Water Dimer dynamics using DFT fails to converge

[SAcompuser]

Hi all,

So I'm currently trying to set up a dynamics simulation of a water dimer (2 molecules of TIP3P waters) using DFT and I've been unable to get the simulation to converge which I imagine is a problem with the geometry setup but I can't see the problem. Can anyone help? I have attached the input pdf file along with the input script I use to prepare and the input script I use to try run the actual dynamics.

[Edoardo Aprà]

Did you disable symmetry in the input file?
It can be achieved with the following lines in the dft input section

dft
sym off
adapt off
end

[SAcompuser]

I have tried what you suggested, but getting the same error it appears. I have tried attached my input scripts and my starting pdb but keep getting errors so here is a copy of my dynamics input script:

echo

#memory global 8 mb stack 16 mb heap 10 mb

permanent_dir ./data

scratch_dir   ./scratch

title "Water Dimer DFT MD"

start  watdim

charge 0

geometry

load watdim_out.pdb

end

md

 system watdim

 vreass 1 298.15 fraction 1.0 once initial 

 step 0.001 equil 0 data 100

 print step 1 stat 1

 record coord 1

 iseed  126

 msa 10

end

dft

  direct

  xc b3lyp

  sym off

  adapt off

end

basis "ao basis" 

 * library "6-31G"

end

qmmm

 region qm

end

#task qmmm dft energy

#task qmmm dft optimize

task qmmm dft dynamics

analyze

 system watdim

 file watdim.trj

 copy watdim.crd

 write 5 watdim_final.pdb

end

 

#task analyze

[Edoardo Aprà]

Could you post the geometry pdb file, too?

[Edoardo Aprà]

Could you also post the complete output/error files in attachment?

Thanks

[SAcompuser]

Also this is the prepare file that I use before I try using DFT:

title "Water Dimer ab initio MD"

start watdim

charge 0

#----------------------------------------------------------------------

#-- SETUP

#----------------------------------------------------------------------

prepare

  charmm

  source ./coor/tip3p.pdb

#  solvent name TIP model TIP3

  new_top new_seq

  new_rst

  #orient

  modify segment 1 quantum

  modify segment 2 quantum

  update lists

  vacuo  # . Disables PBC.

#  periodic

#  boxsize

  write watdim_out.pdb

  write watdim.rst

end

task prepare

[SAcompuser]

Sure

[Edoardo Aprà]

Thanks for these files.

However, since you mentioned it's a DFT convergence problems., it would be better to get all the output and error files, too

[SAcompuser]

Here is the output

[Edoardo Aprà]

This geometry is really odd.

The distance O-H is above 4 angstroms ... are you sure that this is the geometry you want to use?

[SAcompuser]

Yeah you're right. I switched to using my original pdb input file in the dynamics step instead of the output pdb from prepare, and now it appears to converge and dynamics runs! I wonder why the output pdb from prepare is so weird. 

Thanks so much for your help!

[SAcompuser]

I'm always wondering whether it is possible to run a DFT dynamics simulation with periodic boundary conditions?