How do you add a new solver?

[johnjaypl]

So I'm trying the next tutorial. Eigenmodes of an elastic plate.

The tutorial says to do:

Before we can start the set-up we shoud make sure that the menus for Smitc solver are present. If not,
they may be found in file
$ELMERHOME/bin/edf-extra/elasticplate.hml
To load these definitions do the following
File
Definitions
Append -> choose the file
To see what kind of new menu structures you got you may play around with viewer collapsing and opening.

I more of less do that (xml instead of hml)
I don't see any new menu structures to play with. When I follow further down in the tutorial to add an elastic plate equation none is to be found. Also, I didn't see any hml in the bin, it was xml.

I don't know what to make of this note either.

Note that if you want to load an existing project you should load the xml-definitions that were used in
creating the project. Therefore it may be best to place all actively used menu definitions in directory
$ELMERHOME/bin/edf

[raback]

Hi

Thanx for the typo fixes. This tutorial has been used in many courses so its difficult to know why it fails for you. Alternatively you can move the elasticplate.xml from edf-extra directory to edf. This was it will be automatically loaded at the time of the launch. This works the other way around as well i.e. you can remove solvers from the menu structure by moving the corresponding xml file from the edf directory.

It may also be noted that many of the less frequently used solvers do not have a xml definition file at all. Adding them (or the user's own menu structure) is, however, not a major undertaking. It hasn't just been done for many of the solvers.

-Peter

[Hanan]

Hi
How to add the Schrödinger equation to the Elmer FEM program?

[kevinarden]

You write the FORTRAN 90 code and compile it with elmerf90 compiler

https://www.nic.funet.fi/pub/sci/physic ... torial.pdf

[raback]

Hi,

There is a version for density functional theory, DFTSolver.F90, among the code and apparently H20 molecule in test case dft_water. When this was done lo...ong ago FEM had not been applied too much for DFT. No the situation is quite different. There are a lot of code doing quantum mechanics also using FEM. This is derived from Schrödinger's equation with some additional assumptions.

To my understanding choosing low order polynomials for orbitals is not as good a solution that it is generally in engineering. The reason is that all solutions are smooth and have a large "support" and the mesh does not need to present a shape. So many codes use precomputed orbitals (e.g. Gromacs) as the basis. If local support is used often people choose FD (e.g. gPaw) over FE because you don't really need a shape.

Generally we like to see many different physical problems in Elmer bacause this enables multiphysics simulation which lies in the essense of the code. However, to be honest, the quantum mechanical issues are very rarely coupled with the continuum equations. Also the DFT groups are really dedicated to their codes so it is impossible for a generic code to challenge them.

-Peter