Dear abinit friends,
Contrary to the manual (https://docs.abinit.org/variables/paw/#pawovlp), the default value of pawovlp is not correctly processed by abinit:
If pawovlp >5, then pawovlp is correctly processed.
If pawovlp <=5, then pawovlp is unexpectedly set to another value or otherwise incorrectly processed.
I encountered an overlap of 0.04%:
message: |
PAW COMPENSATION DENSITIES ARE OVERLAPPING !!!!
There are 2 pairs of overlapping atoms.
The maximum overlap percentage is obtained for the atoms 22 and 24.
| Distance between atoms 22 and 24 is : 2.17565
| PAW radius of the sphere around atom 22 is: 1.20000
| PAW radius of the sphere around atom 24 is: 1.41465
| This leads to a (voluminal) overlap ratio of 0.04 %
THIS IS DANGEROUS !, as PAW formalism assumes non-overlapping compensation densities.
I could only circumvent this by setting the pawovlp to 6. pawovlp equal to 5 or less led to above error message.
B.t.w. is it possible to force abinit to print also the default values to output file?
pawovlp is not equal to 5 per default
Moderator: bguster
-
woffermans
- Posts: 41
- Joined: Fri Jun 01, 2018 8:22 am
- Contact:
Re: pawovlp is not equal to 5 per default
Dear Willem,
On my side, the tests I did so far show that pawovlp can be increased up to 10 without much troubles, thought it is always advised to test if some results look unusual while changing the default.
Best wishes,
Eric
On my side, the tests I did so far show that pawovlp can be increased up to 10 without much troubles, thought it is always advised to test if some results look unusual while changing the default.
Best wishes,
Eric
-
woffermans
- Posts: 41
- Joined: Fri Jun 01, 2018 8:22 am
- Contact:
Re: pawovlp is not equal to 5 per default
Dear Eric,
Did you have an overlap ratio of 0.04 % and did you set pawovlp to 5 (default value)?
ABINIT should continue calculation with this setting and measurement, to my opinion.
Did you have an overlap ratio of 0.04 % and did you set pawovlp to 5 (default value)?
ABINIT should continue calculation with this setting and measurement, to my opinion.
Re: pawovlp is not equal to 5 per default
Dear Willem,
If the overlap is too big, this might increases the completeness of the basis and drives troubles or errors in some cases, this is why it was decided to stop the calculation to be sure the user is aware there is something going on. The default stop value has been set to 5 to be always on the safe side, but the user can then tune it by changing the pawovlp as he wants. I've already set pawovlp to 10 or even 15 in some of my cases, mostly during the relaxation and then I restart a calculation with a lower value if I want to do DFPT or anything else that will use the ground state WF.
Best wishes,
Eric
If the overlap is too big, this might increases the completeness of the basis and drives troubles or errors in some cases, this is why it was decided to stop the calculation to be sure the user is aware there is something going on. The default stop value has been set to 5 to be always on the safe side, but the user can then tune it by changing the pawovlp as he wants. I've already set pawovlp to 10 or even 15 in some of my cases, mostly during the relaxation and then I restart a calculation with a lower value if I want to do DFPT or anything else that will use the ground state WF.
Best wishes,
Eric