bulk cell opt questions

[Jing Liu]

Dear CP2K users

Recently I'm trying to run the cell optimization for a organic material (Tetraazapyrene). My goal is to optmize the cell parameters of this bulk system. 

However during the test run there are few things that I could not explain, it's very likely that I missed something so if anyone could point them out would be very appreciated.

The test cases are:

Test_1: using the cif file, set pbc x y z both in &CELL and &POISSON, initial cell parameters are taken from cif file, no super cell

Test_2: using the same cif file as Test_1, set pbc x y z both in &CELL and &POISSON, initial cell parameters are taken from the same cif file as Test_1, 2*2*2 super cell, set charge constraint to 0

Reference: single TAPP molecule, geo_opt with KOALA at cc2 level

The problems are:

In Test_1: total energy of the converged run greatly dffered with respect to the reference run. They of course should not be identical but also should be close. Total energy of this run is: -340.2863 H. 

In Test_2: Total energy difference increased, total charge is also wrong despite the constraint was applied. Total energy of this is: -2722.1877 H, total charge of this run is: 0 (for Mulliken) and 1.2 (for Hirshfeld)

In Reference: Total energy is: -1061.3367 H.

All the inp and out files are attached.

[Fabian Ducry]

Dear Jing,

Some pointers to improve your simulations:

 - The charge you find is an artefact of the numerical inaccuracy in your system. EPS_DEFAULT should be tighter, just leave it at the default value of 1.0E-10
 - How did you choose CUTOFF and REL_CUTOFF, did you follow https://www.cp2k.org/howto:converging_cutoff? For CELL_OPT you might need to increase these numbers

 - SZV is a very small basis set, useful for testing but not for anything else. DZVP often produces okish results but not always. You should check the convergence of the basis set.

The total energy is very much dependent on the basis set that you use. So you cannot compare the energy of your "reference" with the one from cp2k and expect them to be the same.

Best,

Fabian

[Thomas Kühne]

Dear Jing, 

honestly I don’t fully understand what you are testing and what is the rational behind it. 

However, first of all you can’t compare the absolute energies (there are not even 

expected to be close), since you are comparing a pseudopotential calculation with an 

all electron reference. Also, I suspect that your reference is a gas phase calculation, 

which can neither compared to your periodic test no. 1 (because of pbc) and also 

not to test no. 2, which is not a proper isolated supercell due to too small cell dimensions 

along the x-direction.

Best, 

Thomas

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/73b9a873-5d94-4fff-a37d-61ba48721201%40googlegroups.com.
<hth.cif><hth_supercell.inp><hth.inp><hth_bulk_cell_opt_supercell_reduced.out><hth_bulk_cell_opt_reduced.out><koala_cc2_geo_opt_last_iter_reduced.out>

==============================

Thomas D. Kühne

Dynamics of Condensed Matter

Chair of Theoretical Chemistry

University of Paderborn

Warburger Str. 100

D-33098 Paderborn

Germany

tdku...@mail.upb.de

+49/(0)5251/60-5726

[Jing Liu]

Dear Fabian

Thank you for your suggestions, I'm running new tests based on them and hopefully the charge inaccuracy can be solved...

As for the basis set, I agree that they should not be directly compared since different basis set of course yield different total energy, however I expect somewhat "smaller" difference, not few hundred hartree... With this much of difference I can only assume one of the calclation is completely nonsense...

Best regards

Liu Jing

在 2020年4月27日星期一 UTC+2下午5:05:03，Fabian Ducry写道：

[Jing Liu]

Dear Thomas

The test cases I have ran are just test cases, their purpose are to test the input for the real work. So ideal is to run them with smallest basis set, get the results as fast as possible, check if there is any abnormal happens (like the non-zero charge in this case). If everything worked then I can move to the real case with sufficient basis set and functional.

You did make a good point on energy comparison...I have to think about the valid reference system...

Thank you for the suggestions!

Best regards

Liu Jing

在 2020年4月27日星期一 UTC+2下午11:42:16，tkuehne写道：

To unsubscribe from this group and stop receiving emails from it, send an email to cp...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/73b9a873-5d94-4fff-a37d-61ba48721201%40googlegroups.com.
<hth.cif><hth_supercell.inp><hth.inp><hth_bulk_cell_opt_supercell_reduced.out><hth_bulk_cell_opt_reduced.out><koala_cc2_geo_opt_last_iter_reduced.out>