[CP2K:2408] Re: NaNs in wave function cube files
Laino Teodoro
teodor... at gmail.com
Thu Nov 19 21:08:36 UTC 2009
Dear Claudio,
what you said:
> It is only when I tried things with periodic HF (through b3lyp) using
> 6-31g** as you can see in the example below that problems came up.
and
> This happens whether you use cutoff or you don't. Energies with and
> without cutoff are very similar so this is not really a problem.
is not correct.
The reality is that if you try a plain GGA-DFT with the same GAPW
setup of your input file it does NOT work!
This means that HFX is NOT the source of this problem.
The problem is a bug (an arithmetic exception) which pops up with
GAPW for certain values of the GTO (6-31Gxx) exponents: this is the
source of the NaNs.
The input here below reproduces the problem (no HFX).
Teo
!---------------------------------------------------
&FORCE_EVAL
METHOD Quickstep
&DFT
LSD
CHARGE -1
MULTIPLICITY 2
BASIS_SET_FILE_NAME EMSL_BASIS_SETS
POTENTIAL_FILE_NAME POTENTIAL
&MGRID
! CUTOFF 200
! REL_CUTOFF 50
&END MGRID
&QS
METHOD GAPW
EPS_PGF_ORB 1.0E-16
&END QS
&PRINT
&MO_CUBES
NHOMO 5
NLUMO 5
&END
&E_DENSITY_CUBE
&END
&END PRINT
&SCF
EPS_SCF 1.0E-4
SCF_GUESS RESTART
MAX_SCF 1
&OT
&END OT
&END SCF
&XC
&XC_FUNCTIONAL BLYP
&END XC_FUNCTIONAL
&END XC
&END DFT
&SUBSYS
&CELL
ABC 31.000 31.000 31.000
PERIODIC XYZ
&END CELL
&COORD
C 2.724300 -0.210525 1.679657
N 3.378813 0.142337 0.367635
C 2.628981 -0.547029 -0.747413
C 3.308675 1.641065 0.186405
C 4.824483 -0.335730 0.427386
C 5.707747 0.076171 -0.750114
C 7.039619 -0.686925 -0.723046
N -2.115674 0.098149 0.566927
S -1.065879 1.188462 -0.004405
O -0.096986 1.688012 0.992480
O -0.543935 0.938005 -1.360293
S -1.679594 -1.425080 0.861473
O -0.223810 -1.659926 0.878041
O -2.517197 -1.975152 1.928867
C -2.318886 2.581971 -0.206516
C -2.273741 -2.356150 -0.672280
F -3.277763 2.255810 -1.085288
F -1.666785 3.662915 -0.676494
F -2.893950 2.911808 0.957784
F -2.046785 -3.675730 -0.512894
F -1.628775 -1.949980 -1.775445
F -3.592764 -2.180301 -0.862609
H 3.269846 0.294602 2.481570
H 2.778648 -1.293048 1.815246
H 1.683606 0.110359 1.652541
H 1.595470 -0.201469 -0.749459
H 2.666731 -1.624615 -0.575470
H 3.104631 -0.308923 -1.698754
H 3.927011 2.115369 0.954315
H 2.265541 1.948040 0.284571
H 3.693613 1.904240 -0.800118
H 4.768472 -1.425574 0.513280
H 5.238050 0.053710 1.365517
H 5.910151 1.153814 -0.709845
H 5.210145 -0.119752 -1.706544
H 7.575766 -0.520996 0.219045
H 6.886641 -1.766428 -0.838317
H 7.681022 -0.342145 -1.545307
&END COORD
&KIND H
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&KIND F
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&KIND C
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&KIND N
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&KIND S
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&KIND O
BASIS_SET 6-31Gxx
POTENTIAL ALL
&END KIND
&END SUBSYS
&END FORCE_EVAL
&GLOBAL
PROJECT HF-ion-pair
PRINT_LEVEL MEDIUM
RUN_TYPE ENERGY
ECHO_INPUT
&END GLOBAL
On 19 Nov 2009, at 17:58, Claudio wrote:
> Normal dft calculations with pbc have always worked fine for me
> ( though basis set was the GTH).
>
> The reason that I am doing this is because I have been experimenting
> with SIC and for my problem this seems to play a very important role
> so I wanted to compare results of PBE including SIC with those from
> b3lyp since b3lyp should fix some of the self interaction problems
> (without needing SIC).
>
> It is only when I tried things with periodic HF (through b3lyp) using
> 6-31g** as you can see in the example below that problems came up.
>
> This happens whether you use cutoff or you don't. Energies with and
> without cutoff are very similar so this is not really a problem.
>
> BTW, what's not correct about the input below?
> It's a pair of ions with an extra electron in a box that is larger
> than the system and a cutoff that is less than 1/2 of the box. This
> should be a valid calculation right?
>
>
> Thanks!!
> Claudio
>
> On Nov 19, 9:45 am, mguidon <manuel... at gmail.com> wrote:
>> Hi Claudio.
>>
>> Does the problem also occur if you run with appropriate periodic
>> settings (i.e. correct box size and correct cutoff radius for the
>> truncated Coulomb operator) ?
>> What about a pure GGA calculation without HFX. Does this work?
>>
>> Manuel
>>
>> On Nov 19, 1:36 pm, Claudio <marg... at gmail.com> wrote:
>>
>>
>>
>>> BTW, I noticed there are no NaNs with 3-21g* instead.
>>
>>> Any ideas?
>>
>>> Thanks!
>>
>>> Claudio
>>
>>> On Nov 18, 3:56 am, Claudio <marg... at gmail.com> wrote:
>>
>>>> Dear all, thanks for your help trying to run periodic b3lyp.
>>
>>>> Though my run appears to converge, it writes seemingly correct
>>>> density
>>>> and spin cubes, wave function cubes are filled with NaNs after the
>>>> geometry block. A typical input file that will generate this errors
>>>> follows. The version of CP2K is yesterday's CVS. An older version
>>>> from a month ago also does this.
>>
>>>> &FORCE_EVAL
>>>> METHOD Quickstep
>>>> &DFT
>>>> LSD
>>>> CHARGE -1
>>>> MULTIPLICITY 2
>>>> BASIS_SET_FILE_NAME /home/claudiom/src/cp2k/tests/QS/
>>>> EMSL_BASIS_SETS
>>>> POTENTIAL_FILE_NAME /home/claudiom/src/cp2k/tests/QS/POTENTIAL
>>>> &MGRID
>>>> ! CUTOFF 200
>>>> ! REL_CUTOFF 50
>>>> &END MGRID
>>>> &QS
>>>> METHOD GAPW
>>>> EPS_PGF_ORB 1.0E-16
>>>> &END QS
>>>> &PRINT
>>>> &MO_CUBES
>>>> NHOMO 5
>>>> NLUMO 5
>>>> &END
>>>> &E_DENSITY_CUBE
>>>> &END
>>>> &END PRINT
>>>> &SCF
>>>> EPS_SCF 1.0E-4
>>>> SCF_GUESS RESTART
>>>> MAX_SCF 60
>>>> &OT
>>>> &END OT
>>>> &OUTER_SCF
>>>> MAX_SCF 1
>>>> &END OUTER_SCF
>>>> &END SCF
>>>> #These are the coefficients used for B3LYP using VWN3, this is the
>>>> default Gaussian definition, but not recommended
>>>> &XC
>>>> &XC_FUNCTIONAL
>>>> &LYP
>>>> SCALE_C 0.81
>>>> &END LYP
>>>> &BECKE88
>>>> SCALE_X 0.72
>>>> &END BECKE88
>>>> &VWN
>>>> FUNCTIONAL_TYPE VWN3
>>>> SCALE_C 0.19
>>>> &END VWN
>>>> &XALPHA
>>>> SCALE_X 0.08
>>>> &END XALPHA
>>>> &END XC_FUNCTIONAL
>>>> &HF
>>>> &SCREENING
>>>> EPS_SCHWARZ 1.0E-10
>>>> SCREEN_ON_INITIAL_P FALSE
>>>> &END SCREENING
>>>> &INTERACTION_POTENTIAL
>>>> POTENTIAL_TYPE TRUNCATED
>>>> CUTOFF_RADIUS 15.000
>>>> T_C_G_DATA /home/claudiom/src/cp2k/tests/QS/t_c_g.dat
>>>> &END INTERACTION_POTENTIAL
>>>> &MEMORY
>>>> MAX_MEMORY 768
>>>> EPS_STORAGE_SCALING 1.0E-1
>>>> &END MEMORY
>>>> FRACTION 0.20
>>>> &END HF
>>>> &END XC
>>>> &END DFT
>>>> &SUBSYS
>>>> &CELL
>>>> ABC 31.000 31.000 31.000
>>>> PERIODIC XYZ
>>>> &END CELL
>>>> &COORD
>>>> C 2.724300 -0.210525 1.679657
>>>> N 3.378813 0.142337 0.367635
>>>> C 2.628981 -0.547029 -0.747413
>>>> C 3.308675 1.641065 0.186405
>>>> C 4.824483 -0.335730 0.427386
>>>> C 5.707747 0.076171 -0.750114
>>>> C 7.039619 -0.686925 -0.723046
>>>> N -2.115674 0.098149 0.566927
>>>> S -1.065879 1.188462 -0.004405
>>>> O -0.096986 1.688012 0.992480
>>>> O -0.543935 0.938005 -1.360293
>>>> S -1.679594 -1.425080 0.861473
>>>> O -0.223810 -1.659926 0.878041
>>>> O -2.517197 -1.975152 1.928867
>>>> C -2.318886 2.581971 -0.206516
>>>> C -2.273741 -2.356150 -0.672280
>>>> F -3.277763 2.255810 -1.085288
>>>> F -1.666785 3.662915 -0.676494
>>>> F -2.893950 2.911808 0.957784
>>>> F -2.046785 -3.675730 -0.512894
>>>> F -1.628775 -1.949980 -1.775445
>>>> F -3.592764 -2.180301 -0.862609
>>>> H 3.269846 0.294602 2.481570
>>>> H 2.778648 -1.293048 1.815246
>>>> H 1.683606 0.110359 1.652541
>>>> H 1.595470 -0.201469 -0.749459
>>>> H 2.666731 -1.624615 -0.575470
>>>> H 3.104631 -0.308923 -1.698754
>>>> H 3.927011 2.115369 0.954315
>>>> H 2.265541 1.948040 0.284571
>>>> H 3.693613 1.904240 -0.800118
>>>> H 4.768472 -1.425574 0.513280
>>>> H 5.238050 0.053710 1.365517
>>>> H 5.910151 1.153814 -0.709845
>>>> H 5.210145 -0.119752 -1.706544
>>>> H 7.575766 -0.520996 0.219045
>>>> H 6.886641 -1.766428 -0.838317
>>>> H 7.681022 -0.342145 -1.545307
>>>> &END COORD
>>>> &KIND H
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &KIND F
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &KIND C
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &KIND N
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &KIND S
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &KIND O
>>>> BASIS_SET 6-31Gxx
>>>> POTENTIAL ALL
>>>> &END KIND
>>>> &END SUBSYS
>>>> &END FORCE_EVAL
>>>> &GLOBAL
>>>> PROJECT HF-ion-pair
>>>> PRINT_LEVEL MEDIUM
>>>> RUN_TYPE ENERGY
>>>> ECHO_INPUT
>>>> &END GLOBAL
>
> --
>
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To post to this group, send email to cp... at googlegroups.com.
> For more options, visit this group at http://groups.google.com/
> group/cp2k?hl=.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20091119/e9346712/attachment.htm>
More information about the CP2K-user
mailing list