SIGSEGV segmentation fault

DL_POLY
#1

Hi guys,

I m trying to run NpT/NsT Berendsen simulations of a DBT doped Anthracene crystal on dlpoly 4.09 on windows 10 (via windows’ cmd). When I simulate a pure Anthracene crystal the simulations run smoothly.
However, if I try to add a DBT molecule in the simulation I get the same problem as Julien:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0xffffffff
#1 0xffffffff
#2 0xffffffff
#3 0xffffffff
#4 0xffffffff
#5 0xffffffff
#6 0xffffffff
#7 0xffffffff
#8 0xffffffff
#9 0xffffffff
#10 0xffffffff
#11 0xffffffff
#12 0xffffffff
#13 0xffffffff
#14 0xffffffff
#15 0xffffffff
#16 0xffffffff

no matter how many steps I use.
Does anyone have any idea?

Thanks,
Tasos

#2

Dear Tasos,

I would suggest to update to 4.10.0 to be sure we do not hunt some strange bugs.

while your error looks like Julien is not the same since you can run some simulations. when you update to 4.10.0 you may get the same.

now on your error without input and output files is difficult. please email them to [email protected]
also since you are using Berendsen please read this
https://pubs.acs.org/doi/abs/10.1021/acs.jctc.8b00446

Regards,
Alin

Without Questions there are no Answers!

#3

Dear Tasos,

first please update, you really use some old version of dlpoly.

second after looking at your inputs the periodic boundary conditions
are wrong. load the working and non working and you will see what I
mean. Your molecules at the fringe of the box are broken.

Regards,
Alin

Without Questions there are no Answers!

#4

Hi Alin,

The molecules are broken on purpose, because there is no other way to contain Anthracene’s formation in a parallelepiped cell. So, the boundary conditions should be fine.
Could there be any other reason for the error?

Thanks,
Tasos

#5

Dear Tasos,

attached two pictures of your system pure and dopped.

pure
pure1846×1506 455 KB

dopped
dopped1967×1296 885 KB

vesta is smart enough to complete the molecules over the PBC.
as you can see that does not happen because i suspect the box is simply to big and “breaks” the molecules. do not confuse this with wrapping.

if you look at the output you can see the energy all over. and then a dihedral failure. all these indicate an input issue.

Regards,
Alin

#6

Dear Tasos,

since I cannot post pictures being to big I answered here

Alin

Without Questions there are no Answers!

#7

Thanks a lot Alin. It makes sense. I’ll try to have another look on my PBC then.

Cheers,
Tasos