In CASTEP, can you start with a converged density for a lower level of theory for a starting guess for HSE06? If you can, what functional is suggested or is it something that will cause more headache than it is worth?

If you know for another code, feel free to leave your experience as well.

  • $\begingroup$ I have worked mainly with VASP and QE but I feel the implementation should be similar - mix a certain amount of local/semi-local functional with a certain amount of exact exchange (as in Hartree Fock). Regarding what functional you could start with, one thing you could try is to start with PBE/LDA (whichever shows better agreement with literature) - You could look at bandstructures for example (omitting the bandgap). $\endgroup$
    – Xivi76
    Nov 3, 2020 at 20:04

1 Answer 1


Yes, you can do this. First, do a normal PBE calculation (HSE is built on-top of PBE) which generates an <seed>.check and <seed>.castep_bin file.

Then add this line

reuse : <name of the check/castep_bin file>

to the param file of your HSE06 calculation. This will let CASTEP read in and reuse the previous charge density and wavefunction for the new calculation.

Starting from an existing charge density and wavefunction will usually reduce the number of SCF cycles needed to reach convergence.

As a side note, you can also do a HSE06 bandstructure on top of a PBE charge density in one single calculation with:

task : spectral
spectral_task : bandstructure
xc_functional : pbe
spectral_xc_functional : hse06

This will make the program first solve the eletronic structure with PBE and use the obtained charge density for the subsequent HSE06 bandstructure calculation.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .