MD++ QnA - Revision history

Caiwei at 06:14, 25 March 2008

2008-03-25T06:14:11Z

← Older revision		Revision as of 06:14, 25 March 2008
Line 1:		Line 1:
	<H1 ALIGN="CENTER"><FONT SIZE="-1"></FONT>		<H1 ALIGN="CENTER"><FONT SIZE="-1"></FONT>
	<BR>		<BR>
			MD++ Frequently Asked Questions</H1>
	QnA about MD++</H1>
	<!--<DIV>		<!--<DIV>
	<P ALIGN="CENTER"><STRONG>Keonwook Kang and Wei Cai</STRONG></P>		<P ALIGN="CENTER"><STRONG>Keonwook Kang and Wei Cai</STRONG></P>

Kwkang at 20:13, 14 February 2008

2008-02-14T20:13:29Z

← Older revision		Revision as of 20:13, 14 February 2008
Line 26:		Line 26:
	export LD_LIBRARY_PATH		export LD_LIBRARY_PATH
	</pre>		</pre>

			== Q. Why stacking fault (SF) energy is inconsistent with ''ab-initio'' calculations or experimental values? ==
			[A] This is not a bug of MD++. The discrepancy of SF energy calculation is originated from how the corresponding potential is designed and fit. I quote Chris's answer below.
			:"... the EAM potential is an semi-empirical potential that is fit to a set parameters. Many of the potentials are fit to things such as the elastic constants, cohesive energy, and like terms. Many of the potentials are NOT fit to stacking fault energies so the value comes out of the details of the specific formulation. Therefore, we do not, in general, expect there to be good agreements between experimental (or ab initio) results and those that come out of EAM. Since the stacking fault energy is closely linked to the cutoff used in the potential, the values produced by EAM potentials vary significantly. <br> ... For some potentials in FCC metals, negative stacking fault energies have been reported (which implies the HCP phase is more stable). I work a lot with gold, and the potential file that comes with MD++ (and thus MDCASK) gives a stacking fault energy 5 to 10 times lower than experimentally reported values. Therefore, we need to be careful anytime we use a potential to make sure that it captures the basics of the physics we want to simulate."

Kwkang: New page:

QnA about MD++

First Written [[...

2008-02-14T19:52:19Z

New page: <H1 ALIGN="CENTER"><FONT SIZE="-1"></FONT> <BR> QnA about MD++</H1>  <P ALIGN="CENTER">First Written [[...

New page

<H1 ALIGN="CENTER"><FONT SIZE="-1"></FONT>
<BR>
QnA about MD++</H1>

<P ALIGN="CENTER">First Written [[ Feb 14 ]], [[2008]]</P>
<P ALIGN="CENTER">Last Modified [[ Feb 14 ]], [[2008]]</P>
<P>
<BR><HR>
== Q. Error with libimf.so ==
[Q] I see an error message "error while loading shared libraries: libimf.so:
cannot open shared object file: No such file or directory" while running MD++.
<br>
[A] This is not a bug of MD++. When you compile MD++ with intel compilers (icc or ifort), you need to specify the library path for intel compilers. For example, if the libraries are installed at '''/opt/intel/cc/10.1.012/lib/''' and '''/opt/intel/fc/10.1.012/lib/''', you run the following command at the bash shell prompt.

$ LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/cc/10.1.012/lib:/opt/intel/fc/10.1.012/lib; export LD_LIBRARY_PATH

You can check the library path by running the commnd '''echo'''.

$ echo $LD_LIBRARY_PATH

Another solution is to add the following lines to the bash startup file '''~/.bash_profile'''.
<pre>
LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/cc/10.1.012/lib:/opt/intel/fc/10.1.012/lib
export LD_LIBRARY_PATH
</pre>