On Mon, 23 Jul 2012 18:31:50 +0200, sylvian kahane <sylviankahane at gmail.com> wrote:
> > When I am looking in the literature Jarlborg (Physica Scripta 37, 795 > (1988)) gives -14.84 Ry/atom and this is what I got in my pw calculation > -800/54=-14.81, but Dacorogna and Cohen (Phys. Rev. B 34, 4996 (1986) > gave > -0.617 Ry/atom, which is not close but resembles my result from cp.x > -47*2/54=-1.74 Ry/atom. This is because the Jarlborg calculations appear to be all electron so match up (roughly) with the PAW calculation and the Cohen calculations use a pseudopotential with only one valence electron (so match up roughly with your pseudopotential calculation). As has been pointed out though total energies produced from pseudopotential calculations are meaningless, you have to look at energies differences between calculations using the same pseudopotentials. Unless you are using the same pseudo as Cohen et al. you shouldn't be looking to reproduce their total energies. > > Why I am bothered by these values ? I got involved in some EOS > calculations > and I am trying to solve what is called as first shock > Hugoniot equation (which takes you from an initial state P1, V1, U1 > (pressure, volume, internal energy) to a final state P2, T2, U2 by a > shock): > > (U2 - U1) = 1/2*(P1+P2)*(V2-V1) if > I remember correctly > > in which you have to put the initial values. For U I use > > U=3/2kT + Etot/N (N=54) > > following some papers by one Lenosky (I don't have references right > here). > So, if in this equation you use -1.74 Ry or -14.8 Ry it makes a hell of a > difference. > This just means that the Etot1 and Etot2 (for use in U1 and U2) need to be calculated using the same pseudopotentials, then assuming that the pseudopotentials are a reasonable approximation for your system, U1 - U2 should be the same across all different sets of pseudos. Best regards, Simon -- Simon Binnie | Post Doc, Condensed Matter Sector Scuola Internazionale di Studi Avanzati (SISSA) Via Bonomea 256 | 34100 Trieste | sbinnie at sissa.it
