User Tools

Site Tools


howto:xas_tdp

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
howto:xas_tdp [2021/08/02 15:40]
abussy [FAQ]
howto:xas_tdp [2021/08/16 14:48] (current)
abussy [Tetrahedral NaAlO$_2$ (K-edge, periodic)]
Line 318: Line 318:
 </code> </code>
  
-There are many performance oriented keywords and subsection in the above input. Most importantly, only one atom is treated at the all-electron level (the one atom from which the excitation takes place), all other are described using pseudopotentials. Also quite important is the usage of the ADMM method. This allows for very efficient evaluation of the HFX energy in the ground state calculation. Finally, the OT iterative solver is used. Since only a handful of eigenvalues are calculated (those within 20.0 eV of the first excitation energy), this scales much better than a full digonalization.+There are many performance oriented keywords and subsection in the above input. Most importantly, only one atom is treated at the all-electron level (the one atom from which the excitation takes place), all other are described using pseudopotentials. Also quite important is the usage of the ADMM method. This allows for very efficient evaluation of the HFX energy in the ground state calculation. Finally, the OT iterative solver is used. Since only a handful of eigenvalues are calculated (those within 20.0 eV of the first excitation energy), this scales much better than a full digonalization. Note that the ''RI_REGION'' keyword is absent (it is set to 0 by default). Since the neighbors of the excited Al atom are described with pseudopotentials, there is no need for extra RI basis function for the projection of the density.
  
 This input file would generate a spectrum such as the one visible on figure 4 of the [[ https://pubs.rsc.org/en/Content/ArticleLanding/2021/CP/D0CP06164F#!divAbstract | reference work]]. This is a much larger calculation than the first example though and would require a few hours on 20-30 processors (mostly to converge the SCF). In you are interested in reproducing this result, input, geometry and pcseg-2/admm-2 basis sets are available {{ :howto:sodal.zip | here}}.  This input file would generate a spectrum such as the one visible on figure 4 of the [[ https://pubs.rsc.org/en/Content/ArticleLanding/2021/CP/D0CP06164F#!divAbstract | reference work]]. This is a much larger calculation than the first example though and would require a few hours on 20-30 processors (mostly to converge the SCF). In you are interested in reproducing this result, input, geometry and pcseg-2/admm-2 basis sets are available {{ :howto:sodal.zip | here}}. 
howto/xas_tdp.txt ยท Last modified: 2021/08/16 14:48 by abussy