When building a simple water dimer I obtain an error from proc.py such as:
ValidationError: Cannot compute projection of different symmetries.
The following has been obtained in python notebook and it is obtained in spite of having specified keywords to neglect symmetry.
w2=psi4.geometry("""
0 1
O
H 1 dOH
H 1 dO1H 2 aHOH
O 1 dOO 2 -100.0 3 180.0
H 4 dOH 3 120.0 2 140.0
H 4 dOH 5 aHOH 1 140.0
no_reorient
no_com
symmetry c1
""")
w2.aHOH=104.5
w2.dO1H=1.1
w2.dOH=0.96
w2.dOO=3.0
psi4.set_memory('1gb')
psi4.set_output_file('w2_opt.log', False)
psi4.energy('b3lyp/6-31g*', molecule=w2)
It appears that the error was due to some previously existing geometry - probably without the “no symmetry” labels and it kept popping up even when using psi4.clean(). However, after restarting the session it went fine.
The keywords here are “obtained in python notebook” and “due to some previously existing geometry - probably without the ‘no symmetry’ labels.”
The matter is exactly as you suppose. You ran a previous computation with symmetry and had the SCF guess set to read in a fragment of the notebook that you haven’t shown us. Psi4 complains because it doesn’t know how to project the orbital guess down to a lower point group, so it can’t carry out the orbital read operation you requested Psi4 to do.
I’m seeing a story about how forgetting about state in a Python notebook can come back to bite you. Was there something you wanted developers to do about this topic…?
thank you!
well I assumed psi4.clean() would do the job but not quite … the issue I had was quite “persistent” as even changing the molecule and apply the .clean() it seemed not to be able to forget about previous state. Thank you once again