[CPMD-list] Output of TDDFT geometry optimisation

[Juerg Hutter]

Hi

geometry optimisation with constraints has some problems
within CPMD. Have a look at the final gradients.
If you project out the gradients along the constraint coordinates
you should get only small numbers. GNMAX was calculated
including the forces on the Constraint.

Concerning the SAOP functional. CPMD is not using the
SAOP potential for the ground state as it is not
providing an energy. In your input you are requesting
a PBE ground state calculation followed by a TDDFT
calculation using the SAOP potential for the
energy state differences (first part of LR-TDDFT eqn)
and then LDA for the kernel part. In this way a total
excited state energy is well defined (and not just an
excitation energy) and we can calculate energy gradients.

regards

Juerg Hutter

----------------------------------------------------------
Juerg Hutter                   Phone : ++41 44 635 4491
Physical Chemistry Institute   FAX   : ++41 44 635 6838
University of Zurich           E-mail: hutter at pci.unizh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------


On Wed, 15 Nov 2006, Latévi Max LAWSON DAKU wrote:

> Dear CPMD users and developers,
>
> Using the CPMD code - to which I am quite new, I am trying to carry
> out a constrained geometry optimization in the first excited state,
> within TDA-TDDFT. I used the following input in an attempt I made:
>
>    &CPMD
>      RESTART WAVEFUNCTION COORDINATES LINRES
>      OPTIMIZE GEOMETRY
>      LBFGS
>      TDDFT
>      CONVERGENCE ORBITALS
>      1.0d-7
>      STORE
>      40
>      STRUCTURE BONDS ANGLES DIHEDRALS
>    &END
>
>    &TDDFT
>      STATES SINGLET
>      2
>      TAMM-DANCOFF
>      DAVIDSON PARAMETER
>      150 1.D-7 50
>      DAVIDSON RDIIS
>      100 15 1.D-3
>      FORCE STATE
>      1
>      TD_METHOD_A SAOP
>    &END
>
>    &DFT
>      FUNCTIONAL PBE
>      LR KERNEL LDA
>    &END
>
> The calculation ended normally but there are things I do not
> understand in the output.
> First, it seems that the job has stopped too early since, as this
> can seen below, GNMAX is well above the convergence criterion
> of 5e-4 :
>
>     ****************************************************************
>     *** TOTAL STEP NR.   431           GEOMETRY STEP NR.     20  ***
>     *** GNMAX=  9.191471E-02 [3.81E-06]     ETOT=    -61.317526  ***
>     *** GNORM=  1.687789E-02               DETOT=    -1.500E-07  ***
>     *** CNSTR=  8.773584E-02                TCPU=       1305.02  ***
>     ****************************************************************
>     ================================================================
>     =              END OF GEOMETRY OPTIMIZATION                    =
>     ================================================================
>
>
> Please note that the geometry really did no more significantly  evolve
> during the last steps of the calculations. Maybe I missed the fact that
> the convergence criteria are different for ground- and excited-state
> optimizations ?! By the way, what does mean the number in brackets
> that follows GNMAX ?
>
> Secondly, I am not sure about the manner in which the ground-state
> total energy is being evaluated during the calculation. In a given
> calculation step, before solving the TDDFT eigenvalue problem, the
> ground state is characterised first:
>
>       1  3.028E-07   3.690E-08     -90.367734   -2.523E-06      4.82
>       2  1.745E-07   1.342E-08     -90.367734   -8.812E-10      4.85
>       3  1.514E-07   7.063E-09     -90.367734   -7.759E-11      5.00
>       4  1.454E-07   3.517E-09     -90.367734   -2.143E-11      5.02
>       5  1.167E-07   2.133E-09     -90.367734   -7.560E-12      5.05
>       6  5.499E-08   1.536E-09     -90.367734   -3.382E-12      5.07
>     ================================================================
>     ==            DAVIDSON DIAGONALISATION OF TDDFT MATRIX        ==
>     ================================================================
>
> I expect that the SAOP potential is being used to determine the
> KS eigensystem and that the energy which is reported (here, -90.367734 au)
> is a post-SCF result obtained with the PBE functional. Is this correct ?
>
> Then, after that the TDDFT forces have been computed, one ends up
> with the following result for the energy (last calculation step), which
> I would like to understand (for instance, no X-C contribution ??):
>
>     (K+E1+L+N+X)                 TOTAL ENERGY =          -61.31752565 A.U.
>     (K)                                    KINETIC  ENERGY =
> 3.82987466 A.U.
>     (E1=A-S+R)      ELECTROSTATIC  ENERGY =          -75.96854907 A.U.
>     (S)                                                        ESELF
> =           81.11826368 A.U.
>    (R)                                                             ESR
> =            4.43642281 A.U.
>     (L)    LOCAL PSEUDOPOTENTIAL ENERGY =          -62.20759089 A.U.
>     (N)          N-L PSEUDOPOTENTIAL ENERGY =            8.97368637 A.U.
>     (X)      EXCHANGE-CORRELATION ENERGY =             .00000000 A.U.
>                   GRADIENT CORRECTION ENERGY =              .00000000 A.U.
>     (E)                EXCITATION ENERGY(TDDFT) =
> 3.074333 e.V.
>
>
> Could someone explain me what is the reported energy referring to ?
> Besides digging into the code, which papers can help me understand
> what is going on ?
>
> Thanks in advance for your answers.
>
> Best regards,
> Max
>
> -- 
> ***********************************************
> Latevi Max LAWSON DAKU
> Departement de chimie physique
> Universite de Geneve - Sciences II
> 30, quai Ernest-Ansermet
> CH-1211 Geneve 4
> Switzerland
>
> Tel: (41) 22/379 6548 ++ Fax: (41) 22/379 6103
> ***********************************************
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>