Author Topic: using RI with hybrid functionals  (Read 5830 times)


  • Jr. Member
  • **
  • Posts: 14
  • Karma: +0/-0
using RI with hybrid functionals
« on: July 29, 2010, 12:26:47 AM »

I'm attempting to run some TDDFT calculations using the B3-LYP functional and the resolution of identity (RI) approximation.  I'm having a hard time finding clear instructions on how run a geometry optimization using RIDFT and a hybrid functional.

I've seen some literature mention using rijk - but it doesn't seem to recognize the basis set I'm trying to use (TZVP).  Any tips?



  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 487
  • Karma: +0/-0
Re: using RI with hybrid functionals
« Reply #1 on: August 06, 2010, 09:03:14 PM »

TDDFT will not run with RI-J or RI-K when hybrid functionals are used. Hybrid functionals like B3-LYP need the exact HF exchange which is by far the most time consuming part, and it also delivers more or less for free the Coulomb contribution too. So RI-J will not help much in such cases. For ground state calculations using ridft, a (so called) linear-scaling algorithm for the exact exchange - that can be used when no Coulomb (J) part is needed - is implemented and can be used together with RI-J. I do not know if and how this can be extended to TDDFT, but it is not implemented for excited states right now. RI-K is more complicated and has been developed for Post-Hartree-Fock methods and is very often used for MP2, CC2, etc. in ridft and ricc2, so it is not available for TDDFT either.