ABINIT Discussion Forums

Dear ABINIT friends,
I want to calculate the partial DOS of AlP compound. I am trying to make these calculations with t38.in input file "https://docs.abinit.org/tests/v4/Input/t38.in". I have some questions about the values of natsph, iatsph and ratsph input variables. What changes should be made in the values of these variables for this compound with two atoms in the primitive unit cell?
Best wishes,
Domakani

Hello Domakani!

There are a few changes that are needed to the input file. The one you attached is for a crystalline solid (Silicon in fact).

To change this for AlP you would need to modify several of the input variables. Some of the variables that you need to modify are:
ntypat, znucl, natom, xred, typat.

Additionally, you will need to determine the appropriate ecut and kgrid to use for this calculation by running a convergence test.

Information about the abinit input variables can be found here:
https://docs.abinit.org/variables/#A

Additional information on convergence tests can be found in the tutorial (see "four basic tutorials"):
https://docs.abinit.org/tutorial/

For the variables you asked about, you can find information about them here:
natsph -> https://docs.abinit.org/variables/gstate/#natsph
iatsph -> https://docs.abinit.org/variables/gstate/#iatsph
ratsph -> https://docs.abinit.org/variables/gstate/#ratsph

Good luck and have a great day!

Dear NPike,
Thank you for your response. I asked about AlP to have an example. I really feel that I need this example to have a better understanding about this problem.
Thank you very much.
Best,
Domakani

Hello Domakani,

I understand your point. I also like to have an example of my own when I am learning about a new code or method.

As I mentioned in my post, some of the variables you will need to modify are ntypat, znucl, natom, xred, typat, etc. You can find more information about each of these input variables in the Abinit website (click variables at the top).

Of course, to properly calculate the DOS of any compound, you need to first have relaxed lattice parameters and run a few convergence tests. (See the Abinit tutorials for information about convergence tests and structural relaxation calculations).

-Nick

Hello Nick,
Thank you for your reply,
In fact, I know any necessary things about ntypat, znucl, natom, xred and typat for this compound. I also can do necessary convergence studies on k points, ecut, and lattice parameters. I could not determine the natsph, iatsph and ratsph input variables for this compound.
Best,
Domakani

Hello,

natsph is a number corresponding to the number of atoms that you want an atomic projected density of states for. I would personally just do all atoms in your unit cell.

iatsph is the index of the atoms you want in the atomic projection of the density of states for.

ratsph is automatic if you use PAW pseudopotentials, if you do not, you need to test a few different values.

For example. if there are 4 atoms in your system and you use PAW pseudos then I would add the following to my input file.
natsph 4
iatsph 1 2 3 4

Hello Nick,
Thank you for your response,
I work with norm-conserving pseudopotentials. So, for ratsph what values should be considered. In fact, how is this variable determined for different systems?
Domakani
Best,
Domakani

Hello domakani,

I suggest you look at https://docs.abinit.org/variables/gstate/#ratsph which provides two physically motivated ways to determine this variable.

-Nick

Dear Nick,

My problem is the same I do not know how to get Radii of the ATomic SPHere(s) in order to compute DOS. As I know, in a different run we have to calculate ratsph then using them in the run of DOS. Is this correct please inform me the correct one
I use norm-conserving pseudopotentials

Regards

Domokani,

When I used these variables in a relaxation program of a slab of Pd [111], I followed the description in the variables section of abinit.org (see earlier answers). The method suggested there and I used was to take the radii < (interatomic distance)/2.

Eliezer