Author Topic: possible bug?  (Read 150 times)

0 Members and 1 Guest are viewing this topic.

Offline k.cerda

  • New ATK user
  • *
  • Posts: 9
  • Country: us
  • Reputation: 0
    • View Profile
possible bug?
« on: August 29, 2021, 07:03 »
I've used counterpoise LCAO calculator to simulate a big structure and took over 36 hrs. I now have the hdf5 file of the converged calculation. When I use the existing converged hdf5 file to calculate analyses such as electron density, potential and many more, it works as expected, uses the converged data to extract these quantities very fastly. But when I try to calculate the total energy of the system, ATK doesn't use the converged data to calculate the total energy, instead it re-starts the counterpoise LCAO SCF loops again which takes lots of hours to complete. Is this a bug or is there a setting to calculate the total energy from converged data of counterpoise corrected LCAO calculation? Edit: The following total energy calculation triggers a restart of the counterpoise LCAO calculation:
Code
# -------------------------------------------------------------
# Analysis from File
# -------------------------------------------------------------
path = '/home/kcerda/20082021/converged.hdf5'
configuration = nlread(path, object_id='MoleculeConfiguration_0')[0]

# -------------------------------------------------------------
# Total Energy
# -------------------------------------------------------------
total_energy = TotalEnergy(configuration)
nlsave('/home/kcerda/20082021/convergedTE.hdf5', total_energy)
nlprint(total_energy)
whereas the following calculation uses the converged data and calculates the potential instantly without a restart of SCF loops:
Code
# -------------------------------------------------------------
# Analysis from File
# -------------------------------------------------------------
path = '/home/kcerda/20082021/converged.hdf5'
configuration = nlread(path, object_id='MoleculeConfiguration_0')[0]

# -------------------------------------------------------------
# Electrostatic Potential
# -------------------------------------------------------------
electrostatic_potential = ElectrostaticPotential(configuration)
nlsave('/home/kcerda/20082021/convergedPOT.hdf5', electrostatic_potential)
« Last Edit: August 29, 2021, 07:08 by k.cerda »

Offline sureshko

  • QuantumATK Staff
  • New ATK user
  • *****
  • Posts: 4
  • Country: dk
  • Reputation: 0
    • View Profile
Re: possible bug?
« Reply #1 on: September 13, 2021, 10:38 »
Thanks for letting us know. Yes, it is a bug and we are updating the TotalEnergy Analysis object to work for CP correction calculators as well.

Offline k.cerda

  • New ATK user
  • *
  • Posts: 9
  • Country: us
  • Reputation: 0
    • View Profile
Re: possible bug?
« Reply #2 on: September 14, 2021, 13:11 »
Thanks for your reply.