cp2k vs. gaussian software

K.AK

unread,

Oct 9, 2023, 5:31:36 AM10/9/23

to cp2k

Dear experts,

I am performing DFT calculations using the QS method.

Also, I did the same optimization using Gaussian.

In the case of Gaussian software, the simulated conditions were below.

Basis set: 6-31G(d,p), Simulation: DFT, B3LYP

But, the results of atomic charge caluculated from Mulliken population are much different.

(I used ALL_POTINTIALS at cp2k/data/ as potentials.)

Please advise about the cp2k simulation method and input file.

I am putting in attachment my input.

Best wishes

K.AK

cp2k.inp

Jürg Hutter

unread,

Oct 10, 2023, 3:21:51 AM10/10/23

to cp...@googlegroups.com

Hi

I made some changes to your input:

EPS_ISO -> use the default value
&MGRID
CUTOFF 400
NGRIDS 4
REL_CUTOFF 60
&END MGRID
I used the default interaction potential. You probably should use
a setting for an isolated molecule (Poisson solver, Periodicity NONE)
to get best comparison to QC programs.
I also used an cubic box, I don't see why you have does angles defined.

I calculated the charges at the initial geometry:

CP2K G16
1 C 0.298435 0.302041
2 C 0.058327 0.056900
3 N -0.521926 -0.521311
4 C 0.648801 0.647720
5 N -0.562059 -0.561722
6 C 0.486531 0.485664
7 F -0.297309 -0.295955
8 O -0.430784 -0.419662
9 H 0.303720 0.295938
10 O -0.441045 -0.438800
11 H 0.297841 0.299063
12 H 0.159468 0.150124

Maybe the different Mulliken charges are due to another problem,
e.g. different geometries after optimization?

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of K.AK <kosh...@gmail.com>
Sent: Monday, October 9, 2023 11:29 AM
To: cp2k
Subject: [CP2K:19322] cp2k vs. gaussian software

--
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<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/b76622de-96aa-4f99-8036-b6dcec35bba1n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/b76622de-96aa-4f99-8036-b6dcec35bba1n%40googlegroups.com?utm_medium=email&utm_source=footer>.

K.AK

unread,

Oct 10, 2023, 3:46:17 AM10/10/23

to cp2k

Dear Jürg Hutter & cp2k community
Thank you for reply.

I confirmed the structure after geometry optimization.

As attached snapshots, I get the much different structure of before and after optimization.

I am putting in attachment the obtained snapshots (left:before, right: after).

Probably, I think that the molecule connectivities did not keep the initial atoms under simulation.

In case of Gaussian program, I confirmed the consistment of structure with before and after optimization.

How do I solve this problem?

Please advise the method to solve it.

regards

K.AK

2023年10月10日火曜日 16:21:51 UTC+9 Jürg Hutter:

Jürg Hutter

unread,

Oct 10, 2023, 3:50:45 AM10/10/23

to cp...@googlegroups.com

Hi

use a larger box.
use a cubic box.
use PERIODIC NONE and a corresponding POISSON SOLVER

make the other changes I suggested.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of K.AK <kosh...@gmail.com>

Sent: Tuesday, October 10, 2023 9:46 AM
To: cp2k
Subject: Re: [CP2K:19332] cp2k vs. gaussian software

Dear Jürg Hutter & cp2k community
Thank you for reply.

I confirmed the structure after geometry optimization.

As attached snapshots, I get the much different structure of before and after optimization.
I am putting in attachment the obtained snapshots (left:before, right: after).

Probably, I think that the molecule connectivities did not keep the initial atoms under simulation.
In case of Gaussian program, I confirmed the consistment of structure with before and after optimization.

How do I solve this problem?
Please advise the method to solve it.

regards

K.AK[before optimization.png][after optimization.png]

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/6cd294cb-df87-42fc-940a-5915617eba91n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/6cd294cb-df87-42fc-940a-5915617eba91n%40googlegroups.com?utm_medium=email&utm_source=footer>.

after optimization.png

before optimization.png

K.AK

unread,

Oct 10, 2023, 11:02:21 AM10/10/23

to cp2k

Dear Jürg Hutter & cp2k users

Thank you for your suggestion.

I try to do simutation with the set conditions you suggested.

regards

K.AK

2023年10月10日火曜日 16:50:45 UTC+9 Jürg Hutter:

K.AK

unread,

Oct 10, 2023, 8:50:47 PM10/10/23

to cp2k

Dear Jürg Hutter & cp2k experts

I tried to simulate optimization with suggested conditions.

But, the atomic charge were a little different in case of gaussian or cp2k software.

I confirmed the structure consistment with before and after optimization.

The results of mulliken atomic sharge are below.

Gaussian cp2k
F -0.27228 -0.29413
O -0.476218 -0.424407

O -0.490938 -0.44221
N -0.571762 -0.522822
H 0.29534 0.29644
N -0.611851 -0.56184
H 0.298201 0.299782
C 0.268481 0.302454
C 0.569844 0.487981
C 0.747199 0.650784
C 0.092956 0.057482
H 0.151035 0.150487

What did the difference of charges be caused?

Is this charge difference an acceptable difference in DFT field?

regards

K.AK

2023年10月11日水曜日 0:02:21 UTC+9 K.AK:

Lucas Lodeiro

unread,

Oct 10, 2023, 11:21:48 PM10/10/23

to cp...@googlegroups.com

AFAIK, a complete match between both codes is a little bit difficult, due to their different approaches to compute terms, use of GAPW scheme among others. But take into account that B3LYP in G program is defined in a "special" way ussing the VWN3 functional instead VWN5, the later is the "correct" one (as is stated in the VWN paper) and the most used definition of B3LYP in most programs... This difference in the functional definition triggers some appreciable differences, and have to taken into account.

For a closer matching with G program you need to use the VWN3 instead of VWN5, check it.

Regards - Lucas

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/22976020-bd34-4b6f-9c42-d7d0744cbc7an%40googlegroups.com.

Jürg Hutter

unread,

Oct 11, 2023, 3:16:38 AM10/11/23

to cp...@googlegroups.com

Hi

for your reference, here are the charges I get after geometry optimization.
I used default settings in both codes.
G16 CP2K
1 C 0.302512 0.302685
2 C 0.056933 0.056588
3 N -0.521581 -0.521206
4 C 0.650616 0.649432
5 N -0.561838 -0.561196
6 C 0.487665 0.486565
7 F -0.293979 -0.294131
8 O -0.424335 -0.423591
9 H 0.296440 0.296449
10 O -0.442591 -0.441756
11 H 0.299573 0.299576
12 H 0.150585 0.150585

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of K.AK <kosh...@gmail.com>

Sent: Wednesday, October 11, 2023 2:50 AM
To: cp2k
Subject: Re: [CP2K:19346] cp2k vs. gaussian software

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/22976020-bd34-4b6f-9c42-d7d0744cbc7an%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/22976020-bd34-4b6f-9c42-d7d0744cbc7an%40googlegroups.com?utm_medium=email&utm_source=footer>.

K.AK

unread,

Oct 11, 2023, 3:56:39 AM10/11/23

to cp2k

Dear Jürg Hutter & cp2k experts

Thank you for your help.

Your results are great consistment.

but, I did not get the same results.

Probably, I think the accurate conditions for gaussian and cp2k program.

If possible, it would be helpful if you could attach the file calculation terms or input failes (gaussian: .gjf , cp2k: .inp).

regards

K.AK

2023年10月11日水曜日 16:16:38 UTC+9 Jürg Hutter:

Matt Watkins

unread,

Oct 11, 2023, 4:11:16 AM10/11/23

to cp2k

Hi,

in the pictures you posted the order of the atoms seems totally different between the two codes. Are you sure you have labelled / specified coordinates for your atoms consistently when running the two codes?

Matt

Jürg Hutter

unread,

Oct 11, 2023, 4:16:52 AM10/11/23

to cp...@googlegroups.com

Here is my G16 command line:

#N RB3LYP/6-31G** 5D opt

How about you posting your cp2k input file. Did you follow my
suggestions?

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of K.AK <kosh...@gmail.com>

Sent: Wednesday, October 11, 2023 9:56 AM
To: cp2k
Subject: Re: [CP2K:19350] cp2k vs. gaussian software

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/9ec510a1-2163-4d0d-b228-4bc62a73edf2n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/9ec510a1-2163-4d0d-b228-4bc62a73edf2n%40googlegroups.com?utm_medium=email&utm_source=footer>.

K.AK

unread,

Oct 11, 2023, 4:40:24 AM10/11/23

to cp2k

Dear Jürg Hutter & cp2k experts

Thank you for your help.

In gaussian, I do not use the restricted and 5D route sections.

My cp2k input is below.

&GLOBAL
PROJECT DFT
RUN_TYPE GEO_OPT
PRINT_LEVEL LOW
&END GLOBAL
&FORCE_EVAL
METHOD QS
&SUBSYS
&CELL
ABC 30 30 30
&END CELL
&COORD
F 10.69500000 2.27100000 4.22400000
O 8.93200000 4.13200000 5.23700000
O 7.05100000 4.66200000 1.12700000
N 8.78700000 3.21200000 1.33800000
H 8.75500000 3.01000000 0.50200000
N 8.01900000 4.37700000 3.16800000
H 7.47000000 4.95200000 3.49700000
C 9.79100000 2.87300000 3.41700000
C 8.91600000 3.82000000 4.04300000
C 7.89700000 4.11800000 1.82400000
C 9.73100000 2.60600000 2.12200000
H 10.33200000 2.00400000 1.74400000
&END COORD
&TOPOLOGY
&CENTER_COORDINATES
&END CENTER_COORDINATES
&END TOPOLOGY
&KIND H
BASIS_SET 6-31Gdp
POTENTIAL ALL
&END KIND
&KIND C
BASIS_SET 6-31Gdp
POTENTIAL ALL
&END KIND
&KIND F
BASIS_SET 6-31Gdp
POTENTIAL ALL
&END KIND
&KIND N
BASIS_SET 6-31Gdp
POTENTIAL ALL
&END KIND
&KIND O
BASIS_SET 6-31Gdp
POTENTIAL ALL
&END KIND
&END SUBSYS
&DFT
BASIS_SET_FILE_NAME ./BASIS_SET_6-31G_DP_FNCHO
POTENTIAL_FILE_NAME ./ALL_POTENTIALS
CHARGE 0
&QS
METHOD GAPW
EPS_DEFAULT 1.0E-14
EPSFIT 1.0E-8
&END QS
&XC
&XC_FUNCTIONAL
&LYP
SCALE_C 0.81 ! 81% LYP correlation
&END
&BECKE88
SCALE_X 0.72 ! 72% Becke88 exchange
&END
&VWN
FUNCTIONAL_TYPE VWN3
SCALE_C 0.19 ! 19% LDA correlation
&END
&XALPHA
SCALE_X 0.08 ! 8% LDA exchange
&END
&END XC_FUNCTIONAL
&HF
&SCREENING
EPS_SCHWARZ 1.0E-12
&END
&MEMORY
MAX_MEMORY 100
&END
&INTERACTION_POTENTIAL ! for condensed phase systems
POTENTIAL_TYPE TRUNCATED ! should be less than halve the cell
CUTOFF_RADIUS 4.0 ! data file needed with the truncated operator
T_C_G_DATA ./t_c_g.dat
&END
FRACTION 0.20
&END HF
&END XC

&MGRID
CUTOFF 400
NGRIDS 4
REL_CUTOFF 60
&END MGRID

&POISSON ! Solver requested for non periodic calculations
PERIODIC NONE
PSOLVER MT ! Type of solver MT > WAVELET
&END POISSON
&SCF
EPS_SCF 1.0E-07
MAX_SCF 20
SCF_GUESS ATOMIC ! Change the initial guess for the wavefunction. RESTART > ATOMIC
&OT
MINIMIZER DIIS
PRECONDITIONER FULL_ALL
&END OT
&PRINT
&RESTART OFF
&END RESTART
&END PRINT
&END SCF
&PRINT
&MULLIKEN ON
FILENAME ./mulliken.dat
PRINT_ALL
&END MULLIKEN
&END PRINT
&END DFT
&END FORCE_EVAL
&MOTION
&GEO_OPT
TYPE MINIMIZATION
MAX_DR 1.8E-03
MAX_FORCE 4.5E-04
RMS_DR 1.2E-03
RMS_FORCE 3.0E-04
OPTIMIZER CG
&END GEO_OPT
&CONSTRAINT
&FIXED_ATOMS
COMPONENTS_TO_FIX XYZ
LIST 1
&END FIXED_ATOMS
&END CONSTRAINT
&END MOTION

Are this input file accurate and consistment with your input?

regards

K.AK

2023年10月11日水曜日 17:16:52 UTC+9 Jürg Hutter:

Jürg Hutter

unread,

Oct 11, 2023, 4:48:05 AM10/11/23

to cp...@googlegroups.com

Hi

you need to use 5D for G16 or else you have a different basis set!

for the CP2K input I suggest:

Wavelet solver and a box of 15 A cube.
Don't use the truncated Coulomb option (INTERACTION_POTENTIAL) !

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of K.AK <kosh...@gmail.com>

Sent: Wednesday, October 11, 2023 10:40 AM
To: cp2k
Subject: Re: [CP2K:19353] cp2k vs. gaussian software

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/d57bc2d6-a1db-4514-a0e7-d59f510b7041n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/d57bc2d6-a1db-4514-a0e7-d59f510b7041n%40googlegroups.com?utm_medium=email&utm_source=footer>.

K.AK

unread,

Oct 11, 2023, 9:45:22 AM10/11/23

to cp2k

Dear Jürg Hutter , Lucas Lodeiro , Matt Watkins, &cp2k users

Thank you for helpful advice.

I confirm the input file and try to simulate optimization with the suggested input information.

regards

K.AK

2023年10月11日水曜日 17:48:05 UTC+9 Jürg Hutter:

Reply all

Reply to author

Forward