dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

On 29/08/11 10:43, Anders Logg wrote:
> On Mon, Aug 29, 2011 at 07:17:27AM -0700, Garth N. Wells wrote:
> 
>> I had in mind that where you have 'uint subdomain' that this should
>> be templated so that any data can be attached.
> 
> I thought of it but couldn't think of any good use for it 

Is that a good reason? ;)

> so I decided
> to go with 'uint'. Is there a compelling reason for it or can we add
> it later if we should need it?
> 

I don't think that it would be too difficult. I've used MeshFunctions
with non-primitive data types, so I imagine that it could also be useful
on subsets of entities.

Garth

>> Also, since objects could
>> be attached to only a very small subset of mesh entities, is it
>> necessary to create mesh functions? What about just a map/vector of tuples?
> 
> The conversion to MeshFunctions is practical since it means we can
> plug it directly into the assembler which relies on MeshFunctions for
> checking the subdomain number of the current cell or facet. It might
> be possible to do it the other way around, iterate over subdomains and
> assemble on each, but that requires a rewrite of the assembler.
> 
> For boundary conditions, the conversion is not necessary.
> 
> --
> Anders
> 
> 
> 
>> Garth
>>
>>
>>
>> On 29/08/11 06:27, noreply@xxxxxxxxxxxxx wrote:
>>> ------------------------------------------------------------
>>> revno: 6145
>>> committer: Anders Logg <logg@xxxxxxxxx>
>>> branch nick: work
>>> timestamp: Mon 2011-08-29 15:25:38 +0200
>>> message:
>>>   Add missing files
>>> added:
>>>   dolfin/mesh/MeshDomains.cpp
>>>   dolfin/mesh/MeshDomains.h
>>>
>>>
>>
>>
>> _______________________________________________
>> Mailing list: https://launchpad.net/~dolfin
>> Post to     : dolfin@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~dolfin
>> More help   : https://help.launchpad.net/ListHelp