Bandstructure and DOS

Q. How can I quickly generate a density-of-states plot.

A Use the program which is shipped with academic source code releases of CASTEP. This somply reads the <seed>.bands file and can directly invoke either xmgrace or gnuplot to make a plot. It has a number of refinements, including a variety of ways to plot a spin-polarized DOS, but uses a Gaussian broadening scheme which smears any sharp features, and can not treat partial densities of states. Usage can be as simple as " -xg <seed>.bands", but see " --help" for more options.

Q. How can I plot partial densities of states

A. Using the OPTADOS postprocessing code. A copy of this is bundled with academic releases of CASTEP and may be installed on some managed HPC systems.

Q. How can I generate an electronic density of states with the sharp features I expect?

A. Again, the answer is OPTADOS, which uses an adaptive broadening scheme.

Q. Why does my OPTADOS density of states plot not look like the one using, even for fixed Gaussian smearing?

A. OPTADOS and take different inputs for their smearing. For OPTADOS, the fixed_smearing parameter is the standard deviation of the Gaussian; for the width specified by -w <width> is the full-width half-maximum (FWHM), which assumes is 2.354 times the standard deviation for a Gaussian (strictly, it should be 2*sqrt(2*ln(2)) times).

Q. How can I generate a bandstructure plot?

A. Use the "" script shipped with CASTEP. As with "" usage can be as simple as " -xg <seed>.bands".

Q. How can I generate a publication quality electronic band structure with the bands properly joined up?

A. Using the orbitals2bands postprocessing utility which is part of the academic source code distribution, and which ought to installed on managed HPC systems. This reads a checkpoint file, detects band crossings and and regenerates the .bands file with re-ordered states.

Q. My bandstructure looks a mess with many very short segments.

A. Did you calculate the bands for a grid of points rather than a path in reciprocal space? and can only with with the <seed>.bands file they are given, and it is your responsibility to set up the CASTEP calculation appropriately.