===============================================================================
*
* LinearThermalExpansionCoeffCubic__TD_522633393614_000
*
* Author: Mingjian Wen, wenxx151@umn.edu, University of Minnesota
* Date: 09/30/2015
* Version: 000
*
===============================================================================
This Test Driver computes the linear thermal expansion coefficient of a cubic
crystal (including fcc, bcc, sc, and diamond) at a specific temperature under a
given hydrostatic pressure (usually zero) of a single given species.
The linear thermal expansion coefficient is defined as:
alpha = (1/L[T]) * lim{epsilon->0} (L[T+epsilon*T]-L[T]) / (epsilon*T),
where L[T] is the length (e.g., lattice constant) at temperature T and
lim{epsilon->0} means that epsilon approaches zero.
Here, LAMMPS with NPT ensemble is used to do the computation, see below for
details about the algorithm.
===============================================================================
*
* Section 1: Summary of input variables
*
===============================================================================
In this Test Driver (TD), the user specifies the following variables through stdin
using a Jinja template file named 'pipeline.stdin.tpl' contained in the directory
of a Test which uses this TD (the sequence in which they are specified must be the
same as below, with 'modelname' being the first):
- 'modelname'
Type: string
Description: Contains the extended-id of a valid KIM Model. This can be specified
by '@< MODELNAME >@' in pipeline.stdin.tpl. When the pipeline mates a Model with
the Test, it will automatically replace '@< MODELNAME >@' with the Model's extended-ID.
- 'species'
Type: string
Description: Contains the abbreviation for the element from which the lattice will
be constructed (e.g. Si, Au, Al, etc.).
- 'mass'
Type: double
Description: The molar mass (in g/mol) of the element specified by 'species',
e.g. 29.98 for Aluminum.
- 'latticetypeinput'
Type: string
Description: The type of cubic lattice to construct. This can be any of "bcc",
"fcc", "sc", or "diamond". Note that this is completely case-insensitive, e.g.
"FcC" and "fCC" are both valid specifiers.
- 'latticeconst'
Type: double
Description: The lattice constant (in meter) about which the initial configuration
will build. Since LAMMPS NPT ensemble is used here, this value will not affect the
final linear thermal expansion coefficient, but a value corresponds to the lattice
constant at the specified temperature would decrease the simulation time.
- 'temperature'
Type: double
Description: The temperature (in Kelvin) at which the linear thermal expansion
coefficient is computed.
- 'pressure'
Type: double
Description: The hydrostatic pressure (in Pa) under which the linear thermal expansion
coefficient is computed. This value is usually taken to be zero.
===============================================================================
*
* Section 2: Invocation of LAMMPS
*
===============================================================================
The `lammps.in.template' is a LAMMPS input file that is used to calculate the
equilibrated volume of the system at a given temperature and pressure. The `rpls_*'
in this file are placeholders that will be replaced.
The LAMMPS script builds a supercell with 10*10*10 conventional (orthogonal)
unit cells with periodic boundary conditions. The system is then evolved under NPT
ensemble. The total timesteps is set to 2000*1000, and at each 1000th step, the
averaged temperature and volume (ignoring the first 5000 unstable steps) will be
checked against the specified values. Once they are pretty close, (Here, we require
that the averaged temperature to be within plus or minus 0.2K of the specified
temperature, and the averaged pressure to be within plus or minus 0.2Bar of the
specified pressure.) halt LAMMPS. Thus, the averaged volume at this temperature
is obtained.
===============================================================================
*
* Section 3: Algorithm to compute alpha
*
===============================================================================
The python script `runner' is used to compute the linear thermal expansion (alpha)
at a given temperature and pressure.
This script first reads the input parameters. Then it generates a temperature list
with 5 temperatures with interval dT, that is [T-2*dT, T-dT, T, T+dT, T+2dT], where
T is the temperature at which we want to compute the linear thermal expansion
coefficient. Here, we let dT = 20K. So if we want to compute alpha at T = 300K, the
temperature list would be [260, 280, 300, 320, 340] K. If T < 2*dT, we let T = 2*dT,
otherwise the first value (possibly the second) in the temperature list coule be
nonphysical negative value. In other words, the computed alpha at all temperatures
lower than 2*dT (40K here) would be the same, because the temperature list are
the same, i.e., [0, dT, 2*dT, 3*dT, 4*dT].
Five subprocesses will then be involved to compute the volume of the system at each
temperature via LAMMPS seperatly and simultaneously. Let's denote the volume list as
[V1, V2, V3, V4, V5]. With the temperatures and the corresponding volumes at hand,
least square regression is used to fit a linear line for the cubic root of the
volume against the temperature. Then the linear thermal expansion coefficient can
be computed by,
alpha = slope/T
where, slope is the slope of the fitted line.
===============================================================================
*
* Files contained
*
===============================================================================
The following files are in the directory (before any compilation is performed):
LICENSE.CDDL
The Common Development and Distribution License (CDDL) Version 1.0 file.
Makefile
Makefile to build executables with GNU or Intel compilers.
README
This file.
runner
A python script that implements the Test Driver.
lammps.in.template
LAMMPS input template file.
result.edn.tpl
Template file to report result in EDN format.
kimspec.edn