}
|
||||
% E_\text{HF}(0, 0) = \frac{1}{2} (2 U - 4 \epsilon).
|
||||
%\end{equation}
|
||||
With this representation, the parametrised RMP Hamiltonian becomes
|
||||
\begin{widetext}
|
||||
\begin{equation}
|
||||
\label{eq:H_RMP}
|
||||
\bH_\text{RMP}\qty(\lambda) =
|
||||
\begin{pmatrix}
|
||||
-2 \delta \epsilon + 2U(1 - \lambda/2) & -\lambda t & -\lambda t & 0 \\
|
||||
-\lambda t & - \delta \epsilon + (1-\lambda)U & 0 & -\lambda t \\
|
||||
-\lambda t & 0 & - \delta \epsilon + (1-\lambda)U & -\lambda t \\
|
||||
0 & -\lambda t & -\lambda t & \lambda U \\
|
||||
\end{pmatrix},
|
||||
\begin{pmatrix}
|
||||
2(U-\epsilon) - \lambda U & -\lambda t & -\lambda t & 0 \\
|
||||
-\lambda t & (U-\epsilon) - \lambda U & 0 & -\lambda t \\
|
||||
-\lambda t & 0 & (U-\epsilon) -\lambda U & -\lambda t \\
|
||||
0 & -\lambda t & -\lambda t & \lambda U \\
|
||||
\end{pmatrix}.
|
||||
\end{equation}
|
||||
\end{widetext}
|
||||
}
|
||||
|
||||
\hughDraft{%
|
||||
Now let's think about the physics of the problem...
|
||||
For the ghost site to truly represent ionised electrons, we need the hopping term to vanish (or become very small).
|
||||
$U$ controls the strength of the HF repulsive potential. A stronger repulsion will encourage the electrons to be forced
|
||||
away from the "atomic" site at a less negative value of $\lambda$.
|
||||
$\epsilon$ controls the strength of attraction to the atom. A stronger attraction to the nucleus will mean that the electrons
|
||||
are more tightly bound to the atom and a more negative $\lambda$ will be needed for auto-ionisation.
|
||||
We therefore expect that the position of the RMP critical point will be controlled by the ratio $\epsilon / U$, with smaller ratios
|
||||
making ionisation occur closer to $\lambda$ origin, and making the divergence in these cases more likely.
|
||||
}
|
||||
|
||||
\hughDraft{%
|
||||
Taking the exact case with $t=0$, the RMP energies becomes
|
||||
% DERIVING BEHAVIOUR OF THE CRITICAL SITE
|
||||
\hugh{%
|
||||
For the ghost site to perfectly represent ionised electrons, the hopping term between the two sites must vanish with $t=0$.
|
||||
This limit corresponds to the dissociative regime in the asymmetric Hubbard dimer (see Ref.~\ref{Carrascal_2018}),
|
||||
and the RMP energies become
|
||||
\begin{subequations}
|
||||
\begin{align}
|
||||
E_{-} &= 2U - 2 \epsilon - U \lambda
|
||||
\\
|
||||
E_{\text{S}} &= U - \epsilon - U \lambda
|
||||
\\
|
||||
E_{+} &= U \lambda
|
||||
E_{+} &= U \lambda,
|
||||
\end{align}
|
||||
\end{subequations}
|
||||
By comparison with Fig.~\ref{fig:RMP_cp}, the critical point can be identified as by solving $E_{-} = E_{+}$,
|
||||
giving
|
||||
as shown in Fig.~\ref{subfig:rmp_cp} (dashed lines).
|
||||
The RMP critical point then corresponds to the intersection $E_{-} = E_{+}$, giving the critical $\lambda$ value
|
||||
\begin{equation}
|
||||
\lc = 1 - \epsilon / U.
|
||||
\lc = 1 - \frac{\epsilon}{U}.
|
||||
\end{equation}
|
||||
Thus, as expected, the critical point lies on the real axis and moves closer to the origin for larger
|
||||
$U / \epsilon$.
|
||||
Clearly the radius of convergence $\rc = \abs{\lc}$ is controlled directly by the ratio $\epsilon / U$,
|
||||
with a convergent RMP series occurring for $\epsilon > 2 U$.
|
||||
The on-site repulsion $U$ controls the strength of the HF potential localised around the ``atomic site'', with a
|
||||
stronger repulsion encouraging the electrons to be ionised at a less negative value of $\lambda$.
|
||||
Large $U$ can be physically interpreted as strong electron repulsion effects in electron dense molecules.
|
||||
In contrast, smaller $\epsilon$ gives a weaker attraction to the atomic site,
|
||||
representing strong screening of the nuclear attraction by core and valence electrons,
|
||||
and again a less negative $\lambda$ is required for ionisation to occur.
|
||||
Both of these factors are common in atoms on the right-hand side of the periodic table, \eg\ \ce{F},
|
||||
\ce{O}, \ce{Ne}, and thus molecules containing these atoms are often class $\beta$ systems with
|
||||
a divergent RMP series due to the MP critical point.
|
||||
}
|
||||
|
||||
\hughDraft{%
|
||||
The position of the critical point is controlled by the ratio $\epsilon / U$.
|
||||
If the magnitude of the ratio becomes greater than one, then the series diverges.
|
||||
We can interpret large $U$ as strong electron repulsion effects in electron dense molecules,
|
||||
such as \ce{F-}. A small $\epsilon$ is also likely to correspond to
|
||||
strong nuclear screening by the core and valence electrons. Both of these factors are
|
||||
common in atoms on the right-hand-side of periodic table,
|
||||
\eg\ \ce{F}, \ce{O}, \ce{Ne}, etc, as well as negatively-charged species, and so we recover the
|
||||
class $\beta$ system classification.
|
||||
}
|
||||
|
||||
\hughDraft{%
|
||||
In practical calculations, one considers the perturbation energy in a finite basis and the critical point is modelled
|
||||
as a cluster of branch points close to the real axis. While we cannot change the size of our basis, we can adjust
|
||||
the extent to which the second site behaves as a ghost by making the hopping term larger.
|
||||
When $t$ is (slightly) non-zero, our modelled critical point becomes an EP point close to the real axis with a sharp
|
||||
associated avoided crossing for real $\lambda$. If $t$ becomes larger, this avoided crossing becomes less sharp and the EPs
|
||||
move away from the real axis. This mirrors the discussion of EPs approaching the real axis in the exact basis.
|
||||
This effect is shown by the dashed lines in Fig.~\ref{fig:RMP_cp}.
|
||||
% EXACT VERSUS APPROXIMATE
|
||||
\hugh{%
|
||||
The critical point in the exact case $t=0$ lies on the real $\lambda$ axis, mirroring the behaviour of a quantum
|
||||
phase transition.\cite{Kais_2006}
|
||||
However, in practical calculations performed with a finite basis set, the critical point is modelled as a cluster
|
||||
of branch points close to the real axis.
|
||||
The use of a finite basis can be modelled in the asymmetric dimer by making the second site a less
|
||||
idealised destination for the ionised electrons with a non-zero hopping term $t$.
|
||||
Taking the small value $t=0.1$ (Fig.~\ref{subfig:rmp_cp}: solid lines), the critical point becomes a
|
||||
sharp avoided crossing with a complex-conjugate pair of EPs close to the real axis (Fig.~\ref{subfig:rmp_cp_surf}).
|
||||
In the limit $t \rightarrow 0$, these EPs approach the real axis, mirroring Sergeev's discussion on finite basis
|
||||
set representations of the MP critical point.\cite{Sergeev_2006}
|
||||
}
|
||||
|
||||
%%====================================================
|
||||
|
||||
Binary file not shown.
Binary file not shown.
Loading…
x
Reference in New Issue
Block a user