Poisson solver

[89aj]

Dear QW and users,

Is it possible to use the poisson solver object separately from the calculator object?

I would like to solve the poisson equation (with Neumann bc) not with the electron density on the rhs but with volume current density which I have obtained from another calculation (GPAW). I wonder if this is possible with ATK?

Thank you,
Anders

[Jess Wellendorff]

Dear Anders,
that is an interesting idea, but it would be a very difficult task. There are lots of Poisson solvers on the market "outhere", I suggest you have a look at e.g. the DOLFIN project: https://fenicsproject.org/olddocs/dolfin/2016.1.0/python/index.html
Regards,
Jess

[bina]

Dear ATK-developers,

I want to know the effect of poisson solver and periodic condition, so i did the several calculation of MoS2 monolayer.
MoS2 monolayer created by removing one unit of MoS2 from bulk and optimized it. here, i have considered periodicity in A and B direction and 25A vaccum created in C direction.
computational details are
K points (10x10x1)
GGA-BPE
LCAO-DZDP and remaining by defaults parameters considered.
total energy of MoS2 (unit cell) monolayer are
E = -1176.97 (FFT2D with dirichlet BC)
E = -1177.17   (multigrid with neumann)
E= -1176.98  (FFT with periodic BC)

I read from the previous post that milti-grid  used for device application.
So, Please explain me in details about the poisson solver and boundary condition.
what poisson solver and BC should use for 2D monolayer.
 

[Petr Khomyakov]

So, Please explain me in details about the poisson solver and boundary condition.
what poisson solver and BC should use for 2D monolayer.

There is no short answer to this. The best way to learn about it to read textbooks, or at least google this information.

Deciding on the choice of boundary conditions is fully up to you. Whatever boundary conditions are adopted, you have to investigate whether your results are converged with respect to the vacuum thickness. 

Note that using multigrid is more computationally demanding, but could be a way to get more accurate results, since a truly 2D system is not periodic in the out-of-plane direction.

[Jess Wellendorff]

You might also have a look at this tutorial where some aspects of the choice of boundary conditions in a particular device calculation is discussed: https://docs.quantumwise.com/tutorials/graphene_nanoribbon_device/graphene_nanoribbon_device.html