~nickpapior/siesta/trunk-buds-format0.92 : revision 527.1.17

205

206

\item

207

``Unconstrained minimization approach for electronic computations

208

that scales linearly with system size"

208

that scales linearly with system size''

209

P. Ordej\'on, D. A. Drabold, M. P. Grumbach and R. M. Martin,

210

Phys. Rev. B \textbf{48}, 14646 (1993);

211

``Linear system-size methods for electronic-structure calculations"

211

``Linear system-size methods for electronic-structure calculations''

212

Phys. Rev. B \textbf{51} 1456 (1995), and references therein.

213

214

Description of the order-\textit{N} eigensolvers

216

217

\item

218

``Self-consistent order-$N$ density-functional

219

calculations for very large systems"

219

calculations for very large systems''

220

P. Ordej\'on, E. Artacho and J. M. Soler,

221

Phys. Rev. B \textbf{53}, 10441, (1996).

222

223

Description of a previous version of this methodology.

224

225

\item

226

``Density functional method for very large systems with LCAO basis sets"

226

``Density functional method for very large systems with LCAO basis sets''

227

D. S\'anchez-Portal, P. Ordej\'on, E. Artacho and J. M. Soler,

228

Int. J. Quantum Chem., \textbf{65}, 453 (1997).

229

230

Description of the present method and code.

231

232

\item

233

``Linear-scaling ab-initio calculations for large and complex systems"

233

``Linear-scaling ab-initio calculations for large and complex systems''

234

E. Artacho, D. S\'anchez-Portal, P. Ordej\'on, A. Garc\'{\i}a and

235

J. M. Soler, Phys. Stat. Sol. (b) \textbf{215}, 809 (1999).

236

238

used in the code, and brief review of applications as of March 1999.

239

240

\item

241

``Numerical atomic orbitals for linear-scaling calculations"

241

``Numerical atomic orbitals for linear-scaling calculations''

242

J. Junquera, O. Paz, D. S\'anchez-Portal, and E. Artacho, Phys. Rev. B

243

\textbf{64}, 235111, (2001).

244

245

Improved, soft-confined NAOs.

246

247

\item

248

``The \siesta\ method for ab initio order-$N$ materials simulation"

248

``The \siesta\ method for ab initio order-$N$ materials simulation''

249

J. M. Soler, E. Artacho, J.D. Gale, A. Garc\'{\i}a, J. Junquera, P. Ordej\'on,

250

and D. S\'anchez-Portal, J. Phys.: Condens. Matter \textbf{14}, 2745-2779 (2002)

251

253

254

\item

255

``Computing the properties of materials from first principles

256

with \siesta", D. S\'anchez-Portal, P. Ordej\'on,

256

with \siesta'', D. S\'anchez-Portal, P. Ordej\'on,

257

and E. Canadell, Structure and Bonding \textbf{113},

258

103-170 (2004).

259

391

the whole \siesta\ tree. Go to your favorite working directory

392

and:

393

394

\texttt{\$ mkdir h2o}

395

396

\texttt{\$ cd h2o}

397

398

\texttt{\$ cp path-to-package/Examples/H2O/h2o.fdf .}

394

\begin{fdfexample}

395

$ mkdir h2o

396

$ cd h2o

397

$ cp path-to-package/Examples/H2O/h2o.fdf

398

\end{fdfexample}

399

%$

399

400

401

You need to make the siesta executable visible in your path.

401

402

You can do it in many ways, but a simple one is

402

403

\texttt{ln -s path-to-package/Obj/siesta .}

404

\begin{fdfexample}

405

$ ln -s path-to-package/Obj/siesta

406

\end{fdfexample}

407

%$

404

408

405

409

\noindent

406

410

We need to generate the required pseudopotentials.

8375

8379

8376

8380

\section{TRANSIESTA}

8377

8381

8378

The present \siesta\ release includes the possibility of

8379

performing calculations of electronic transport properties using the

8380

\tsiesta\ method. This Section describes how to compile the

8381

code to be able to use these capabilities, and a reference guide to

8382

the relevant \fdflib\ options. We describe here only the additional options

8383

available for \tsiesta\ calculations, while the rest of the Siesta

8384

functionalities and variables are described in the previous sections

8385

of this User's Guide.

8382

\siesta\ includes the possibility of performing calculations of

8383

electronic transport properties using the \tsiesta\ method. This

8384

Section describes how to compile the code to be able to use these

8385

capabilities, and a reference guide to the relevant \fdflib\

8386

options. We describe here only the additional options available for

8387

\tsiesta\ calculations, while the rest of the Siesta functionalities

8388

and variables are described in the previous sections of this User's

8389

Guide.

8386

8390

8387

8391

\subsection{Brief description}

8388

8392

8586

8590

8587

8591

\item%

8588

8592

Importantly the $k$-point sampling need typically be much higher in

8589

a \tbtrans\ calculation to achieve a converged transmission function.

8593

a \tbtrans\ calculation to achieve a converged transmission

8594

function.

8595

8590

8596

\end{itemize}

8591

8597

8592

8598

\subsection{Electrodes}

8611

8617

against the electrode coordinates and the program stops if there is a

8612

8618

mismatch to a certain precision ($10^{-4}\,\mathrm{Bohr}$).

8613

8619

8620

8614

8621

\paragraph{Principal layer interactions} %

8615

8622

\index{transiesta!electrode!principal layer}%

8623

8616

8624

It is \emph{extremely} important that the electrodes only interact

8617

8625

with one neighboring supercell due to the self-energy

8618

8626

calculation. \tsiesta\ will print out a block as this

8676

8684

second for a 3 electrode setup, and so on. See the help (\program{-h})

8677

8685

for the program for additional options.

8678

8686

8687

8679

8688

\subsubsection{General options}

8680

8689

8681

8690

\begin{fdfentry}{SolutionMethod}[string]<transiesta>

8856

8865

Control whether the forces are calculated. If \emph{not} \tsiesta\

8857

8866

will use slightly less memory and the performance slightly

8858

8867

increased.

8859

%

8860

8868

8861

8869

\end{fdflogicalT}

8862

8870

8863

8871

\begin{fdfentry}{TS!ChargeCorrection}[string]<none|buffer|fermi>

8864

8872

8865

Any excess/deficiency of charge can be re-adjusted after each \tsiesta\ cycle to reduce

8866

charge fluctuations in the cell.

8873

Any excess/deficiency of charge can be re-adjusted after each

8874

\tsiesta\ cycle to reduce charge fluctuations in the cell.

8867

8875

8868

8876

\begin{fdfoptions}

8869

8877

8876

8884

8877

8885

\option[fermi] %

8878

8886

Correct the filling by calculating a new Fermi-level (reference energy). \\

8879

We approximate the contribution to be constant around the Fermi level and find

8887

We approximate the contribution to be constant around the Fermi

8888

level and find

8880

8889

\begin{equation}

8881

8890

\label{eq:fermi-shift}

8882

8891

\mathrm{d}E_F = \frac{Q'-Q}{Q|_{E_F}},

8883

8892

\end{equation}

8884

where $Q'$ is the charge from a converged \tsiesta\ calculation and $Q|_{E_F}$ is the

8885

equilibrium charge at the current Fermi level, $Q$ is the supposed charge to reside in

8886

the calculation. Fermi correction utilises \ref{eq:fermi-shift} for the first correction

8887

and all subsequent corrections are based on a cubic spline interpolation to much faster

8888

converge to the ``correct'' Fermi level.

8893

where $Q'$ is the charge from a converged \tsiesta\ calculation

8894

and $Q|_{E_F}$ is the equilibrium charge at the current Fermi

8895

level, $Q$ is the supposed charge to reside in the

8896

calculation. Fermi correction utilizes Eq. \eqref{eq:fermi-shift} for

8897

the first correction and all subsequent corrections are based on a

8898

cubic spline interpolation to much faster converge to the

8899

``correct'' Fermi level.

8889

8900

8890

8901

This method will create a file called \file{TS\_FERMI} and only works with the BTD

8891

8902

solver.

9057

9068

all described in the MUMPS manual.

9058

9069

9059

9070

The following list of orderings are available (without detailing

9060

their differences):

9061

\fdf*{auto}, \fdf*{AMD}, \fdf*{AMF}, \fdf*{SCOTCH}, \fdf*{PORD}, \fdf*{METIS}, \fdf*{QAMD}.

9071

their differences): %

9072

\fdf*{auto}, \fdf*{AMD}, \fdf*{AMF}, \fdf*{SCOTCH}, \fdf*{PORD},

9073

\fdf*{METIS}, \fdf*{QAMD}.

9062

9074

9063

9075

\end{fdfentry}

9064

9076

9079

9091

9080

9092

\end{fdfentry}

9081

9093

9094

\subsubsection{Poisson solution for fixed boundary conditions}

9095

9096

\tsiesta\ requires fixed boundary conditions and forcing this is an

9097

intricate and important detail.

9098

9099

\begin{fdfentry}{TS!Poisson.Position}[string]<cell|central>

9100

9101

\note this only have meaning for a two electrode calculation with

9102

both electrodes having the same semi-infinite direction.

9103

9104

Define how the correction of the Poisson equation is

9105

superimposed. The default is to only apply the linear correction

9106

within the boundaries of the electrodes.

9107

9108

\end{fdfentry}

9109

9110

\begin{fdfentry}{TS!Hartree.Fix}[string]<plane|elec-plane|elec-box>

9111

9112

As the fixed boundary conditions requires a fixed reference

9113

potential. For two electrode calculations this defaults to taking

9114

the plane at one of the electrodes basal-planes (\fdf*{plane}).

9115

9116

For anything but two electrodes this defaults to \fdf*{elec-plane}

9117

because the plane should be at a fixed position in the cell.

9118

9119

\note generally this shouldn't need to be changed.

9120

9121

\end{fdfentry}

9122

9123

\begin{fdfentry}{TS!Hartree.Fix!Frac}[real]<$1.$>

9124

9125

Fraction of the correction that is applied.

9126

9127

\note this is an experimental feature and shouldn't be used.

9128

9129

\end{fdfentry}

9130

9131

9082

9132

\subsubsection{Electrode description options}

9083

9133

9084

9134

As \tsiesta\ supports $N$ electrodes you need to specify all

9282

9332

9283

9333

\end{fdfentry}

9284

9334

9335

\begin{fdfentry}{TS!Elecs!DM.Init}[string]<diagon|bulk>

9336

9337

The density matrix elements in the electrodes may be forcefully set

9338

to the bulk values by reading in the DM of the corresponding

9339

electrode. This may be set to \fdf*{bulk} to forcefully set the bulk

9340

values.

9341

9342

\note this should only be set to \fdf*{bulk} for equilibrium

9343

calculations.

9344

9345

\end{fdfentry}

9346

9347

\begin{fdfentry}{TS!Elecs!Coord.EPS}[length]<$10^{-4}\,\mathrm{Bohr}$>

9348

9349

When using Bloch expansion of the self-energies one may experience

9350

difficulties in obtaining perfectly aligned electrode coordinates.

9351

9352

This parameter controls how strict the criteria for equivalent

9353

atomic coordinates is. If \tsiesta\ crashes due to mismatch between

9354

the electrode atomic coordinates and the scattering region

9355

calculation, one may increase this criteria. This should only be

9356

done if one is sure that the atomic coordinates are almost similar

9357

and that the difference in electronic structures of the two may be

9358

negligible.

9359

9360

\end{fdfentry}

9361

9362

9285

9363

\subsubsection{Chemical potentials}

9286

9364

\label{sec:ts:chem-pot}

9287

9365

9725

9803

9726

9804

\end{description}

9727

9805

9728

\subsection{Utilities for analysis: \texorpdfstring{\tbtrans}{TBtrans}}

9806

\subsection{Utilities for analysis:

9807

\texorpdfstring{\tbtrans}{TBtrans}}

9729

9808

\index{tbtrans@\textbf{tbtrans}}

9730

9809

9731

The \texttt{tbtrans} code can be found in the directory \program{Util/TBTrans}.

9732

9733

A much optimized version of \texttt{tbtrans} is found in the directory

9734

\texttt{Util/TBTtrans\_rep} which is recommended to be used. Only in this version can the

9735

repetition be used. It has been completely rewritten to be as fast as possible as well as

9736

keeping the maintenance to a minimum. This version has been completely rewritten by Nick

9737

Papior Andersen (referred to by NPA in the following).

9738

9739

It is used in order to obtain, in a post-processing

9740

way, the transport properties after a \tsiesta\ run. It was

9741

developed by M. Brandbyge. It has been made to conform

9742

with the modifications introduced in \tsiesta\ in late October 2012.

9743

9744

In order to run it, it requires the electrode's \sysfile{TSHS} files

9745

(may be just one file if the left and right electrodes are equal),

9746

and the

9747

scattering region's \sysfile{TSHS} file. These are generated as explained

9748

above. The location of these files are specified by the (already

9749

discussed) \textbf{TS.HSFileLeft}, \textbf{TS.HSFileRight} input options,

9750

and by:

9751

\begin{description}

9752

\itemsep 10pt

9753

\parsep 0pt

9754

9755

\item [\textbf{TS.TBT.HSFile}](\textit{string}): Scattering region \sysfile{TSHS} file.

9756

\index{TS.HSFileLeft@\textbf{TS.HSFileLeft}}

9757

9758

\textit{Default value:} \texttt{\textit{SystemLabel}.TSHS}

9759

\end{description}

9760

respectively for the left and right electrodes and the scattering

9761

region \sysfile{TSHS} file.

9762

9763

The energy scale in \texttt{tbtrans} is shifted so that the Fermi level

9764

of the system, if no voltage were applied, is zero. When computing the

9765

transmission function of a zero bias calculation, the transmission at

9766

the Fermi level is then given by $T(E=0)$. When there is a finite bias,

9767

the Fermi energy of the left electrode is placed at $V/2$, and that of

9768

the right electrode at $-V/2$.

9769

9770

The voltage is specified by \textbf{TS.Voltage}. The energy window and

9771

number of points for the computation of the transmission function is

9772

specified by

9773

\begin{description}

9774

\itemsep 10pt

9775

\parsep 0pt

9776

9777

\item [\textbf{TBT\_kgrid\_Monkhorst\_Pack}] \index{TBT\_kgrid\_Monkhorst\_Pack@\textbf{TBT\_kgrid\_Monkhorst\_Pack}}

9778

\index{kgrid\_Monkhorst\_Pack@\textbf{kgrid\_Monkhorst\_Pack}}

9779

(\textit{block}): Block equivalent

9780

to \textbf{kgrid\_Monkhorst\_Pack}, however, this is used to have a separate

9781

k-sampling in the tbtrans utility. If it does not exist it will use

9782

\textbf{kgrid\_Monkhorst\_Pack}.

9783

9784

\textbf{Notice}: \textbf{TBT\_kgrid\_Monkhorst\_Pack} can only be used with the version

9785

created by NPA.

9786

9787

\textit{Default value:} \textbf{kgrid\_Monkhorst\_Pack}

9788

9789

\item [\textbf{TS.TBT.Emin}](\textit{physical}): Lowest energy value of the

9790

computed transmission function. \index{TS.TBT.Emin@\textbf{TS.TBT.Emin}}

9791

9792

\textit{Default value:} \texttt{-2.0 eV}

9793

9794

\item [\textbf{TS.TBT.Emax}](\textit{physical}): Highest energy value of the

9795

computed transmission function. \index{TS.TBT.Emax@\textbf{TS.TBT.Emax}}

9796

9797

\textit{Default value:} \texttt{2.0 eV}

9798

9799

\item [\textbf{TS.TBT.NPoints}](\textit{integer}): Number of energy points

9800

of the transmission function between \textbf{TS.TBT.Emin} and \textbf{TS.TBT.Emax}. \index{TS.TBT.NPoints@\textbf{TS.TBT.NPoints}}

9801

9802

\textit{Default value:} \texttt{100}

9803

9804

\item [\textbf{TS.TBT.CalcElectrodeValenceBandBottom}](\textit{logical}):

9805

If true will calculate the valence band bottom of the electrode.

9806

9807

\textbf{Notice}: Can only be used with the NPA version.

9808

9809

\textit{Default value:} \texttt{true}

9810

9811

\item [\textbf{TS.TBT.COOP}](\textit{logical}): Calculate the COOP on the PDOS region.

9812

Creates the files:

9813

\begin{itemize}

9814

\item \sysfile{COOP}

9815

\item \sysfile{COOPL}

9816

\item \sysfile{COOPR}

9817

\end{itemize}

9818

9819

\textbf{Notice}: Can only be used with the NPA version.

9820

9821

\textit{Default value:} \texttt{false}

9822

9823

\item [\textbf{TS.TBT.AtomPDOS}](\textit{logical}): Calculate the \emph{onsite} projected

9824

density of states in the PDOS region. Creates the files:

9825

\begin{itemize}

9826

\item \sysfile{TOTDOS}

9827

\item \sysfile{ORBDOS}

9828

\end{itemize}

9829

9830

Notice, that this will not sum up to PDOS in \file{.TRANS}, as this is onsite density of

9831

states.

9832

9833

\textbf{Notice}: Can only be used with the NPA version.

9834

9835

\textit{Default value:} \texttt{false}

9836

\end{description}

9837

Note that it is important to specify the voltage, since this

9838

information is not stored in the \file{.TSHS} files. The current will be

9839

computed using the resulting transmission function, so be sure to make

9840

it suited for the integration in the bias window (the energy window

9841

defined by \textbf{TS.TBT.Emin} and \textbf{TS.TBT.Emax} being bigger than

9842

or equal to the applied bias).

9843

9844

The k-point sampling is defined by the \textbf{TBT\_kgrid\_Monkhorst\_Pack}

9845

block, and if that is non-existing it will use the

9846

\textbf{kgrid\_Monkhorst\_Pack} block. The averaged (over $k$-points)

9847

transmission function is printed in the file \sysfile{AVTRANS}.

9848

9849

An additional options is:

9850

\begin{description}

9851

\item [\textbf{TS.TBT.NEigen}](\textit{integer}): Number of eigenvalues of the transmission

9852

matrix to be computed. If larger than 0 \file{.TEIG} and \file{.AVTEIG} will be created

9853

which holds the $k$-point eigenchannels and the averaged eigenchannels,

9854

respectively. \index{TS.TBT.NEigen@\textbf{TS.TBT.NEigen}}

9855

9856

\textit{Default value:} \texttt{0}

9857

\end{description}

9858

9859

9860

To summarize, here we give a list of the parameters read by \tbtrans\

9861

from the input file (the fdf flags, NPA indicates the version

9862

created by Nick Papior Andersen):

9863

\begin{itemize}

9864

\item \textbf{ElectronicTemperature}

9865

\item \textbf{TS.Voltage}

9866

\item \textbf{TS.TBT.ReUseGF} (NPA)

9867

\item \textbf{TS.UseBulkInElectrodes} (NPA)

9868

\item \textbf{TBT\_kgrid\_Monkhorst\_Pack} (block) has precedence of

9869

\textbf{kgrid\_Monkhorst\_Pack} (NPA)

9870

\item \textbf{kgrid\_Monkhorst\_Pack} (block)

9871

\item \textbf{TS.HSFileLeft}

9872

\item \textbf{TS.HSFileRight}

9873

\item \textbf{TS.TBT.HSFile}

9874

\item \textbf{TS.TBT.Emin}

9875

\item \textbf{TS.TBT.Emax}

9876

\item \textbf{TS.TBT.NPoints}

9877

\item \textbf{TS.TBT.NEigen}

9878

\item \textbf{TS.TBT.CalcIEig}

9879

\item \textbf{TS.TBT.PDOSFrom}

9880

\item \textbf{TS.TBT.PDOSTo}

9881

\item \textbf{TS.TBT.COOP} (NPA)

9882

\item \textbf{TS.TBT.AtomPDOS} (NPA)

9883

\item \textbf{TS.TBT.CalcElectrodeValenceBandBottom} (NPA)

9884

\item \textbf{TS.BufferAtomsLeft}, \textbf{TS.BufferAtomsRight}

9885

\item \textbf{TS.NumUsedAtomsLeft}, \textbf{TS.NumUsedAtomsRight}

9886

\item \textbf{TS.ReplicateA1Left}, \textbf{TS.ReplicateA2Left} (NPA)

9887

\item \textbf{TS.ReplicateA1Right}, \textbf{TS.ReplicateA2Right} (NPA)

9888

\item \textbf{SpinPolarized}

9889

\end{itemize}

9890

9891

\subsubsection{Output}

9892

9893

\texttt{tbtrans} generates several output files. The output files are

9894

named \texttt{<SystemLabel>.\emph{ext}}, defined using the SystemLabel

9895

FDF command, and .\emph{ext} depends on the type of the output.

9896

9897

\begin{description}

9898

\itemsep 10pt

9899

\parsep 0pt

9900

9901

\item[\sysfile{TRANS}]: The transmission for each $k$ point,

9902

as well as the total density of states in the scattering region and the projected

9903

density of states for all orbitals in the region specified by \fdf{TS.TBT.PDOSFrom}

9904

and \fdf{TS.TBT.PDOSTo}.

9905

9906

\item[\sysfile{AVTRANS}]: The $k$ point averaged

9907

transmission and density of states according to \file{.TRANS}.

9908

9909

\item[\sysfile{LDOS}]: The $k$ point averaged and energy weighted density of states of

9910

the left electrode.

9911

9912

\item[\sysfile{RDOS}]: The $k$ point averaged and energy weighted density of states of

9913

the right electrode.

9914

9915

\end{description}

9916

9917

\subsubsection{Compiling \texorpdfstring{\tbtrans}{TBtrans}}

9918

9919

In the \texttt{Util/TBTrans} or \texttt{Util/TBTrans\_rep} (for the NPA version)

9920

directory,

9921

simply type \texttt{make} if your main \siesta\ compilation

9922

directory is the top \texttt{Obj} directory. If you have used another

9923

object directory \texttt{MyObjDir}, type \texttt{make OBJDIR=MyObjDir}.

9810

Please see the separate \tbtrans\ manual.

9924

9811

9925

9812

\section{ANALYSIS TOOLS}

9926

9813

9932

9819

\section{SCRIPTING}

9933

9820

9934

9821

In the \texttt{Util/Scripting} directory we provide an experimental

9935

python scripting framework built on top of the 'Atomic Simulation

9936

Environment' (see \texttt{https://wiki.fysik.dtu.dk/ase2}) by the Campos

9822

python scripting framework built on top of the ``Atomic Simulation

9823

Environment'' (see \texttt{https://wiki.fysik.dtu.dk/ase2}) by the Campos

9937

9824

group at DTU, Denmark.

9938

9825

9939

9826

(NOTE: ``ASE version 2'', not the new version 3, is needed)

10013

9900

10014

9901

\end{itemize}

10015

9902

9903

10016

9904

\section{REPORTING BUGS}

10017

\index{bug reports} Your assistance is essential to help improve the

10018

program. If you find any problem, or would like to offer a suggestion

10019

for improvement, please follow the instructions in the file \texttt{Docs/REPORTING\_BUGS}

9905

\index{bug reports}

9906

9907

Your assistance is essential to help improve the program. If you find

9908

any problem, or would like to offer a suggestion for improvement,

9909

please follow the instructions in the file

9910

\texttt{Docs/REPORTING\_BUGS}.

9911

9912

Since \siesta\ has moved to Launchpad you may also follow the

9913

instructions presented at:

9914

\url{https://answers.launchpad.net/siesta/+faq/2779}.

9915

10020

9916

10021

9917

10022

9918

\section{ACKNOWLEDGMENTS}

10163

10059

\newpage

10164

10060

\section{APPENDIX: NetCDF}

10165

10061

\index{NetCDF format}

10166

>From the NetCDF User's Guide:

10062

10063

From the NetCDF User's Guide:

10167

10064

10168

10065

\begin{quotation}

10169

10066

The purpose of the Network Common Data Form (netCDF) interface is to

10243

10140

While it might seem a hassle to install the library, the added

10244

10141

functionality is significant: speedup in diagonalization with k-points

10245

10142

by storing the eigenvectors, optional restarts with charge density

10246

information instead of a density-matrix, new analysis tools, etc.

10143

information instead of a density-matrix, new analysis tools, etc.

10247

10144

10248

10145

10249

10146

\newpage