p2y segmentation error

[martinspenke]

Dear Daniele,

i get this segmentation error when running "p2y -S -N"

Code: Select all

 
 <---> P(W) 2 Y(ambo) Ver. 5.0
 <---> DBs path set to .
 <---> Index file set to data-file.xml
 <---> Header/K-points/Energies...done
 <---> Cell data...done
 <---> Atomic data...done
 <---> Symmetries...[SI no]...[I yes]...[-I no]...[TR yes]Segmentation fault

Attached is my QE input.

Best regards
Martin

[Daniele Varsano]

Dear Martin,
we will look at it in detail. My suspect is that the problem arise because you are forcing the xc potential with the input_dft variable, using an lda pseudopotential. You can try to look if the problem disappear using a blyp functional avoiding to use the input_dft variable. Unfortunately I've checked that in the qe espresso pseudo repository there is only a blyp functional of the hgh form, that usually turns to be rather hard.
Anyway we will look at it and hopefully solve it soon.
Best,
Daniele

[martinspenke]

Dear Daniele,

Unfortunately even if i avoid using input_dft variable and use Si.blyp-hgh.UPF, i still get the same error.

Best regards
Martin

[Daniele Varsano]

OK,
thanks for the test, we will look at it in detail and let you know.
Daniele

[Davide Sangalli]

Dear Martin,
we should have fixed this issue in the GLP.

If you are using the svn version simply update the soure (svn up).
Otherwise you should modify the file "interfaces/int_modules/mod_xc2y.F" as below (replace the lines with a minus with the lines with a plus).

Let us know if it works.

Best regards,
Davide

FIX:

- integer, parameter :: N_Abinit_XC = 28,N_PW_X = 10, N_PW_C = 13, N_PW_GX = 14, N_PW_GC = 11
+ integer, parameter :: N_Abinit_XC = 28,N_PW_X = 10, N_PW_C = 13, N_PW_GX = 15, N_PW_GC = 11

-& (/ 'NOGX', 'B88 ', 'GGX ', 'PW91', 'PBX ', 'PBE ', 'RPB ', 'HCTH', 'OPTX',&
+& (/ 'NOGX', 'B88 ', 'BLYP', 'GGX ', 'PW91', 'PBX ', 'PBE ', 'RPB ', 'HCTH', 'OPTX',&

-& (/ NOXC, XC_GGA_X_B88, XC_GGA_X_PW91, XC_GGA_X_PW91, XC_GGA_X_PBE, XC_GGA_X_PBE, XC_GGA_X_PBE_R,&
+& (/ NOXC, XC_GGA_X_B88, XC_GGA_X_B88, XC_GGA_X_PW91, XC_GGA_X_PW91, XC_GGA_X_PBE, XC_GGA_X_PBE, XC_GGA_X_PBE_R,&

[martinspenke]

Dear Davide,
many many thanks, it works now perfectly, as always you give pin point instructions.

I also have another annoying problem in QE :
Every time i want to use hybrid functional in Quantum Espresso like " input_dft='B3LYP' ", QE complains which is due to improper compiling.
Some thing like this :

Code: Select all

*** glibc detected *** pw.x: malloc(): memory corruption: 0x0000000002742d90 ***
======= Backtrace: =========
/lib/libc.so.6(+0x78bf6)[0x7f4c56a7cbf6]
/lib/libc.so.6(+0x7cd18)[0x7f4c56a80d18]
/lib/libc.so.6(__libc_malloc+0x6e)[0x7f4c56a81bce]
/lib/libc.so.6(__backtrace_symbols+0x11f)[0x7f4c56b0703f]
[luca:28900] *** Process received signal ***
[luca:28900] Signal: Aborted (6)
[luca:28900] Signal code:  (-6)
[luca:28901] *** Process received signal ***
[luca:28901] Signal: Aborted (6)
[luca:28901] Signal code:  (-6)
pw.x(__exx_MOD_exx_grid_init+0x1664)[0x44e594]
pw.x(setup_+0x1adc)[0x4e458c]
pw.x(MAIN__+0x2ca)[0x43774a]
pw.x(main+0x2a)[0x88cdfa]
/lib/libc.so.6(__libc_start_main+0xfd)[0x7f4c56a22c8d]
pw.x[0x4373b9]
======= Memory map: ========
00400000-008e1000 r-xp 00000000 08:08 1323842                            /home/luca/Downloads/espresso-5.0.2/PW/src/pw.x
00ae0000-00ae1000 r--p 004e0000 08:08 1323842                            /home/luca/Downloads/espresso-5.0.2/PW/src/pw.x
00ae1000-00ae9000 rw-p 004e1000 08:08 1323842                            /home/luca/Downloads/espresso-5.0.2/PW/src/pw.x

Can you please tell me which compiler and/or library versions (gcc, libgfortran, etc) are the best suitable for compiling QE ?

With best wishes,
Martin

[Davide Sangalli]

Dear Martin,
this time we have to thank Myrta for the fix. She is the expert for the libxc interface.

Anyway I'm not at all a QE expert. Usually I work with gfortran (I've tryed different versions) but I've never used hybrids.
I guess the best thing is to ask on the pw forum.

Davide