DOS calculation - which ALGO
Posted: Thu Mar 31, 2016 12:41 pm
Hi all.
I have a probably naive question, but I am really puzzled and I cannot find a clear (at least for me) answer neither in the forum neither in the vasp documentation.
Problem: calculation of the Density of States and the selection of the proper algorithm (ALGO flag).
Commonly after a Self Consistent calculations (converged over the k-point mesh) (ALGO=N or Conjugate) I calculate the DOS within a NON Self Consistent calculation increasing the K point sampling and number of empty bands imposing the code read the pre-converged charge density (ICHARG=11).
From the vasp documentation for a DOS run ICHARG=11 is set and ALGO is not specified so by default it takes ALGO = N and NELMIN = 2.
Thus the calculation will do at least a 2 step SCF run, by reading the pre-converged charge density and recalculating the occupations and the eigenvalues for the new mesh of k points imposed.
To me this procedure is NOT equivalent to a NON-Self Consistent calculation normally used to calculate the DOS ....am I wrong?
Normally I just take the DOS calculated after a SCF calculation with a dense k-point mesh..but in the present case the size of my system does not me allow this trick.
Should I better use for example another ALGO like ALGO= Eigenval?
At the moment I am doing some cross-check runs, but if someone has the answer ready I will really appreciate a clarification.
Thanks a lot.
Federico Iori
I have a probably naive question, but I am really puzzled and I cannot find a clear (at least for me) answer neither in the forum neither in the vasp documentation.
Problem: calculation of the Density of States and the selection of the proper algorithm (ALGO flag).
Commonly after a Self Consistent calculations (converged over the k-point mesh) (ALGO=N or Conjugate) I calculate the DOS within a NON Self Consistent calculation increasing the K point sampling and number of empty bands imposing the code read the pre-converged charge density (ICHARG=11).
From the vasp documentation for a DOS run ICHARG=11 is set and ALGO is not specified so by default it takes ALGO = N and NELMIN = 2.
Thus the calculation will do at least a 2 step SCF run, by reading the pre-converged charge density and recalculating the occupations and the eigenvalues for the new mesh of k points imposed.
To me this procedure is NOT equivalent to a NON-Self Consistent calculation normally used to calculate the DOS ....am I wrong?
Normally I just take the DOS calculated after a SCF calculation with a dense k-point mesh..but in the present case the size of my system does not me allow this trick.
Should I better use for example another ALGO like ALGO= Eigenval?
At the moment I am doing some cross-check runs, but if someone has the answer ready I will really appreciate a clarification.
Thanks a lot.
Federico Iori