CCL: ORCA

 Sent to CCL by: "Raji  Raji" [raji=anal.chem.tohoku.ac.jp]
 Dear CCL Members,
      Few hours ago I have posted a message regarding my doubts about ORCA
 calculations. I have received such a nice explanations from Dr. Frank (ORCA
 developing team). Thank you very much Dr. Frank and ORCA developing team. Here I
 would like to summarise those results for your future reference. In addition
 with orca_plot utility programme, I found that the orca_2mkl utility programme
 works well to plot the MOs using Molekel.
 Regards,
 Raji.
 Dear Raj,
     1) For TDDFT/TDA calculations with B3LYP/6-311G, in the COSMO part, I have
 to mention both solvent dielectric constant and refractive index? or just
 dielectric constant is enough?
 .. you have to provide the refractive index as well in CIS/TD-DFT; this is for
 the program to figure out the "fast" and "slow" solvent
 response.
 Its given in the manual that the refractive index is used only for CIS or MRCI
 modules.
 .. that is true; the MRCI module uses a somewhat approximate perturbation theory
 which needs more testing before i generally recommend it.
 Since in ORCA, CIS keyword is applicable for both CIS and TDDFT calculations, I
 got confused. So the refractive index is also necessary for TDDFT/TDA with
 hybrid functionals? By the way I am using RI approximation too.
 .. both CIS and TD-DFT/TDA pass through the same code. If you have a DFT SCF
 calculation you will get TD-DFT/TDA, if you have a Hartree-Fock ground state SCF
 calculation you'll get a CIS excited state calculation. So, no worries necessary
 - the program decides for you!
 2) When I use RI approximation (RIJONX for the same hybrid function and basis
 set), it will only be applied to the Coulomb term;
 .. this is a sensible question. In RIJONX mode with hybrid functionals the RI
 approximation is only applied to the Coulomb term during the SCF iterations
 (this still saves a lot of time for larger molecules since the exchange is
 nearly linear scaling with molecule size in contrast to the Coulomb term). In
 the TD-DFT calculations RI is applied to both - the Coulomb and the exchange
 term. In our experience, the errors that you obtain in TD-DFT via the RI
 approximation are truly minimal even with the standard Coulomb fitting basis
 sets (for standard basis sets SV/C is also ok; it is only when you have
 additional diffuse functions in the basis set that you really should also
 augment the fitting basis). Technically speaking one should probably have
 different aux sets for the Coulomb and exchange parts but numerical evidence did
 so far not support that this is necessary. You can also use automatically
 generated auxiliary basis sets (AutoAux keyword), which are a little larger than
 the incredibly nice standard fit sets of the Ahlrichs group but AutoAux
 generates "general purpose" fit sets. Of course, as always in science
 - if you are in doubt of your results, you have to test the accuracy of the
 calculations by running a representative calculation with and without RI!
 3) When I do population analysis, I want to analyse the results using molekel.
 so in this case I have to use
 ! RKS B3LYP 6-311G OPT RIJONX SV/C XYZFile
 %plots Format Cube
       MO("abc.cube",4,0) ;
 end
 this means orca will generate a cube file (abc.cube) only for the MO 4 of the
 closed shell RKS operator? Lets say 37th orbital is HOMO for my system, and if I
 am interested from HOMO-4 to LUMO+4 then I have to mention the following :
     MO("33.cube",33,0) ;
     MO("34.cube",34,0) ;
     ......................
     MO("42.cube",42,0) ;
 .. yes this is all correct if you want to obtain these files via the ORCA input.
 Of course you can use the simple way of generating the cube files interactively
 by running orca_plot MyJob.gbw -i then you will get a "stone age" menu
 which lets you choose the ouput format, number of grid points and MOs to plot.
 This is how i usually do it and once you are familiar with orca_plot it is a
 really fast way from a calculation to a pretty picture.
 Hope that helps,
 have fun with ORCA!
 Frank