MLWF calc: glibc detected invalid pointer

raul_l · Post by **raul_l** » Thu Jun 13, 2013 5:06 pm

Perhaps somebody can check if the following is a bug. I'm doing a standard calculation of producing MLWFs, but my calculation terminates if I use

Code: Select all

typat 1 2 3 3 3 3 1 2 3 3 3 3
xred
0.500 0.3027  0.75 # Cd
0.000 0.17847 0.25 # W
0.242 0.372   0.384 # O
0.202 0.096   0.951 # O
0.758 0.372   0.116 # O
0.798 0.096   0.549 # O
0.500 0.6973  0.25 # Cd
0.000 0.82153 0.75 # W
0.758 0.628   0.616 # O
0.798 0.904   0.049 # O
0.242 0.628   0.884 # O
0.202 0.904   0.451 # O

instead of

Code: Select all

typat 1 1 2 2 3 3 3 3 3 3 3 3
xred
0.500 0.3027  0.75 # Cd
0.500 0.6973  0.25 # Cd
0.000 0.17847 0.25 # W
0.000 0.82153 0.75 # W
0.242 0.372   0.384 # O
0.202 0.096   0.951 # O
0.758 0.372   0.116 # O
0.798 0.096   0.549 # O
0.758 0.628   0.616 # O
0.798 0.904   0.049 # O
0.242 0.628   0.884 # O
0.202 0.904   0.451 # O

(The only difference is in the order.)
I'm not sure if this is related to abinit or wannier90. I have already described the problem here. I have attached the input which reproduces the problem, using the following datasets taken from the abinit website:
048-Cd-gpbe-campos-mod
074-W-gpbe-nsp-campos-mod
008-O-gpbe--bm

The first calculation, run in parallel, produces the required WFK and DEN files and finishes normally in either case. The second part, for generating MLWFs in serial, only finishes normally if for 'typat' and 'xred' parameters I first specify the coordinates of one atom type before moving to the next atom type.