Dear Sirous,
I think that these settings do not have any effect on the RPA energies from the RPATDDFT program at all. They are just printed because the program calls a subroutine from the (iterative) coupled cluster code where the values of THRDEN and THRVAR matter and are being printed by default. However, in case of the RPATDDFT program it uses the adiabatic fluctuation dissipation theorem to calculate the correlation energies. This doesn't require an iterative calculation but an integration over the coupling strength. In order to get numerically better result one then needs more abscissa points for the integration, which you can do by using, e.g:
{rpatddft; orb,7000.2;
INTAC,NLAMBDA=10
ecorr,DRPAI-AC;
}
so that now you are using 10 quadrature points. The default set in the program is NLAMBDA=7. I think that the default of 7 points should be very accurate already for all but the most difficult (multi reference) cases.
Best wishes,
Andreas