[molpro-user] Freq calculation

[Hans-Joachim Werner]

I tried your calculation and the differences in the energies of the opt and sp calculations are of the order of 10^{-7} E_h. 
The reason can be slightly different convergence of the Hartree-Fock, since in the optimization it starts with the orbitals
of the previous iteration, while in your sp it starts with those from the non-density fitted HF. Furthermore, you included
the bond- and dihedral angles in your optimization, and very small deviations from the exact values (120, 0, 180) may cause
tiny energy changes. In principle, these differences should disappear with tighter convergence thresholds, but in the current
case with a diffuse basis set this is limited by the numerical accuracy. It may help to omit the diffuse functions on the hydrogen atoms.

Some other general comments:

- the gradients and all other information for each displacement can be found in the log file.

- you included a conventional HF calculation before DF-HF. This causes all two-electron integrals to be explicitly computed and stored.
  This HF calculation should be omitted.

- you included the bond- and dihedral angles in the optimization. This is avoided by either defining them as „inactive“ in the optimization
  or to put constants into the z-matrix (e.g. 0 instead of dih4). In a numerical gradient calculation this would save a lot of time.

- for very accurate optimized MP2 energies it may be necessary to tighten the CPHF convergence threshold. For example, 

  basis,avtz,h=vtz
  geometry={…}
  {df-hf;accu,15}
  {df-mp2;cphf,accu=1.d-9}
  optg,gradient=1.d-6;

 yields 
 CC2=                 1.39421141 ANGSTROM
 HC7=                 1.08238902 ANGSTROM

and the opt and sp energies agree within 1.d-8 hartree.

Best regards
Joachim Werner







> Am 17.09.2015 um 18:31 schrieb samala nagaprasad reddy <snpreddy63 at gmail.com>:
> 
> Hi
>   I have a couple of questions.
> 
> 1) I am calculating optimization and frequency calculation of benzene. After optimization when I calculate the single point energy calculation the SP energy is lower than the optimized energy. I am taking the optimized Z matrix to run the SP energy calculation. How it is possible? What I am missing? Here I am giving my input of optimization and output files of both SP and optimization calculations.
> 
> Optimization input
> 
>  ***,coupling constants
>   memory,450,m
>  cc2    =    1.394207,
>  ccc3   =    120.000,
>  dih4   =      0.000,
>  hc7    =    1.082115,
>  dih7   =    180.000
>   basis,avtz
>   geomtype = zmat
>   geometry = {angstrom,
>  c;
>  c,   1, cc2;
>  c,   2, cc2,        1, ccc3;
>  c,   3, cc2,        2, ccc3,          1, dih4;
>  c,   4, cc2,        3, ccc3,          2, dih4;
>  c,   5, cc2,        4, ccc3,          3, dih4;
>  h,   2, hc7,        3, ccc3,          4, dih7;
>  h,   3, hc7,        2, ccc3,          1, dih7;
>  h,   4, hc7,        3, ccc3,          2, dih7;
>  h,   5, hc7,        4, ccc3,         3, dih7;
>  h,   6, hc7,        5, ccc3,         4, dih7;
>  h,   1, hc7,        2, ccc3,         3, dih7;
>  }
>  {hf,ENERGY=1.0E-10,orbital=1.0E-8;}
>  {df-hf,ENERGY=1.0E-10,orbital=1.0E-8;}
>  {df-mp2,ENERGY=1.0E-10,orbital=1.0E-8;}
>  optg;
>  frequencies;
> 
> SP energy input
> 
> ! Cite this work as:
>  ***,coupling constants
>  memory,450,m
> cc2    =    1.394247,
> cc3    =    1.394247,
> ccc3   =    120.000,
> cc4    =    1.394247,
> ccc4   =    120.000,
> dih4   =      0.000,
> cc5    =    1.394247,
> ccc5   =    120.000,
> dih5   =      0.000,
> cc6    =    1.394247,
> ccc6   =    120.000,
> dih6   =      0.000,
> hc7    =    1.082198,
> hcc7   =    120.000,
> dih7   =    180.000,
> hc8    =    1.082198,
> hcc8   =    120.000,
> dih8   =    180.000,
> hc9    =    1.082198,
> hcc9   =    120.000,
> dih9   =    180.000,
> hc10   =    1.082198,
> hcc10  =    120.000,
> dih10  =    180.000,
> hc11   =    1.082198,
> hcc11  =    120.000,
> dih11  =    180.000,
> hc12   =    1.082198,
> hcc12  =    120.000,
> dih12  =    180.000
> basis,avtz
> !gprint,orbitals,civectors
>  geomtype = zmat
>  geometry = {angstrom,
> c;
> c,   1, cc2;
> c,   2, cc3,        1, ccc3;
> c,   3, cc4,        2, ccc4,          1, dih4;
> c,   4, cc5,        3, ccc5,          2, dih5;
> c,   5, cc6,        4, ccc6,          3, dih6;
> h,   3, hc7,        2, hcc7,          1, dih7;
> h,   4, hc8,        3, hcc8,          2, dih8;
> h,   5, hc9,        4, hcc9,          3, dih9;
> h,   6, hc10,       5, hcc10,         4, dih10;
> h,   1, hc11,       2, hcc11,         3, dih11;
> h,   2, hc12,       3, hcc12,         4, dih12;
> }
> {hf,ENERGY=1.0E-10,orbital=1.0E-8;}
> {df-hf,ENERGY=1.0E-10,orbital=1.0E-8;}
> {df-mp2,ENERGY=1.0E-10,orbital=1.0E-8;}
> 
> 
> 2) In the frequency (numerical) calculation, it distort the molecule to get Hessian. I want the forces at each displacement. When I check the output files, it is not wring the information. How can I get the information?
> 
> I look forward to hear from you
> 
> Thank you
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user