readme.tex

已经编译好的levoo程序

\begin{itemize}
\item Wavefunctions (in reciprocal representation) are to be stored on a disk at the
end of the band-structure run as a sequence of files \texttt{fort.20}, \texttt{fort.21},
\texttt{fort.22}, etc; one file per each \textbf{k}-point in the same order
as they are given in the \texttt{fort.15} file.
\item All eigenvalues ordered properly in ascending order should be delivered after
the band-structure run to the file \texttt{band.out} which has the following
format: {[}k-point number, five eigenvalues{]} on every line. Note that this
is compulsory as the utility \textbf{take} (see below) does it anyway. It is
important to have, however, properly ordered eigenvalues on the last \textbf{CASTEP}
iteration.
\item The electronic density should be in \texttt{fort.16} file.
\end{itemize}
To compile the code you must prepare a standard file \texttt{param.inc} in your
\emph{}\textit{\emph{working}} directory and then submit a shell-script \textbf{comp}
(supplied). After the compilation is completed, you will find in your working
directory two executables: \textbf{castepx} which is the \textbf{CASTEP} itself
and \textbf{check} which is a check utility described below. 


\subsection{Utility \textit{check}}

This utility simulates a run of \textbf{CASTEP}: it reads all the data including
pseudopotentials, density, etc., generates plane waves, etc. and then stops
when all the preliminary work is done. If it reveals that some parameters set
in the \texttt{param.inc} file are insufficient, it will give a FATAL message
with an advice about the least values for them. At the same time it gives error
messages if it cannot open a file or cannot read it successfully. To compile
the utility, submit the shell-script \textbf{check.comp} from your current directory


\subsection{Utility \textit{take}}

This utility \textit{\emph{is used to assess the convergence of the band-structure
run of}} \textbf{\textit{\emph{CASTEP}}} \textit{\emph{and}} is compiled from
your current working directory using a shell-script \textbf{take.comp}. Then,
you can either run it with the name of the \textbf{CASTEP} output in the command
line or without it in which case you will be asked to specify this name explicitly:

\begin{verbatim}
 take   [name of the CASTEP output file]
\end{verbatim} 

The utility gives the following information about each \textbf{CASTEP} band-structure
iteration: 

\begin{itemize}
\item actual iteration number \textit{in the file} and the \textbf{CASTEP} iteration
number; the former number differs from the latter if you keep not all the output
in your output file (e.g., only several last iterations);
\item then, for each \textbf{\( \mathbf{k} \)}-point (\texttt{NKP}) per line it gives
\begin{equation}
\label{9}
delta=\frac{1}{{\texttt {NBANDS}}}\sum _{n=1}^{{\texttt {NBANDS}}}\left| \epsilon _{n\mathbf{k}}^{{\texttt {NKP}}}-\epsilon _{n\mathbf{k}}^{{\texttt {NKP-1}}}\right| 
\end{equation}
 which shows the convergence of the eigenvalues \( \epsilon _{n\mathbf{k}}^{{\texttt {NKP}}} \)
on the current (\texttt{NKP}-th) iteration with respect to the previous one
((\texttt{NKP-1})-th); it also compares the criteria with the previous iteration
and checks if all the eigenvalues are properly ordered. If they are not ordered
properly, it tells you how many weird eigenvalues have been met. (Note that
if \textbf{CASTEP} does the subspace rotation, this information is redundant.) 
\end{itemize}
Usually, \( delta=10^{-3} \) is fine.

At the end of the output you are asked to produce a file \texttt{band.0} with
eigenvalues (of the same format as \texttt{band.out}) from \textit{any} desired
iteration. This option allows you to check the convergence of the DOS with the
iteration number by preparing \texttt{band.out} files from different iterations
and subsequently running \textbf{lev00}.


\subsection{Utilities \texttt{\textit{lev1}} \textit{(old) and} \texttt{\textit{lev2}}
\textit{(new)}}

Either utility is used to prepare data files for \textbf{lev00} for the calculation
of the partial charge density as well as DOS/LDOS using \textbf{CASTEP} wavefunctions.
To compile this utility you should run the shell-script \textbf{lev1.comp} from
the current directory using the same \texttt{param.inc} file. \textbf{lev2}
should be compiled in directories LEV2\_f77 (f77 version, compile every time
you get a new \texttt{param.inc} file) or LEV\_f90 (f90 version, to be compiled
only once).

First of all, run \textbf{CASTEP} in the band-structure mode to write the wavefunctions.
Note that the \( \mathbf{k} \)-points mesh as given by \textbf{tetr} need to
be used for the DOS/LDOS calculation. Then you can run \textbf{lev1.} It is
an interactive menu-driven program and five options are available: DOS, DOp,
SRF and MAP, and at the end of the run a file \texttt{psi2.{[}option{]}}is produced
needed for the subsequent run of the plotter \textbf{lev00}. Note that the utility
\textbf{lev1} needs the file \texttt{fort.14} which contains all information
necessary to perform the Fast Fourier Transform (FFT) from the reciprocal to
the real space.

All states available NBANDS can participate. However, it is possible to do a
sophisticated job by creating \emph{islands} of bands, each island being a continuous
manifold of states which can overlap, and then run the calculation for every
island. This way I can do a specific job on particular states. If all states
are to be included, just specify one island with all states from 1 to NBANDS.

The five options available are listed below:

\begin{enumerate}
\item \texttt{\large DOS}: \( s \)-projected DOS as in Eqs. (\ref{4}) and (\ref{6})
(with \( l=0 \)) is calculated using conserving algorithm for the space integration.
The projection is made either on a single Slater type AO, 
\begin{equation}
\label{10}
R(r)=N(\xi )r^{n-1}e^{-\xi r}
\end{equation}
or on a linear combination of Gaussians
\begin{equation}
\label{11}
R(r)=\sum ^{N_{G}}_{i=1}C_{i}N(\alpha _{i})e^{-\alpha _{i}r^{2}}
\end{equation}
Here \( N \) is the normalisation of a single orbital which depends on the
exponential, \( \xi  \) or \( \alpha _{i} \). Noninteractive option. All the
information needed should be provided in the file \texttt{fort.14.DOS} (see
below). Up to \texttt{Ntask0} spheres (set in \texttt{dos\_task.inc}) can be
defined for one run of \textbf{lev1. }
\item \texttt{\large PRO}: local DOS, Eqs. (\ref{4}) and (\ref{5}), and \( s,p,d \)-projected
DOS, Eqs. (\ref{4}) and (\ref{6}), are calculated using nonconserving algorithm
for the space integration. The projection is made on a single AO; its radial
part is made either of a single Slater type AO, or of a linear combination of
Gaussians. Noninteractive option. All the information needed should be provided
in the file \texttt{fort.14.PRO} (see below). Up to \texttt{Ntask0} spheres
can be defined for one run of \textbf{lev1. }
\item \texttt{\large DOp}: this is DOS projected on a layer, see Eqs. (\ref{4}) and
(\ref{7}). Interactive option: you will be asked about each layer, its position
and thickness. Up to \texttt{Ntask0} layers can be specified for one run of
\textbf{lev1.} Islands of bands \( n \) should be provided in the file \texttt{fort.14a}
(see below).
\item \texttt{\large MAP}: partial electronic density as in Eq. (\ref{1}) or (\ref{2}).
Islands of bands \( n \) should be provided in the file \texttt{fort.14a} (see
below).
\item \texttt{\large SRF}: partial electronic density which is 2D integrated over
the plane parallel to the slab, i.e. 
\begin{equation}
\label{12}
\rho _{n}(i_{3})=\frac{1}{N_{1}N_{2}}\sum _{i_{1}=1}^{N_{1}}\sum _{i_{2}=1}^{N_{2}}\rho _{n}(i_{1},i_{2},i_{3})).
\end{equation}
where \( N_{1} \) and \( N_{2} \) are the number of grid points along vectors
\( \mathbf{a}_{1} \) and \( \mathbf{a}_{2} \) of the cell and \( i_{3} \)
runs from 1 to \( N_{3} \) (\( N_{1} \), \( N_{2} \), \( N_{3} \) correspond
to NGX, NGY, NGZ in \texttt{param.inc}). You will be asked to provide islands
of bands (as in the file \texttt{fort.14.SRF}) and an output file \texttt{psi2.SRF}
will be produced. Islands of bands \( n \) should be provided in the file \texttt{fort.14a}
(see below).
\end{enumerate}
Formats of the input files mentioned in this section are provided below: 

\begin{itemize}
\item \texttt{fort.14a}: islands of states are provided, i.e. 

\begin{itemize}
\item number of islands of states (1 integer)
\item the first and the last states for the 1st island (2 integers)
\item the first and the last states for the 2nd island (2 integers)
\item etc.
\end{itemize}
\item \texttt{fort.14.DOS}: 

\begin{itemize}
\item islands of states as above;
\item number of spheres (1 integer);
\item then for each sphere:

\begin{itemize}
\item on 1 line: Cartesian coordinates of the sphere (in \AA), sphere radius (in \AA),
1 or 0 (for Slater or Gaussian AO, respectively)
\item if Slater, Eq. (\ref{10}), then on the next line: \( n \) and \( \xi  \)
(in \AA\( ^{-1} \))
\item if Gaussian, Eq. (\ref{11}), then on the next line: \( N_{G} \) and on the
following line all the coefficients and exponentials (in \AA\( ^{-2} \)) are
provided as \( (C_{i},\, \alpha _{i},i=1,...,N_{G}) \).
\end{itemize}
\end{itemize}
\item \texttt{fort.14.PRO}: 

\begin{itemize}
\item islands of states as above;
\item number of spheres (1 integer);
\item then for each sphere:

\begin{itemize}
\item on 1 line: Cartesian coordinates of the sphere (in \AA), sphere radius (in \AA),
1 or 0 (for Slater or Gaussian AO, respectively), one integer \( K \) corresponding
to the fine grid \( K^{3} \) of the nonconserving method, and three flags (0
for 'no' and 1 for 'yes') for whether to do \( s \), \( p \) and \( d \)
projected DOS.
\item then separately for every angular momenta with nonzero flag: specify the radial
part of the orbital \( R(r) \) (see \texttt{fort.14.DOS})
\end{itemize}
\end{itemize}
\item \texttt{fort.14.DOp}: 

\begin{itemize}
\item islands of states as in other cases;
\item number of layers (1 integer);
\item then for each layer you provide the following information (one line per layer):

\begin{itemize}
\item Cartesian position of the layer on the \( z \)-axis (in \AA);
\item layer thickness (in \AA);
\item 0 for conserving and 1 for non-conserving;
\item numbers of grid points along each of the x,y,z directions necessary for the
non-conserving method (3 integers); note that in the case of the conserving
method this data will be ignored.
\end{itemize}
\end{itemize}
\end{itemize}

\subsection{How to split a large job into a set of smaller jobs. Utility \texttt{\textit{sum}}}

If your \textbf{CASTEP} job is been estimated to be too large to run with all
\( \mathbf{k} \)-points simultaneously, you can split it into parts. Final
files \texttt{band.out} and \texttt{psi2.{[}option{]}} are then compiled together
by a special utility \textbf{sum} (hopefully, it works although it has been
long time ago!).

Let us consider this case in detail: 

\begin{itemize}
\item First of all, you should split the \textbf{k}-point sampling into subsets. For
example, a 10 \textbf{k}-points job can be split into a set of 10 jobs with
1 \textbf{k}-point each or into a set of 3 jobs with 3, 3 and 4 \textbf{k}-points
each, etc.
\item Run \textbf{CASTEP} for each \textbf{k}-point subset and then rename all the
output files, namely \texttt{band.out} and \texttt{psi2.{[}option{]}}, as follows:
\texttt{band.out.1, band.out.2,} etc. and \texttt{psi2.DOS.1}, \texttt{psi2.DOS.}2,
etc.
\item Compile the \textbf{sum} routine using a shell-script \textbf{sum.comp} (provided)
from your working directory and then run it as \begin{verbatim}
 sum  [number of subsets] 
\end{verbatim} The number of subsets is optional and will be asked if not given.
To compile the \textbf{sum} utility you can use any of the \texttt{param.inc}
files used to compile any of your CASTEP codes since the number of \textbf{k}-points,
\texttt{NKPTS}, does not matter; only the grid size and the number of bands
do matter.
\end{itemize}
In the case that the option MAP is on, you should be a bit more careful. It
is a good idea to refrain from running \textit{all} CASTEP jobs one after another
because this may lead to a collection of very large \{\texttt{psi2.MAP.N}\}
files. To save the disk space, you can run the first two jobs, then \textbf{sum}