Dear users, I would like to know why calculations with a Gaussian basis are in general more expensive than a numerical basis like the type used in SIESTA.
Both Gaussian and numerical basis sets are atom-centered so one would expect their performance not to be too far apart. Is the SIESTA basis set faster because the radial part of the basis is stored on a log grid? Or is it because the confining potential shortens the radial extent of the basis set, thereby making it more compact? Any inputs from the experts would be appreciated. Thank you.
-- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)