David Strubbe edited Complex DFT.tex  over 9 years ago

Commit id: b20dd9fbf3d3f9b4d7386e4b0c86e9f48df01fc6

deletions | additions      

       

\newcommand{\fdiff}[2]{\frac{\delta #1}{\delta #2}} % functional derivative  In this section we discuss the calculation of resonant electronic states by  means of the complex scaling method complex-scaling method,  as implemented in Octopus. By resonant states ``resonant states,"  we mean metastable electronic states of finite systems, such as atoms or molecules, with a characteristic energy  and lifetime.  Mathematically, resonances can be defined as poles of the scattering  matrix or cross-section at complex energies~\cite{PhysRev.56.750,Hatano01022008}.  If a pole is close to the  real energy axis axis,  it will produce a large, narrow peak in the cross-section of scattered continuum states close to that energy.  Another states.  One  way to create a resonance resonances can arise  is to apply from application of  an electric field strong enough to ionize the system through tunnelling.  Resonant states may temporarily capture incoming electrons or electrons  excited from bound states, making them important intermediate states 

outgoing component but not an incoming one.  They can be determined by solving the  time-independent Schrödinger  equation with the boundary condition that the wave wavefunction  must asymptotically have the form  \begin{align}  \phi(r) \sim \frac{\mathrm{e}^{\ii k}}{r}\quad\textrm{as}\quad r\rightarrow\infty\ , 

from DFT. However it turns out that a suitably chosen analytic  continuation of a Siegert state is localized, and this form can be used  to derive information from the state.  This is the idea beyond behind  the \emph{complex scaling}\cite{aguilarcombes,Balslev:1971ez} \emph{complex-scaling} \cite{aguilarcombes,Balslev:1971ez}  method where states and operators are represented  by means of the transformation  \begin{align}  \hat R_\theta\, \psi(\vec r) = \ee^{\ii N \theta / 2} \psi(\vec r \ee^{\ii\theta})\ ,  \end{align}  where $N$ is the number of spatial dimensions to which the scaling operation  is applied, and $\theta$ is a fixed \emph{scaling angle} which deteremines determines  the path in the complex plane along which the analytic continuation is taken.  The transformation maps the Hamiltonian to a non-Hermitian operator  $\hat H_\theta = \hat R_\theta \hat H \hat R_{-\theta}$. 

resonance~\cite{simon1973resonances,Reinhardt_1982,Ho19831}.  A typical example of a spectrum of the transformed Hamiltonian $\hat H_\theta$  is shown on in  Figure~\ref{fig:cs-spectrum}, and the corresponding potential and lowest bound and resonant states on in  Fig.~\ref{fig:cs-potential-wfs}. The bound-state energies are unchanged while the continuum rotates by  $-2 \theta$ around the origin. Finally resonances appear as isolated  eigenvalues in the fourth quadrant once $\theta$ is sufficiently large  to ``uncover'' them from the continuum.  Importantly, matrix elements (and in  particular energies) of states are independent of $\theta$ as long as  the states are localized and well presented represented  numerically --- this ensures that all physical  bound-state characteristics of the untransformed Hamiltonian are retained. 

for independent particles or  in combination with DFT and  selected  xc XC  functionals~\cite{Larsen:2013cw}. The energy functional in KS-DFT consists of several  terms that are all expressible as integrals of the density or the  wavefunctions with the kinetic operator and various potentials. 

\varphi_{\theta n}(\vec r)\\  + \ee^{-\ii\theta} \frac12  \iint \dee \vec r \, \dee \rprime \,  \frac{n_\theta(\vec r)n_\theta(\rprime)}{\Vert\vec r)n_\theta(\rprime)}{\left| \vec  r - \rprime\Vert}\\ \rprime \left|}\\  \quad+ E_\xc^\theta[n_\theta]  + \int\dee \vec r\, v_{\mathrm{ext}}(\vec r \ee^{\ii \theta}) n_\theta(\vec r)\ ,  \end{multline}  %  with the, now complex, the now-complex  electron density \begin{align}  n_\theta(\vec r) = \sum_n f_n \varphi_{\theta n}^2(\vec r)\ ,  \end{align}  with  occupation numbers $f_n$, and complex-scaled KS states $\varphi_{\theta n}(\vec r)$. Note that no complex conjugation is performed of on  the left component in matrix elements such as the density or kinetic energy.  In order to define the complex-scaled xc XC  potential, it is necessary to perform an analytic continuation procedure~\cite{Larsen:2013cw}. In standard DFT DFT,  the KS equations are obtained by taking the derivative of the energy functional with respect  to the wavefunctions. Solving the equations corresponds to searching  for a stationary point, with the idea that this minimizes the energy. 

continued pseudopotentials, and where the Hartree potential  \begin{align}  v_\Ha^\theta(\vec r) &=  \ee^{-\ii\theta}\int\dee \rprime\, \frac{n_\theta(\vec r')}{\Vert r')}{\left|  \vec r'-\vec r\Vert} r\right|}  \end{align}  is determined by solving the Poisson equation defined by the complex density.  Together with the xc potential 

stationary points~\cite{WM07} of the  functional~\cite{Whitenack_2010,WW11}.  Fig.~\ref{fig:cs-ionization-He} shows calculated ionization rates for the He~1s~state He~1$s$~state  in a uniform Stark-type electric field as a function of field strength.  In the limit of weak electric fields, the simple approximation  by Ammosov, Delone and Krainov~\cite{adk}, whichis  depends only on the ionization potential, approaches the accurate reference calculation  by Scrinzi and co-workers~\cite{PhysRevLett.83.706}.  This demonstrates that the ionization rate is determined largely by the 

Meanwhile exact exchange, which is known to produce  accurate ionization energies, predicts ionization rates  much closer to the reference calculation. The key property   of the xc XC  functional that allows accurate determination of decay rates from complex-scaled DFT therefore appears to be that it must yield  accurate ionization potentials, which is linked to its ability  to reproduce the correct asymptotic form of the potential at large distances