[lammps-users] Error: STOP Too many torsion angles statement executed

LAMMPS LAMMPS Mailing List Mirror
1

hi,

I am trying to simulate a “block” of alpha phase RDX molecules using REAXFF potential, with shrink wrapped boundaries on all sides. Beyond a certain system size (the highest i could simulate was ~200 molecules) , LAMMPS reports the error:

" STOP Too many torsion angles statement executed"

I couldn’t find an explanation in the errors section (http://lammps.sandia.gov/doc/Section_errors.html#9_1) of the manual. The input file and the data file is attached with this email.

I have looked at the structure created from the data file and it seems to be correct. It would great if someone can point out the reason for the error.

thanks in advance.

in.reax.rdx (393 Bytes)

testmod2.data.rdx (214 KB)

2

This is an error internal to the ReaxFF library, not a LAMMPS error, per se.

I think the solution is to recompile the Reax lib with a boosted setting, but
don't recall the details. Aidan - can you help?

However, I think this also indicates you are running an unphysical system.

Steve

3

This is an error that is generated when the total number of torsion angles
on a processor exceeds the value specified in lib/reax/reax_defs.h:

#define NTOMAXDEF 65000

You can increase this number and recompile, or you can try running on more
processors.

Aidan

4

hi

Thanks for the help, Steve & Aidan.

Following Steve’s suggestion, i was looking at the structure to see of if the system was unphysical. I created the structure by replicating the single molecule in the LAMMPS example file according to the P centered orthorhombic structure.

On some investigation, i see that there is a disparity in the structure provided in the example file and the classic reference for RDX structure (Choi and Prince , Acta Crystallographica B28, 1972, 2857). On an “eyeball norm” i can see that according to the reference one of the nitro group is at an angle of 18 degrees with respect to the carbon plane, while in the structure provided with the examples, the the three nitro groups seems to be at same angle.

Is the structure optimized in any way?

I dont know if this is of any concern, i am just being curious.

5

Aidan will have to answer this one - he has better eyeballs for RDX.

Steve

6

There is nothing "unphysical" about the error message you observed. It is a
natural consequence of allocating memory at compile time, which is what the
ReaxFF implementation does. Try increasing NTOMAXDEF or run on more
processors.

Independent of that, you should convince yourself that the RDX structure
that you are using is correct. That's your job, not ours.

Aidan