Do you use the serial or the parallel version of the MKL ?
The MKL that is used for matrix operations is internally parllelized, the
behaviour is
usually controlled by the environment variables OMP_NUM_THREADS or
MKL_NUM_THREADS
as well as some other MKL rlated ones (check the MKL manual).
The us
In case you wonder that there is no t2g state:
Pnma has only twofold rotational axes (+inversion + 3 mirrorplanes),
and the point group of Gamma (D2h) has no triply degenerate irreducible
representation.
the three dxy, dxz, dyz basis functions belong to the representations b1g, b2g,
b3g, respecti
ions (you don't need "all" possible variations).
Even with medium sized supercells one can already learn a lot about
tendencies for clustering,
Gerhard Fecher schrieb:
> First a question, how is the bandstructure defined in a random alloy when the
> infinite periodicity i
First a question, how is the bandstructure defined in a random alloy when the
infinite periodicity is lost ?
To specify 60 % of the sites to be occupied by A, you need a supercell say with
100 atoms and
to set 60 of the sites by A and 40 by B atoms (always one atom no fraction !)
this you have
Check if the libraries are realy in the path that you give in the options and
that they fit youre bash.rc settings
try something like (use the path were you have the libraries):
current:FOPT:-FR -mp1 -w -prec_div -pc80 -pad -DINTEL_VML -xHost -O3 -m64 -ip
-traceback
current:LDFLAGS:-L/opt/inte
The professional version of 11.1 that includes the MKL works well and is fast,
I do not have any problems in compiling or running Wien2k (9.2 or older)
however, I figured out that it is sometimes very helpful to read the release
notes and the manuals.
Ciao
Gerhard
===
Some rule about the linking were changed wit the latest Intel compilers and MKL,
Please go to the webpage
http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/
this page will tell you the correct linking options for the mkl.
Ciao
Gerhard
Dr.
I have some short questions:
I am looking for the eigenvalues and eigenvectors at the high symmetry points.
using WFPRI in case.in1 produces an ASCII case.vec file in addition to the
binary case.vector
that contains the eigenvalues and eigenvectors. How are the eigenvectors
organized ?
(Note:
If Wien2k wants to link to the 64 bit version of the MKL then you told
site_config to do so.
Check the lines with LDFLAGS and RLIBS, there you should give the path to the
32 bit libraries.
The path where the libraries are starts with -L/wherever/youre/libraries/are
followed by the libraries to b
For the 2 structures below, the result should be the same.
as in space group 216 the change of the atoms in positions 4a, 4b with those
on 4c,4d
results in the same symmetry (all positions have Td symmetry)
Occupying 4a,4b or 4c,4d (like you gave below) with the same species will
always result i
download the latest Version of the Intel professional compiler, it includes
both compiler and mkl
be sure that youre operating system is installed in the 64bit Verison otherwise,
the Intel installer will not install the em64t Version of the mkl (as it would
not make sense)
and you have to use the
pecific details of the OS (and these things are never standard) many
parameters such as stack-size limits, LD_LIBRARY_PATHS and other
environmental variables are not inherited from the calling shell.
On Fri, Aug 7, 2009 at 6:39 AM, Gerhard Fecher wrote:
> The answer of Intel is found in all
NO !
Ciao
Gerhard
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.a
The answer of Intel is found in all release notes of the Fortran Compilers 8.0
and higher:
... On Linux also note that many 32bit Linux distributions ship with a pthread
static library ibpthread.a) that at runtime will fix the
stacksize to 2093056 bytes regardless of the ulimit setting. To
avoid
why not using batch -f xyz .job
with a jobfile containing
cd some directory
run_lapw shatever
This will run in background one of the jobs after the other
batch is a special of at and comes with the usual Linux distributions.
Ciao
Gerhard
Dr. Gerhard H. Fec
scalapack comes usually with the mkl
check the directory where you installed the mkl
you may also play with the number of threads (environment variable
OMP_NUM_THREADS) you allow the mkl to use
but don't use it together with parallel.
Ciao
Gerhard
Dr. Gerhar
you need to link the mkl and not a fortran file from some directory
Ciao
Gerhard
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
Von: wien-bounces at zeus
are youre environment variables correctly set LD_LIBRARY_PATH etc.
is the MKL in the directory that you specified in siteconfig ?
Ciao
Gerhard
Dr. Gerhard H. Fecher
Institut of Inorganic and Analytical Chemistry
Johannes Gutenberg - University
55099 Mainz
The messages suggest "Check compile.msg in this directory", did you do that ?
without knowing the options you used for compilation, it is impossible to say
why
these massges appear !
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounc
search the mailing list for OCTAVE_PATH
and follow the instructions given by Peter posted at Thu Jul 10 09:43:49 CEST
2008
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von NORIZA Ahma
[dhguo at
cat.hokudai.ac.jp]
Gesendet: Montag, 22. September 2008 06:23
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] Why ONLY bandstructure can give the image? --Howto create
the link between wien2k and gnuplot?
Dear Gerhard Fecher,
Thank you for your hints. I am not sure whether I have started w2
next possibillity is that some library that is needed to convert the graphics
files is missing or not accesible
to gnuplot or w2web.
The reason that one plot works but not the other is:
the spaghetti plot uses jpg as intermediate graphics and converts it to ps
the dosplot (for example, others too
A simple question, did you install gnuplot after starting w2web,
in that case kill w2web
check who is allowed to execute gnuplot
and start w2web again.
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at
Realy not a good idea,
if you have C on the surface, what C is it ? CO, hydrocarbon, graphite ? All
have a different binding energy of the C 1s !
I suggest, if youre sample is not to be supposed to contain C, you should clean
it.
XPS energies have to be calibrated using the Au (or Ag) Fermi ene
Nice idea and possibly it would be interesting to see how Wien2k performs in
such an environment if it does at all.
But there is no need to wait for a box. For example:
Go to http://www.opensuse.org/ continue to http://software.opensuse.org/
Download the 64-bit (!) Version (DVD)
There are also
I experienced the same for some large cases in parallel calculations.
The error vanished after I deleted all files in the scratch directory but
changing nothing else.
Interesting to note, it happens only once for a given system and then never
again, so its hard to follow why.
Ciao
Gerhard
The first point is interesting as I faced that problem recently, too.
The problem was solved after I cleaned the complete directory (I deleted all
files by hand, clean_lapw did not help).
As it appeared only once, I thought that this is just a follow-up error caused
by some accidential mistake I
check the case.spaghetti_ene files, there you find band by band all kx, ky, kz,
k, and E values
of the k-path you defined for the calculation of the bands
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.a
the options as above with only differing in the -xProcessor switches.
Ciao
Gerhard
Von: Yang Li [YLi at mercury.brooklyn.cuny.edu]
Gesendet: Montag, 28. Juli 2008 18:41
An: Gerhard Fecher
Betreff: Re: [Wien] Wien2k_08 on SuSE Linux 11
Hi, Dr. Fecher
MKL was not buggy,
most of the installations were buggy or inappropriate, that's a big difference.
I have no problems with IFORT 10.1, MKL 10.0.3 on freshly installed 64-bit SUSE
10.3 or 11.0 systems,
not even with -static linking.
Problems occure on updated Linux systems and too many old instal
I didn't try 11 it up to know, but there should be no principal difference.
If your CPU allows, take care to install the 64-bit version, that prevents most
of the troubles with segmentation faults etc.
There may be also differences if you install Suse from scratch or as an update,
in the latter
Dear Peter,
It was without -orb
convergence criteria were -ec 0.1 -cc 0.001
I tried both large and small mixing, the new and the old Broyden mixing schemes
I took care about all possible bad QTL values, and changed number of k-values
and plane waves,
the result was reproducible.
The reason th
Hallo Peter,
I just realized that the analysis of the total magnetic moments is not shown as
graph,
the reason is that it is printed on the screen but MMTOT is not written to the
file analyse.out
(concerns the latest version, it was ok in older versions but I do not know
from which version this
Dear Peter,
it seems that optimize uses case.struct instead of case_initial.struct, at
least in the last Wien2k version if called from w2web, independent which file
one asks for.
and usually case.struct contains the last structure of the optimization.
the easiest way is to copy case_initial.struc
If you include spin-orbit interaction
then only one kind of bands exists (there are no up or down bands anymore)
however, you still can calculate the spin resolved DOS.
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theo
check the files
at.allow and at.deny, maybe you are not allowed to use it.
On some systems the user that likes to use at or batch needs in addition to be
in the group at.
The demon should be started during the boot sequence and not every time you
like to use it.
Ciao
Gerhard
Handbook-Not-Read-Error
see page 26 of the users guide
3.11.4 Bandstructure
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von azar hadidi [k_hadidi3x at yahoo.com]
Gesendet: Sonntag,
something is wrong with the series in which the libraries are called
-lmkl_em64t -lguide -lpthread -lmkl_lapack -lguide -lpthread
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von zhao
check a textbook on angular momenta
e.g. the one of Zare
or textbooks on atomic physics e.g. Cowan
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von fatemeh.mirjani [f_mirjani at
ph.iut.ac
Did you try
EXPORT OMP_NUM_THREADS = 2 (or 4)
with the test cases.
(or at least before starting w2web)
In my case it sems to work with a freshly installed Suse 10.3 64-bit, but not
with older Versions.
my MKL was 10.0.2
did you try 10.0.3 ? I had not yet time to test it.
I still have some proble
Did you check that the mkl libraries are realy in opt/intel/mkl/9.1.021/lib/32
Most probably you need to use ldconfig to tell youre system where the shared
libraries are
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theoc
if youre Linux is 64 bit, you should use the emt64 Versions of the compiler
(usually installed in fce) and MKL (in lib64emt)
If it is not then I wonder why.
Ciao
Gerhard
No, -ulimit does not work, just before it comes again.
ceterum censeo ...
Von: wien
Interesting answer, but obviously not from the Intel compiler manual.
I think its a better idea to use ldconfig to set up the shared libraries
correctl, isn't it.
Ciao
Gaerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theoc
L Linker Flags:$(FOPT)
-L/opt/intel/mkl/9.1.023/lib/em64t -lpthread
P Preprocessor flags '-DParallel'
R R_LIB (LAPACK+BLAS): -lmkl_lapack -lmkl_em64t -lguide -lvml
-pthread
Pablo de la Mora
Gerhard Fecher escribi?:
> Youre R_LIB Opt
Be aware, that this might not be the only additional library you may need to
run Wien2k with
MKL 10 in a stable way.
Due to the changed threading in MKL 10, I suggest to read very carefully the
Release Notes and MKL Users Guide,
in particular the parts on Linking, Performance and Threading.
Actu
2008 21:38
An: A Mailing list for WIEN2k users
Betreff: [Wien] Where is the MKL?
I am trying to compile the WIEN2k with the Intel package:
Fortran: fce 10.0.023
Math Kernel Library mkl 9.1.023
(this was due to the advice of Gerhard Fecher not to use the latest
versions, but ifort 10.0 and mkl 9.1
hiroshima-u.ac.jp]
Gesendet: Dienstag, 1. April 2008 18:39
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] SIGSEGV, segmentation fault occurred!
Very thanks Gerhard Fecher.
I have downgrade mkl to 9.1 version and now it looks work well.
I try to compile as dynamic. I use the some parameter as
Both, youre ifort and mkl versions have still bugs or at least some problems
with Wien, see my earlier comments in that forum.
I suggest to downgrade to the more stable versions: ifort 10.0 mkl 9.1,
both are avalable for free from the Intel webpage.
Why do you use the 32-bit libraries if you have
: Ricardo Faccio [rfaccio at fq.edu.uy]
Gesendet: Mittwoch, 12. M?rz 2008 15:34
An: Gerhard Fecher
Betreff: lapw2 compilation
Dear Gerhard
In principle I installed wien2k in a Intel(R) Core(TM)2 CPU 6420 @
2.13GHz with 2Gb Ram (DELL optiplex 330). I have OpenSuse 10.3, 64 bit. I
installed ifort
For MKL 10.0, do not set OMP_NUM_THREADS or MKL_NUM_THREADS
For MKL 10.0 you may need -lmkl_core
FOR IFORT 10.1 the otimization will not work (see previous messages)
There is no need to use the -Bstatic or -Bdynamic switches on em64t systems.
use -traceback and/or -traceback -g to see where the e
It was in this forum.
Just to mention, in MKL10 the new environment variable MKL_NUM_THREADS does not
work either.
Actually, I have the most stable version running when compiled and linked with
IFORT 10.0 (not 10.1) and MKL 9.1.
On em64t systems one can link with this combination everything wit
Seems I have to make the question more complicated.
Do you have a mirror and a friend at hand ?
Ok then go put your friend between you and the mirror such that he or she faces
the mirror,
try to answer my question again.
You've got it ?
have fun
Gerhard
_
just read
Richard M. Martin
Electronic Structure (Basic Theory and Practical Methods)
Cambridge University Press 2004
Ciao Gerhard
PS.: If you stand in front of a mirror, then youre right and left hands are
interchanged.
Why not youre head and feet?
Von:
Did you set LD_LIBRARY_PATH correctly ?
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von Ricardo Faccio [rfaccio at fq.edu.uy]
Gesendet: Montag, 3. M?rz 2008 19:50
An: wien at zeus.theochem
Principally it should be possible to use the script for volume optimization (or
an adopted one).
Just make the klists and in1 files with different extensions and copy them
similar to the structure files.
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.
This means that the number you give for the k-points is not the one the program
uses for the integration:
Assume a cubic system then 1000 k-points result in a 10x10x 10 mesh,
due to the symmetry operations, there are many k-pints in the cube that are
equivalent,
and if one knows their multiplicit
Check the MKL manual to find that you have to use different options in version
10 compared to version 9
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von Oleg Rubel [rubel at
Physik.U
ve
>found something more on l2main to explain the problem or may be someone
>else ?
>Also, I am not so use to modify the makefile so if you did a specific
>modification for l2main in your make file, I will appreciate to have it.
>Regards
>Florent
>
>Gerhard Fecher a ?crit :
&g
Just from experience:
It is helpfull to use charge consitence -cc 0.001 (at least) for AFM cases in
SP calculations
otherwise one ends up very easily in a ferrimagnetic state.
Ciao
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
z
rest, it makes sense to calculate a
formation energy involving that gas phase. It might be more relevant
from a predictive point of view than the 0K prediction.
> May be its hard to breathe at 0K because oxygen will be solid, too.
One could try to swallow oxygen/nitrogen (20/80) ice cubes as a subs
Just to contribute also to the confusion:
In my lab, the nitrogen is liquid and comes in big barrels. If the gaseous
phase of N2 at room temperature is the ground state, does it mean that the
nitrogen in my barrel is deexcited into a sub-ground state
In the textbook on an introduction to solid s
As the Intel compiler manual tells,
for 32 bit systems the setting of ulimit may not influence the behavior at
runtime at all.
Note: The size of {\it "unlimited"} varies by {\scshape Linux}
configuration, so you may need to specify a larger, specific number to
{\it ulimit} (for example, 9
you also may take care about using the -xW switch,
I do not know whether the processor specific compiler switches work well with
AMD
it might be better to use only -O3 to force more optimizations (or the default
-O2, just check using the lapw1 benchmark).
maybe someone else has used and is exper
This remarks just tell which optimization the compiler is doing for you.
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von zubaer [zubaexy at gmail.com]
Gesendet: Montag, 7. Januar 2008 23:0
Similar things happen to me
4 threads seem to be very unstable (also if using the new MKL_NUM_THREADS
instead of OMP_NUM_THREADS)
Do you have all memory slots filled ? What main board ?
My Timings are on 2 dual-core CPUs Xeon 5160 for Wien2k07.3 ifort 10.1.008 (-FR
-mp1 -w -prec_div -pc80 -pad
zswap',dgemm belong to the BLAS/LAPACK libraries
that means: the path to the MKL is missing in the Linker options !
Gerhard
Von: wien-bounces at zeus.theochem.tuwien.ac.at [wien-bounces at
zeus.theochem.tuwien.ac.at] im Auftrag von Arun Kr. Chatterjee [ar
101 - 167 of 167 matches
Mail list logo