[Wien] Fwd: Again, scf no ghostband,x lapw2 ghost.:(

=?euc-kr?B?vue5zsij?= yangmino at samsung.com
Thu Feb 15 01:15:41 CET 2007


Dear wien user. 

In addition, I found one more thing.  The error just took place in parallel jobs.  Today, I tested them in single mode job, and no error was found.  
In the wien digest, something simillar situation was found and it suggest use scratch diretory in my home folder or continue x lapw2 -p -qtl right after run_lapw -p ~ in a qeue job.  But they are not my case. 
Something about library is wrong?  Here I add my compile information and the files in the library of MKL8.0.2 and F90


SHELL = /bin/sh
FC = /opt/intel/intel_fce_90/bin/ifort
MPF = /sw/EM64T/mpich-gm/bin/mpif90
CC = cc
FOPT =  -FR -mp1 -w -prec_div -pc80 -pad -ip -O3
FPOPT =  -FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML
DParallel = '-DParallel'
FGEN = $(PARALLEL)
LDFLAGS = -L/opt/intel/intel_f90/lib -i-static -lguide_stats -lsvml -lpthread
R_LIBS = -L/opt/intel/mkl8.0.2/lib/em64t -lmkl_lapack64 -lmkl_em64t -lguide -lguide_stats -lpthread
C_LIBS = $(R_LIBS)
RP_LIBS = $(R_LIBS) -L /sw/intel/cmkl/9.0/lib/em64t -lmkl_scalapack -lmkl_blacs -lmkl_em64t
CP_LIBS = $(RP_LIBS)

mkl 8.0.2
node1
libguide.a     libmkl_em64t.a      libmkl_lapack64.so  libmkl.so          libmkl_vml_p4n.so
libguide.so    libmkl_ias.so       libmkl_lapack.a     libmkl_solver.a    libvml.so
libmkl_def.so  libmkl_lapack32.so  libmkl_p4n.so       libmkl_vml_def.so

intel_f90
crtxi.o            libcprts.so       libcxa.so.5        libifcoremt_pic.a  libifport.so    libompstub.a
crtxn.o            libcprts.so.5     libguide.a         libifcoremt.so     libifport.so.5  libsvml.a
for_main.o         libcxa.a          libguide.so        libifcoremt.so.5   libimf.a        libsvml.so
icrt.internal.map  libcxaguard.a     libguide_stats.a   libifcore_pic.a    libimf.so       libunwind.a
icrt.link          libcxaguard.so    libguide_stats.so  libifcore.so       libirc.a        libunwind.so
init.o             libcxaguard.so.5  libifcore.a        libifcore.so.5     libirc_s.a      libunwind.so.5
libcprts.a         libcxa.so         libifcoremt.a      libifport.a        libirc.so

------- Original Message -------
Sender : 양민호<yangmino at samsung.com>  전문연구원/AE Center/삼성종합기술원
Date   : 2007-02-14 14:45
Title  : Again, scf no ghostband,x lapw2 ghost.:(

Dear wien user. 

I found several mails concerning about ghost band in mailing list.  But still I could not solve the problem. 

My case is cubic HfO2 and its elnes simulation with telnes.  I just want to simulate the Oxygen K edge considering cole-hole effects of the crystal as it is insulator with a big band gap.  I made supercell using the unitcell, did init_lapw it without any error and did run_lapw -p -ec 0.0001 -in1new 2 also without any error.  After 7 iteration in K-point parallel calculation, the energy converged. In this case I used RktMax=5 and 1 Kpoint. 

But the problem occured me when I ran x lapw -p -c -qtl after I did 'x kgen' to get the 6 6 6 k-mesh.
The error massage of "forrtl: severe (64) : input conversion error, unit 1002, file /scratch/ch_chbic.help032.

I checked /scratch/case.help032 and case.output2, and found the indication of ghost band near BAND 910. [Q(UE)=0.85 greater than Q(U)=0.17] 
I checked case.scf2 and can see the massage "QTL-B VALUE .EQ. 7202.95009 in Band of energy -0.03731 ATOM= 3 L=0. 
And of course there is not signifant bing QTL-B VALUE in case.scf.  

I did not increase the energy window in case.in1c. And as I want to see the DOS over fermi energy, even I have to increase the energy window.  

My question is which one I have to believe for editting case.in1c to remove ghostband between case.output2 and case.scf2. 

I attached the neccessary file below.  I put -1.0 charge in case.inm and reduce the occupancy from 2 to 1 in the file case.inc at the final atom (it is Oxygen).

Best regards. 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ch_cubic.in1c
Type: application/octet-stream
Size: 3052 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20070215/9ed665b5/ch_cubic.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ch_cubic.inc
Type: application/octet-stream
Size: 3374 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20070215/9ed665b5/ch_cubic-0001.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ch_cubic.inm
Type: application/octet-stream
Size: 171 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20070215/9ed665b5/ch_cubic-0002.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ch_cubic.in2c
Type: application/octet-stream
Size: 1480 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20070215/9ed665b5/ch_cubic-0003.obj


More information about the Wien mailing list