[Wien] The "segmentation fault" in LAPW2

[Yang, Jinbo]

-----Original Message-----
From: wien-admin at zeus.theochem.tuwien.ac.at
[mailto:wien-admin at zeus.theochem.tuwien.ac.at]On Behalf Of
danata at kinetic.ac.donetsk.ua
Sent: Tuesday, January 06, 2004 9:09 AM
To: wien at zeus.theochem.tuwien.ac.at
Subject: [Wien] The "segmentation fault" in LAPW2


Dear Peter Blaha and all Wien2k users,

I'm calculating the optical properties of the periodic 2-dimentional 
multilayered magnetic structures, such as [Ni(1ML)/Cu(1ML)]_{x} , x is the 
number of the periods.  

I use the supercell approach to simulate the 2-D structure with 10-12 
vacuum layeres in z- direction. I converge the scf with the charge 
convergence criteria 

 runsp_lapw -i 80 -cc 0.0001 -so

, increasing the k-point number in 
case.klist to get the k-point convergence.

 The problem is that at some stage, after the successfully converged scf 
with k-list, generated for 300 k-points in the hole BZ (ndiv=(14 14 
1)) 
the further increase of k-point number leads to the "segmentation fault" 
in LAPW2 -up. 
 
For example, this problem  appeares for the structure 
[Ni(1ML)/Cu(1ML)]_{3} for the k-list with 580 k-points in BZ, ndiv=(16 16 
1), while the case of [Ni(1ML)/Cu(1ML)]_{2} was converged up to 2000 
k-points in BZ, ndiv= (25 25 3) without any troubles.

I would be very much obliged to you if you could help me to avoid this 
problem. 

Please, find the files "case.struct" and "case.dayfile"  for 
[Ni(1ML)/Cu(1ML)]_{3} in the attachment.

Thank you in advance.

Best regards,

Nataliya Dadoenkova.