Page 1 of 1

How is abinit calculating integrated electronic DOS ?

Posted: Sat Nov 12, 2011 10:32 am
by hicpalm
Hi,

I am trying to figure out how is abinit calculating the integrated density of electronic states.

Here is a portion of DOS file obtained with the tetrahedron method with a step of 0.0005 Ha:

Code: Select all

 #      energy(Ha)       DOS        integrated DOS       

        -1.3445     0.00161287              0       |     0.00000081
         -1.344     0.00180538              0       |     0.00000171
        -1.3435     0.00198086              0       |     0.0000027
         -1.343     0.00213036              0       |     0.00000376
        -1.3425     0.00224577              0       |     0.00000489
         -1.342   502.36035661     0.10684559       |     0.25118507
        -1.3415 1,947.00799737     0.60955971       |     1.22468906
         -1.341     0.00233338              2       |     1.22469023
        -1.3405     0.00227067              2       |     1.22469137
          -1.34      0.0021659              2       |     1.22469245
        -1.3395     0.00202504              2       |     1.22469346


The 4th column is the results of the simple way of integrating the DOS : at a given step the new integrated
density of states is the previous one plus the the new density times the energy step. if the energy step
is small enough like the one we are using, this should be a very good approximation. However, at the end,
we can see that we are far from the value of 2 displayed by abinit.

thanks for suggestions !

Re: How is abinit calculating integrated electronic DOS ?

Posted: Mon Nov 21, 2011 8:57 pm
by micah.prange
I think the problem is arising because your integrand (the DOS) is too rapidly varying to be integrated numerically on your grid. Try using abinit to output the broadened DOS (choose broadening parameter an order of magnitude larger than the output grid spacing) and I bet you will find good agreement between your own calculations and abinint's. I am not familiar with the details of abinit, but I'm guessing that the integrated DOS is calculated first and that the DOS itself is then found by differentiation.

Best,
Micah Prange

Re: How is abinit calculating integrated electronic DOS ?

Posted: Tue Nov 22, 2011 1:37 pm
by hicpalm
thank you,
I will try it.
regards.