Let me take a deeper look some time today.
Ray Shan, Ph.D.
Support Scientist, Materials Design, Inc.
+1 760 495 4924 Ext. 219
rshan@…1795…
You are using fix reax/c/species with the settings “10 10 100” so that neighbor list reconstruction is going to be invoked every 100 steps. This means your neighbor list might not be reconstructed frequent enough. Try with “1 1 100” and re-run your simulation.
As Aidan pointed out, your tdamp and pdamp values are still too large. Try with a stronger (I.e. Smaller values) damping. Why is it that your box is shrinking to almost half of its original volume under NPT? Does it imply that your initial structure is too far from equilibrium? This is something you need to work out yourself.
In brief: these are the things I would try next:
- Change fix reax/c/species and neigh_modify commands so that you re-build neighbor list more frequently.
- Change your NPT damping factors.
- Figure out why your volume contracted so much during the second stage.
Ray
It is always better to make Pdamp too big rather than too small. You could try removing fix reax/c/species to see if that helps.
Aidan