ABINIT Discussion Forums

Dear all,

I'm trying to calculate the Fermi surface of rutile TiO2. However, the calculations stops with the message

chkinp : ERROR -
The number of bands has to be constant for the output of the Fermi surface.
Action : set all the nbands to the same value in your input file

I have explicitly set the nband variable in each dataset to 26 (input is in the attachment). Because of that, I do not understand why this happens. Can anyone help me to overcome this problem?

I'm using abinit 6.2 version on a Linux cluster.

Thank you all in advance!

Igor Lukacevic

I have also checked the same calculation using the input file from test6 t14.in (MgB2 case). I got the same error in the output file. Log file give a lot of

forrtl: error (69): process interrupted (SIGINT)
Image PC Routine Line Source
libpthread.so.0 00002B62B144889B Unknown Unknown Unknown
libibverbs.so.1 00002B62B2E5EA02 Unknown Unknown Unknown
libmthca-rdmav2.s 00002B62B4B89DF6 Unknown Unknown Unknown
libibverbs.so.1 00002B62B2E64AA1 Unknown Unknown Unknown
libmpich.so.1.0 00002B62B2BE3806 Unknown Unknown Unknown
libmpich.so.1.0 00002B62B2BE7D37 Unknown Unknown Unknown
libmpich.so.1.0 00002B62B2BC727C Unknown Unknown Unknown
libmpich.so.1.0 00002B62B2B952A3 Unknown Unknown Unknown
libmpich.so.1.0 00002B62B2B9500A Unknown Unknown Unknown
abinit 000000000152AE95 Unknown Unknown Unknown
abinit 00000000015261C3 Unknown Unknown Unknown
abinit 0000000000F29D53 Unknown Unknown Unknown
abinit 000000000040DDC7 Unknown Unknown Unknown
abinit 000000000040CCCC Unknown Unknown Unknown
libc.so.6 00002B62B42F2994 Unknown Unknown Unknown
abinit 000000000040CBD9 Unknown Unknown Unknown
DONE



Could this be that the issue is in the abinit installation on the cluster?

These are the configure options used

./configure --prefix=/usr/local/abinit6 --enable-mpi --with-mpi-level=1 --with-mpi-prefix=/opt/mpich-1.2.7p1/intel/ --enable-64bit-flags --enable-smp --disable-debug --with-linalg-libs="-L/opt/intel/mkl/10.2.1.017/lib/em64t -lmkl_intel_lp64 -liomp5 -lpthread -lguide -lmkl_intel_thread -lmkl_core -lm" --with-dft-flavor="none" --with-trio-flavor="none"

Igor L.

I'm sending also the output and log files. Maybe someone will see where the problem lies.

This calc. was done with version 6.10.3.

Thank you in advance!

Yours,
Igor Lukacevic

Dear Igor,

The check on the number of bands is not correct when there are multiple datasets
with different number of k-points. Change chkinp.F90 as shown below.

Best regards,
Matteo


=== modified file 'src/57_iovars/chkinp.F90'
--- src/57_iovars/chkinp.F90 2011-12-06 15:24:05 +0000
+++ src/57_iovars/chkinp.F90 2011-12-18 14:57:49 +0000
@@ -1880,7 +1880,7 @@
end if

! Make sure all nband are equal (well it is always enforced for metals)
- if (any(dt%nband(:) /= maxval(dt%nband(:)) )) then
+ if (any(dt%nband(1:nkpt*nsppol) /= maxval(dt%nband(1:nkpt*nsppol)) )) then
write(message,'(6a)')ch10,&
& ' chkinp : ERROR -',ch10,&
& ' The number of bands has to be constant for the output of the Fermi surface.',ch10,&

Dear Matteo,

thank you for the solution. I have a further question for you... a friend of mine has abinit 6.2 version installed on his cluster, and he tried the same calc., but didn't receive this error. How can this be? I also tried with this abinit version (see first post).

Best,

Igor