lammps/doc/fix_phonon.html

222 lines
9.4 KiB
HTML
Raw Normal View History

<HTML>
<CENTER><A HREF = "http://lammps.sandia.gov">LAMMPS WWW Site</A> - <A HREF = "Manual.html">LAMMPS Documentation</A> - <A HREF = "Section_commands.html#comm">LAMMPS Commands</A>
</CENTER>
<HR>
<H3>fix phonon command
</H3>
<P><B>Syntax:</B>
</P>
<PRE>fix ID group-ID phonon N Noutput Nwait map_file prefix keyword values ...
</PRE>
<UL><LI>ID, group-ID are documented in <A HREF = "fix.html">fix</A> command
<LI>phonon = style name of this fix command
<LI>N = measure the Green's function every this many timesteps
<LI>Noutput = output the dynamical matrix every this many measurements
<LI>Nwait = wait this many timesteps before measuring
<LI>map_file = <I>file</I> or <I>GAMMA</I>
<PRE> <I>file</I> is the file that contains the mapping info between atom ID and the lattice indices.
</PRE>
<PRE> <I>GAMMA</I> flags to treate the whole simulation box as a unit cell, so that the mapping
info can be generated internally. In this case, dynamical matrix at only the gamma-point
will/can be evaluated.
</PRE>
<LI>prefix = prefix for output files
<LI>one or none keyword/value pairs may be appended
<LI>keyword = <I>sysdim</I> or <I>nasr</I>
<PRE> <I>sysdim</I> value = d
d = dimension of the system, usually the same as the MD model dimension
<I>nasr</I> value = n
n = number of iterations to enforce the acoustic sum rule
</PRE>
</UL>
<P><B>Examples:</B>
</P>
<PRE>fix 1 all phonon 20 5000 200000 map.in LJ1D sysdim 1
fix 1 all phonon 20 5000 200000 map.in EAM3D
fix 1 all phonon 10 5000 500000 GAMMA EAM0D nasr 100
</PRE>
<P><B>Description:</B>
</P>
<P>Calculate the dynamical matrix from molecular dynamics simulations
based on fluctuation-dissipation theory for a group of atoms.
</P>
<P>Consider a crystal with <I>N</I> unit cells in three dimensions labelled <I>l
= (l<sub>1</sub>,l<sub>2</sub>,l<sub>3</sub>)</I> where <I>l<sub>i</sub></I>
are integers. Each unit cell is defined by three linearly independent
vectors <B>a</B><sub>1</sub>, <B>a</B><sub>2</sub>, <B>a</B><sub>3</sub> forming a
parallelipiped, containing <I>K</I> basis atoms labelled <I>k</I>.
</P>
<P>Based on fluctuation-dissipation theory, the force constant
coefficients of the system in reciprocal space are given by
(<A HREF = "#Campana">Campa&ntilde;&aacute;</A> , <A HREF = "#Kong">Kong</A>)
<center><b>&Phi;</b><sub>k&alpha;,k'&beta;</sub>(<b>q</b>) =
k<sub>B</sub>T
<b>G</b><sup>-1</sup><sub>k&alpha;,k'&beta;</sub>(<b>q</b>),</center>
</P>
<P>where <B>G</B> is the Green's functions coefficients given by
</P>
<center><b>G</b><sub>k&alpha;,k'&beta;</sub>(<b>q</b>) =
<<b>u</b><sub>k&alpha;</sub>(<b>q</b>)&#149;<b>u</b><sub>k'&beta;</sub><sup>*</sup>(<b>q</b>)>,</center>
<P>where <...> denotes the ensemble average, and
<center><B>u</B><sub>k&alpha;</sub>(<b>q</b>) = &sum;<sub>l</sub>
<b>u</b><sub>lk&alpha;</sub> exp(i<B>qr</B><sub>l</sub>)</center>
</P>
<P>is the &alpha; component of the atomic displacement for the <I>k</I>th atom
in the unit cell in reciprocal space at <B>q</B>. In practice, the Green's
functions coefficients can also be measured according to the following
formula,
</P>
<center><b>G</b><sub>k&alpha;,k'&beta;</sub>(<b>q</b>) =
<<b>R</b><sub>k&alpha;</sub>(<b>q</b>)&#149;<b>R</b><sup>*</sup><sub>k'&beta;</sub>(<b>q</b>)>
- <<b>R</b>><sub>k&alpha;</sub>(<b>q</b>)&#149;<<b>R</b>><sup>*</sup><sub>k'&beta;</sub>(<b>q</b>),
</center>
<P>where <B>R</B> is the instantaneous positions of atoms, and <<B>R</B>> is the
averaged atomic positions. It gives essentially the same results as
the displacement method and is easier to implement in an MD code.
</P>
<P>Once the force constant matrix is known, the dynamical matrix <B>D</B> can
then be obtained by
</P>
<center><b>D</b><sub>k&alpha;, k'&beta;</sub>(<b>q</b>) = (m<sub>k</sub>m<sub>k'</sub>)<sup>-1/2</sup> <b>&Phi;</b><sub>k&alpha;,k'&beta;</sub>(<b>q</b>)</center>
<P>whose eigenvalues are exactly the phonon frequencies at <B>q</B>.
</P>
<P>This fix uses positions of atoms in the specified group and calculates
two-point correlations. To achieve this. the positions of the atoms
are examined every <I>Nevery</I> steps and are Fourier-transformed into
reciprocal space, where the averaging process and correlation
computation is then done. After every <I>Noutput</I> measurements, the
matrix <B>G</B>(<B>q</B>) is calculated and inverted to obtain the elastic
stiffness coefficients. The dynamical matrices are then constructed
and written to <I>prefix</I>.bin.timestep files in binary format and to the
file <I>prefix</I>.log for each wavevector <B>q</B>.
</P>
<P>A detailed description of this method can be found in
(<A HREF = "#Kong2011">Kong2011</A>).
</P>
<P>The <I>sysdim</I> keyword is optional. If specified with a value smaller
than the dimensionality of the LAMMPS simulation, its value is used
for the dynamical matrix calculation. For example, using LAMMPS ot
model a 2D or 3D system, the phonon dispersion of a 1D atomic chain
can be computed using <I>sysdim</I> = 1.
</P>
<P>The <I>nasr</I> keyword is optional. An iterative procedure is employed to
enforce the acoustic sum rule on &Phi; at &Gamma;, and the number
provided by keyword <I>nasr</I> gives the total number of iterations. For a
system whose unit cell has only one atom, <I>nasr</I> = 1 is sufficient;
for other systems, <I>nasr</I> = 10 is typically sufficient.
</P>
<P>The <I>map_file</I> contains the mapping information between the lattice
indices and the atom IDs, which tells the code which atom sits at
which lattice point; the lattice indices start from 0. An auxiliary
code, <A HREF = "http://code.google.com/p/latgen">latgen</A>, can be employed to
generate the compatible map file for various crystals.
</P>
<P>In case one simulates an aperiodic system, where the whole simulation box
is treated as a unit cell, one can set <I>map_file</I> as <I>GAMMA</I>, so that the mapping
info will be generated internally and a file is not needed. In this case, the
dynamical matrix at only the gamma-point will/can be evaluated. Please keep in
mind that fix-phonon is designed for cyrstals, it will be inefficient and
even degrade the performance of lammps in case the unit cell is too large.
</P>
<P>The calculated dynamical matrix elements are written out in
<A HREF = "units.html">energy/distance^2/mass</A> units. The coordinates for <I>q</I>
points in the log file is in the units of the basis vectors of the
corresponding reciprocal lattice.
</P>
<P><B>Restart, fix_modify, output, run start/stop, minimize info:</B>
</P>
<P>No information about this fix is written to <A HREF = "restart.html">binary restart
files</A>.
</P>
<P>The <A HREF = "fix_modify.html">fix_modify</A> <I>temp</I> option is supported by this
fix. You can use it to change the temperature compute from thermo_temp
to the one that reflects the true temperature of atoms in the group.
</P>
<P>No global scalar or vector or per-atom quantities are stored by this
fix for access by various <A HREF = "Section_howto.html#4_15">output commands</A>.
</P>
<P>Instead, this fix outputs its initialization information (including
mapping information) and the calculated dynamical matrices to the file
<I>prefix</I>.log, with the specified <I>prefix</I>. The dynamical matrices are
also written to files <I>prefix</I>.bin.timestep in binary format. These
can be read by the post-processing tool in tools/phonon to compute the
phonon density of states and/or phonon dispersion curves.
</P>
<P>No parameter of this fix can be used with the <I>start/stop</I> keywords
of the <A HREF = "run.html">run</A> command.
</P>
<P>This fix is not invoked during <A HREF = "minimize.html">energy minimization</A>.
</P>
<P><B>Restrictions:</B>
</P>
<P>This fix assumes a crystalline system with periodical lattice. The
temperature of the system should not exceed the melting temperature to
keep the system in its solid state.
</P>
<P>This fix is part of the USER-PHONON package. It is only enabled if
LAMMPS was built with that package. See the <A HREF = "Section_start.html#start_3">Making
LAMMPS</A> section for more info.
</P>
<P>This fix requires LAMMPS be built with an FFT library. See the
<A HREF = "Section_start.html#start_2">Making LAMMPS</A> section for more info.
</P>
<P><B>Related commands:</B>
</P>
<P><A HREF = "compute_msd.html">compute msd</A>
</P>
<P><B>Default:</B>
</P>
<P>The option defaults are sysdim = the same dimemsion as specified by
the <A HREF = "dimension">dimension</A> command, and nasr = 20.
</P>
<HR>
<A NAME = "Campana"></A>
<P><B>(Campa&ntilde;&aacute;)</B> C. Campa&ntilde;&aacute; and
M. H. M&uuml;ser, <I>Practical Green's function approach to the
simulation of elastic semi-infinite solids</I>, <A HREF = "http://dx.doi.org/10.1103/PhysRevB.74.075420">Phys. Rev. B [74],
075420 (2006)</A>
</P>
<A NAME = "Kong"></A>
<P><B>(Kong)</B> L.T. Kong, G. Bartels, C. Campa&ntilde;&aacute;,
C. Denniston, and Martin H. M&uuml;ser, <I>Implementation of Green's
function molecular dynamics: An extension to LAMMPS</I>, <A HREF = "http://dx.doi.org/10.1016/j.cpc.2008.12.035">Computer
Physics Communications [180](6):1004-1010
(2009).</A>
</P>
<P>L.T. Kong, C. Denniston, and Martin H. M&uuml;ser,
<I>An improved version of the Green's function molecular dynamics
method</I>, <A HREF = "http://dx.doi.org/10.1016/j.cpc.2010.10.006">Computer Physics Communications [182](2):540-541
(2011).</A>
</P>
<A NAME = "Kong2011"></A>
<P><B>(Kong2011)</B> L.T. Kong, <I>Phonon dispersion measured directly from
molecular dynamics simulations</I>, <A HREF = "http://dx.doi.org/10.1016/j.cpc.2011.04.019">Computer Physics Communications
[182](10):2201-2207,
(2011).</A>
</P>
</HTML>