OPTGEOM fail during ELASTCON
-
Hi,
I encountered a problem with the geometry optimization failing during Elasticity calculation.
(ERROR **** OPTGEOM **** SCF FAILED - OPTGEOM STOP). I tried to insert NOOPTGEOM Keyword to skip optimization and test input, but it still tries to optimize.
If someone has tips I am happy to try.
I use CRYSTAL14. -
Hi,
The keyword PRESSURE requires the keyword PREOPTGEOM to pre-optimize the structure at the desired pressure of 15 GPa.
It is recommended to increase the level of printing in the output with the PRINT keyword to better see what is happening.
Please, try with the following input:
ELASTCON
PRINT
PREOPTGEOM
PRESSURE
15
END
ENDand let me know what you get.
-
Hi,
I got the new output, but it seems to have the same error code and the PRINT command did not really add new information as far as I could see. -
So, I tried multiple stuff:
when maintaining the geometry input including supercell and just doing geometry optimization the described error remians.
I constrained the optimization to just do optimization of lattice constants in the hope it would help, as the atoms are on special positions.
I removed the supercell description and got the CHOLSK error (basis set linearly dependent) I used Towler basis set, so I changed to pobTZVP basis, and now I get the error message 'ERROR **** f90main **** GEOMETRY OPTIMIZATION FAILED' (what does this mean?)
I use the EXTPRESS command during optimization because I am considering high pressure behaviour.
Fe_hcp_geom_0.d12
Fe_hcp_geom_0.out -
Dear denkos,
Your calculation stopped when the code reached the following part:
... ****************************************************************** * OPT END - FAILED * E(AU): -2.527750102645E+03 POINTS 100 * ****************************************************************** ...The default maximum number of optimisation steps is 100. So the calculation didn't actually fail, it just ran out of optimisation steps. This can be increased with setting the MAXCYCLE keyword to a higher number (in your OPTGEOM block).
Also, it is always worth monitoring the convergence criteria for the gradients. If you, for example, grep the behaviour of the RMS on the gradient for the last 10 steps in your geometry optimisation and you will find:
MAX GRADIENT 0.002030 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.001267 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.000622 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.000342 THRESHOLD 0.000450 CONVERGED YES MAX GRADIENT 0.000778 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.001712 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.001129 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.000575 THRESHOLD 0.000450 CONVERGED NO MAX GRADIENT 0.000332 THRESHOLD 0.000450 CONVERGED YES MAX GRADIENT 0.000716 THRESHOLD 0.000450 CONVERGED NOThat way you can get insights into what is going on with the geometry (you can also load the output file into MOLDRAW and see the geometry/atom positions being updated). You might want to tighten the step size to avoid jumps when the geometry is close to a (local) minimum.
Hope this helps.
Cheers,
Aleks
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login