Energy vs Volume ... the usual problem (?)

[schani]

Hi!

I intend to compare the ideal volume for alpha-quartz extracted from an equation-of-state fit to that found by the optimizer.
I have done ion-only optimizations at several different volumes about +-9% around the experimental minimum. To get the pulay-stress out of the game, I always used the WAVECAR of the last run with ISTART=2. Interestingly, I had to use different CPU numbers between succeeding steps, because otherwise I would get a "coefficients changed"-error.

For my calculation I now get a smooth E(V)-curve, that can be very well fitted using a Murnaghan-Birch EOS, indicating that there is no problem with changing basis sets. The interesting thing is, that the calculation predicts different equilibrium volumes depending on whether I start from the lowest or highest volume (by circa 5-7%). I repeated using different settings, with or without a 4x4x4 MP-K-point-mesh, at high energy offset (800eV) and with or without real-space-projection. The result is always the same.

Does anyone have an idea what could be the problem here?

[schani]

Oops... just found the problem myself.