Dear TURBOMOLE support, question about the converged energy,thank you very much!

[ComingNine]

Dear TURBOMOLE support, question about the converged energy,thank you very much!

Dear TURBOMOLE support,

   Please have a look at the output of one ridft calculation of C2H5-OH

Code: [Select]

Starting SCF iterations

          Overall gridpoints after grid construction =          3901   

 ITERATION  ENERGY          1e-ENERGY        2e-ENERGY     NORM[dD(SAO)]  TOL
   1  -115.62050121264    -236.25587770     80.640718764    0.000D+00 0.877D-09
                            Exc = -15.4761720067     Coul =  96.1168907710   
                              N = 17.999727430   
                            current damping = 0.700
 
          max. resid. norm for Fia-block=  2.497D-05 for orbital      5a'       
          max. resid. fock norm         =  8.537D-05 for orbital     21a'       

 ITERATION  ENERGY          1e-ENERGY        2e-ENERGY     NORM[dD(SAO)]  TOL
   2  -115.62050121553    -236.25604933     80.640890393    0.835D-04 0.556D-09
                            Exc = -15.4761923167     Coul =  96.1170827092   
                              N = 17.999727416   
                            current damping = 0.700
 
          Norm of current diis error: 0.53008E-04
          max. resid. norm for Fia-block=  1.111D-05 for orbital      6a'       
          max. resid. fock norm         =  3.344D-05 for orbital     21a'       

ENERGY CONVERGED !

          Overall gridpoints after grid construction =         14447   

 ITERATION  ENERGY          1e-ENERGY        2e-ENERGY     NORM[dD(SAO)]  TOL
   3  -115.62066272068    -236.25591853     80.640598086    0.399D-04 0.461D-09
                            Exc = -15.4763379432     Coul =  96.1169360292   
                              N = 17.999999749   
                            current damping = 0.750
 
          Norm of current diis error: 0.33078E-04
          max. resid. norm for Fia-block=  4.671D-06 for orbital      5a'       
          max. resid. fock norm         =  1.461D-05 for orbital     21a'       

 End of SCF iterations
   convergence criteria satisfied after    3 iterations

...
                  ------------------------------------------
                 |  total energy      =   -115.62066272068  |
                  ------------------------------------------
                 :  kinetic energy    =    114.60891611891  :
                 :  potential energy  =   -230.22957883958  :
                 :  virial theorem    =      1.99124943087  :
                 :  wavefunction norm =      1.00000000000  :
                  ..........................................

   I am wondering, after the line "ENERGY CONVERGED", why ridft will calculate again, and why there is a large difference between the values below and above the line "ENERGY CONVERGED"?

   I mean, the "total energy      = " value is not the actually converged values. Could you help to explain a little to me?

   Thank you very much for your time and effort!

Best wishes,

Xichen

[uwe]

Hi,

please note that this forum is NOT the Turbomole support. To get support for Turbomole please just send an email to turbomole (at) cosmologic (dot) de

This user forum was thought to be a platform where users can share their (Turbomole related) experiences and thoughts.

To answer your question anyway: This is due to the gridsize 'm3' in the $dft keyword section. What m3 does is to use a small grid for the DFT quadrature (i.e. the numerical integration of the density functional) during the SCF iterations. After convergence, ridft will switch to grid size 3 and do a single iteration to determine the final energy.

This is the default setting in Turbomole since it has been shown that the grid size does not have a large influence on the density but on the total energy only. So SCF converges to the same density with the small grid than it would for he larger grid. For structures and relative energies the final step for the total energy is sufficient (although you are of course right, if you look a the energies in the output it looks strange...).

For higher order properties or TDDFT it is recommended to use gridsize 4 - without an 'm' to be sure not to get too large errors from the quadrature. This is mentioned several times in the documentation and very shortly in the tutorial.

Regards,

Uwe

[ComingNine]

Dear Uwe,

   Thank you very much for your excellent help!

Best wishes,

Xichen