CODEWITHSUNDEEP
physicsmodelingmodelica
Joel Berger
Joel Berger

Reputation: 20280

Modelica - how to implement a constructor for a record

What is the best way to implement a constructor for a record? It seems like a function should be able to return a record object in the instantiation of the record in some later model higher up the tree, but I can't get that to work. For now I just use a bunch of parameters at the top of the record that populate the variables stored in the record, but it seems like that will only work in simple cases.

Can anyone shed a little light? Perhaps I shouldn't be using a record but a model. Also does anyone know how the PDE functionality is coming? The book only says that it is coming, but I have seen some other things around.

Upvotes: 3

Views: 791

Answers (2)

saroele
saroele

Reputation: 10109

I can only comment on your question regarding the book of Peter Fritzson. He confirmed that he's working on an update and he hopes to get it ready 'in the course of 2011'.

Original post here:

http://openmodelica.org/index.php/forum/topic?id=50

And thanks for initiating the modelica tag, I might be useful in the near future for me too... :-)

regards, Roel

Upvotes: 1

Michael Tiller
Michael Tiller

Reputation: 9421

I don't seem to have the clout to add tags (which makes sense, since my "reputation" is lower than yours) so sorry about that. I thought I had actually added one at one point, but perhaps I am mistaken.

I think you need to be clear what you mean by constructor since it has a very specific meaning in Modelica. If I understand your question correctly, it sounds like what you want to do is create an instance of a record that has some fields that are specified in the constructor arguments and from those arguments a bunch of other fields in the record are computed. Is that correct?

If so, there is a mechanism to do this. You mention "the book" but it isn't clear which one you mean. If it is mine, it definitely has no mention of these so called "record constructors" because it is too old. I do not know if Peter Fritzson's book mentions them either. However, they do exist and are documented in Section 12.6 of the Modelica 3.2 specification.

As for PDEs, there has been work into this kind of thing but nothing has really been done within the design group on this topic. I would add that if you want to solve either elliptical or parabolic PDEs on regular grids, this isn't too hard even with the current language. The only real drawback is that most tools probably don't handle sparsity very efficiently. Irregular grids would also be possible, but then you get into complicated basis functions. Finally, hyperbolic PDEs are, in my opinion, quite tricky (in any environment) due to the implicit physical constraints between time and space which are difficult to express (i.e. the CFL condition).

I hope that answers your questions so far.

Upvotes: 3

Related Questions