Disagreement between total energy and stress tensor in distorted unit cell

[hipertrofia]

Hi,

I am doing calculations of total energy with fixed cutoff energy for different lattice vectors. If I am correct this means the basis set is recalculated each time. What is the effect of this on the energy curve as a function of cell deformation? Am I not introducing a discontinuity in the energy due to the fact that the basis set is reoptimized each time?

The derivative of this energy, for very high energy cutoffs, coincides with the stress tensor given by VASP (which is obtained with a fixed basis set). However, for very low symmetry situations, such as all three lattice vectors scaling differently between different calculations, the results obtained with the different methods differ considerably.

I have tried many things, from increasing the cutoff to number of kpoints, NGX, etc.

I would like to do a calculation as in tutorial accuracy.pdf (slide 6) but ISTART = 2 won't work for me when the lattice parameters have changed respect to the calculation that generated WAVECAR.

Also, is there an option to manually disable the kinetic energy corrections in order to evaluate their impact on the results?

Many thanks.

[admin]

as WAVECAR stores the coefficients of the wavefunction, it is crucially important that the basis functions of the calculations reading/writing the file match perfectly. As the basis set changes upon changing the unit cell geometry, WAVECAR cannot be used in this case.
the kinetic energy cannot be disabled in the calculation, all you can do is to write a script which subtracts it from the total energy.