Excited state optimization by TDDFT

HB H

unread,

Sep 19, 2020, 10:37:15 PM9/19/20

to cp2k

Hi,

I am trying to optimize the geomitry of the first excited singlet state (S1) by TDDFT.

Is there a way to do it, such as below?

(I know we can calcualte the absorption spectrum by changing the RUN_TYPE from GEO_OPT to ENERGY.)

Thx!

&GLOBAL

PREFERRED_FFT_LIBRARY FFTW3

PREFERRED_DIAG_LIBRARY SL

PROJECT example

RUN_TYPE GEO_OPT

PRINT_LEVEL MEDIUM

&END GLOBAL

&FORCE_EVAL

METHOD Quickstep

&PROPERTIES

&TDDFPT

NSTATES 1

MAX_ITER 50

CONVERGENCE 1.0e-5

&MGRID

CUTOFF 300

&END MGRID

&END TDDFPT

&END PROPERTIES

&DFT

BASIS_SET_FILE_NAME GTH_BASIS_SETS

POTENTIAL_FILE_NAME POTENTIAL

LSD

CHARGE 0

MULTIPLICITY 1

&MGRID

CUTOFF 300

&END MGRID

&QS

METHOD GPW

EPS_DEFAULT 1.0E-12

EXTRAPOLATION ASPC

EXTRAPOLATION_ORDER 3

EPS_PGF_ORB 1.0E-20

&END QS

&SCF

MAX_SCF 50

SCF_GUESS RESTART

EPS_SCF 1.0E-6

&OT

MINIMIZER DIIS

Rotation

&END OT

&OUTER_SCF

MAX_SCF 10

EPS_SCF 1.0E-6

&END OUTER_SCF

&END SCF

&XC

&XC_FUNCTIONAL PBE

&END XC_FUNCTIONAL

&VDW_POTENTIAL

DISPERSION_FUNCTIONAL PAIR_POTENTIAL

&PAIR_POTENTIAL

TYPE DFTD3

CALCULATE_C9_TERM .TRUE.

REFERENCE_C9_TERM .TRUE.

LONG_RANGE_CORRECTION .TRUE.

VERBOSE_OUTPUT .FALSE.

REFERENCE_FUNCTIONAL PBE

PARAMETER_FILE_NAME dftd3.dat

R_CUTOFF 9

EPS_CN 1.0E-6

&END PAIR_POTENTIAL

&END VDW_POTENTIAL

&END XC

&PRINT

&MO_CUBES MEDIUM

NHOMO 3

NLUMO 3

ADD_LAST NUMERIC

&END MO_CUBES

&END PRINT

&END DFT

&SUBSYS

&CELL

ABC 18.81000 18.81000 18.81000

ALPHA_BETA_GAMMA 90 90 90

PERIODIC XYZ

&END CELL

&TOPOLOGY

CONNECTIVITY OFF

&GENERATE

CREATE_MOLECULES

&END GENERATE

COORD_FILE S0-GS-CS.xyz

COORD_FILE_FORMAT XYZ

&CENTER_COORDINATES TRUE

&END CENTER_COORDINATES

&END TOPOLOGY

&KIND C

BASIS_SET TZV2P-GTH

POTENTIAL GTH-PBE-q4

&END KIND

&KIND H

BASIS_SET TZV2P-GTH

POTENTIAL GTH-PBE-q1

&END KIND

&KIND O

BASIS_SET TZV2P-GTH

POTENTIAL GTH-PBE-q6

&END KIND

&KIND N

BASIS_SET TZV2P-GTH

POTENTIAL GTH-PBE-q5

&END KIND

&END SUBSYS

&END FORCE_EVAL

Matt W

unread,

Sep 20, 2020, 10:51:35 AM9/20/20

to cp2k

Hello,

excited state gradients are not currently implemented.

Matt

HB H

unread,

Sep 20, 2020, 9:51:36 PM9/20/20

to cp...@googlegroups.com

Dear Matt,

Thanks for your reply, that really helps!

However, I have one more question. When I was trying to calculate the absorption spectrum with B3LYP, the error occurrs:

"derivatives bigger than 2 not implemented, xc/xc_xpbe_hole_t_c_lr.F:368". This message will not show up when the PBE functional was used.

Is that error also correlated with the "excited state gradients" , which is not compatible with hybrid functionals?

The input is attached below, and I applied ADMM to accelate the calculation speed:

Thank you again!

&GLOBAL
PREFERRED_FFT_LIBRARY FFTW3
PREFERRED_DIAG_LIBRARY SL
PROJECT example

RUN_TYPE ENERGY

PRINT_LEVEL MEDIUM
&END GLOBAL

&FORCE_EVAL
METHOD Quickstep

&PROPERTIES
&TDDFPT

NSTATES 3

MAX_ITER 50
CONVERGENCE 1.0e-5

&MGRID
CUTOFF 300
&END MGRID

&END TDDFPT
&END PROPERTIES

&DFT
BASIS_SET_FILE_NAME GTH_BASIS_SETS

BASIS_SET_FILE_NAME ./BASIS_ADMM

POTENTIAL_FILE_NAME POTENTIAL
LSD
CHARGE 0

MULTIPLICITY 3

&MGRID
CUTOFF 300
&END MGRID

&QS
METHOD GPW
EPS_DEFAULT 1.0E-12
EXTRAPOLATION ASPC
EXTRAPOLATION_ORDER 3
EPS_PGF_ORB 1.0E-20
&END QS

! use ADMM
&AUXILIARY_DENSITY_MATRIX_METHOD
METHOD BASIS_PROJECTION
ADMM_PURIFICATION_METHOD NONE
&END

&SCF
MAX_SCF 50
SCF_GUESS RESTART
EPS_SCF 1.0E-6
&OT
MINIMIZER DIIS
Rotation
&END OT
&OUTER_SCF
MAX_SCF 10
EPS_SCF 1.0E-6
&END OUTER_SCF
&END SCF

&XC
&XC_FUNCTIONAL

&LYP
SCALE_C 0.81
&END
&BECKE88
SCALE_X 0.72
&END
&VWN
SCALE_C 0.19
FUNCTIONAL_TYPE VWN3
&END
&XALPHA
SCALE_X 0.08
&END
&END XC_FUNCTIONAL

&HF
FRACTION 0.2
&SCREENING
EPS_SCHWARZ 1.0E-6
SCREEN_ON_INITIAL_P TRUE
&END SCREENING
&INTERACTION_POTENTIAL
POTENTIAL_TYPE TRUNCATED
CUTOFF_RADIUS 6.0
T_C_G_DATA ./t_c_g.dat
&END INTERACTION_POTENTIAL
&END HF

&XC_GRID
XC_DERIV SPLINE2_SMOOTH # this is needed for the 2nd derivatives of the XC functional
&END XC_GRID

&VDW_POTENTIAL
DISPERSION_FUNCTIONAL PAIR_POTENTIAL
&PAIR_POTENTIAL
TYPE DFTD3
CALCULATE_C9_TERM .TRUE.
REFERENCE_C9_TERM .TRUE.
LONG_RANGE_CORRECTION .TRUE.
VERBOSE_OUTPUT .FALSE.

REFERENCE_FUNCTIONAL B3LYP

PARAMETER_FILE_NAME dftd3.dat
R_CUTOFF 9
EPS_CN 1.0E-6
&END PAIR_POTENTIAL
&END VDW_POTENTIAL
&END XC

....

&KIND C
BASIS_SET TZV2P-GTH
POTENTIAL GTH-BLYP-q4
BASIS_SET AUX_FIT cFIT3

&END KIND
&KIND H
BASIS_SET TZV2P-GTH

POTENTIAL GTH-BLYP-q1
BASIS_SET AUX_FIT cFIT3

&END KIND
&KIND O
BASIS_SET TZV2P-GTH

POTENTIAL GTH-BLYP-q6
BASIS_SET AUX_FIT cFIT3

&END KIND
&KIND N
BASIS_SET TZV2P-GTH

POTENTIAL GTH-BLYP-q5
BASIS_SET AUX_FIT cFIT3

&END KIND

&END SUBSYS

&END FORCE_EVAL

Matt W <mattwa...@gmail.com> 于2020年9月20日周日下午10:51写道：

--
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/SCHS1-VcFx4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/cc3c75b3-52f5-49d7-becd-f8d9ad3349f5n%40googlegroups.com.

Matt W

unread,

Sep 21, 2020, 5:08:21 AM9/21/20

to cp2k

Hello,

no this is not related. There seems to me to be a problem introduced recently which causes problems with the truncated exact exchange operator and ADMM.

Matt

HB H

unread,

Sep 21, 2020, 7:17:49 AM9/21/20

to cp...@googlegroups.com

Hi Matt,

Thx for your reply, and you are very kind~

Maybe I will try previous version of CP2K😋.

Cheers, have a nice day!

Haibei

Matt W <mattwa...@gmail.com> 于2020年9月21日周一下午5:08写道：

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/f5634f78-22a6-48d7-9f8a-28af0bb3b260n%40googlegroups.com.

Matt W

unread,

Sep 21, 2020, 11:05:50 AM9/21/20

to cp2k

As I posted in the other thread:

As a workaround for now if you add

EXCH_CORRECTION_FUNC PBEX

into the ADMM section the calculation runs (at least in the developer version).

Matt

HB H

unread,

Sep 21, 2020, 9:18:17 PM9/21/20

to cp...@googlegroups.com

Hi Matt,

Thx for the update. I have tested all of EXCH_CORRECTION_FUNC keywords, and only BECKE88X and PBEX work well. The rest keywords (i.e. DEFAULT, NONE and OPTX) are all crashed due to the same error massage.

Meanwhile, in my case where B3LYP was applied in TDDFPT, the PBEX gave me a very bad excitation energy.

TDDFPT WAVEFUNCTION OPTIMIZATION

Step Time Convergence Conv. states
--------------------------------------------------

State Exc. energy (eV) Convergence (eV)
1 -3517.6249333 -3.5214E+03
2 -41.0249219 -4.5343E+01
3 -28.5182648 -3.2893E+01

1 131.6 1.2941E+02 0

State Exc. energy (eV) Convergence (eV)
1 ************** -1.8816E+06
2 ************** -1.7503E+06
3 ************** -1.3091E+06

.......

Therefore we don't have too many choices, and must be very careful when chosen EXCH_CORRECTION_FUNC keywords.

Hopefully this can be fixed soon~

Thank you again, Matt!

Haibei

Matt W <mattwa...@gmail.com> 于2020年9月21日周一下午11:05写道：

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/b871e638-cc25-4c21-b185-603f9c80174bn%40googlegroups.com.

Reply all

Reply to author

Forward