Doubts related to energy trends with XC functionals and cutoffs

[SRKC Sharma Yamijala]

Dear CP2K users and developers,

I am trying to calculate the energy difference between Au20 clusters
(tetrahedral and planar geometries i.e. Et - Ep) on a graphene quantum dot
substrate. After performing the initial tests with both BLYP functional and
PBE functional with 60 Ry Rel_cutoff, I got the below results.

       With BLYP functional                  With PBE functional:
  Cutoff (Ry)      Et - Ep (eV)
  300.000000        0.476312             300.000000        0.836360
  350.000000        0.478999             350.000000        0.838100
  400.000000        0.482285             400.000000        0.840190
  450.000000        0.484214             450.000000        0.840969
  500.000000        0.481904             500.000000        0.840789
  550.000000        0.485268             550.000000        0.841879
  600.000000        0.485097             600.000000        0.842565
  650.000000        0.485540             650.000000        0.842448
  700.000000        0.485891             700.000000        0.842935
  750.000000        0.485847             750.000000        0.842933
  800.000000       -1.372659             800.000000       -1.218138


Though it seems, from these results, that 550 Ry is sufficient for
calculating energy differences, I do have a doubt regarding the 800 Ry.
Also, I want to let you know that all these calculations have been
performed with 1 SCF cycle as mentioned in the tutorial and the absolute
energies of each system got converged (with in 0.0001 Ha) only for 700 and
750 Ry.

If possible could you please let me know
(i) whether performing production runs with 550 Ry and 60 Ry for cutoff and
Relcutoff is okay or not.

(ii) Calculations show PBE and BLYP need the same cutoff for a converged
calculation. I have been repeatedly reading that BLYP needs more cutoff to
get better calculations in the forum. Anything wrong or is it  a special
case?

(iii) I found that there is no change in the absolute energy (with 0.0001
Ha), for any Rel_cutoff between 40 to 90 Ry (results not shown here). I
have tried with cutoffs 350 Ry and 700 Ry. For both of these cutoffs
absolute energies are not changing with relcutoff. Is it expected?

(iv) Can one expect a ~ 0.4 eV increment in the energy difference just by a
change in the XC functional.

 Input file is also attached for clarity.


Thanking you for your support,
Sincerely,
Sharma.

P.S. Sorry for the long mail.



********************************************************
*Chaitanya Sharma,*
*Prof. Pati'*s group,
Chemistry and Physics Materials unit,
JNCASR, BANGLORE,
Lab:: (080-2208) 2581, 2809
https://sites.google.com/site/sharmasrkcyamijala/
*********************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140924/d7f0240e/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: au20_planar_on_bi_layer_6N-GQD.inp
Type: chemical/x-gamess-input
Size: 2156 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140924/d7f0240e/attachment.inp>