[dune-pdelab] trouble with dune-pdelab trunk

[Steffen Müthing]

Hi Oleh,

The functionality of lfsu.getChild(k) is still there, it has only been renamed to
lfsu.child(k) for consistency with the majority of the DUNE interface (no get...
prefixes).

Moreover, you should never try to copy-construct a local function space,
you are only supposed to access them by reference. So, if you want to
access the i-th degree of freedom of the j-th child space, you can write

x(lfsu.child(j), i);
r.accumulate(lfsv.child(j), i, /* some value */);

If you want to create an array of the child spaces, I'm afraid you can really
only save pointers.

A PowerGFS of size 40 should work fine. If you run into performance problems
compared to a PowerGFS of something like size 10 (especially during compilation),
let me know, because that shouldn't happen.

Steffen


Am 09.08.2011 um 15:41 schrieb Oleh Krehel:

> Hi,
> 
> I have another question. This is copied from the website:
> 
> Access to vector entries:
> 
> | x[lfsu.localIndex(i)]                               // old way
> | x(lfsu,i)                                           // new way
> 
> Access to matrix entries:
> 
> | jac(lfsu.localIndex(i),lfsv.localIndex(j))          // old way
> | jac(lfsu,i,lfsv,j)                                  // new way
> 
> Accumulate residual:
> 
> | r[lfsv.localIndex(j)] += value                      // old way
> | r.accumulate(lfsv,j,value)                          // new way
> 
> In my case "lfsu" is PowerGridFunctionSpace<lfsuk,40>
> (is it reasonable in terms of efficiency to use this?).
> Can I index the child space as well as dof inside space
> (i.e. x(lfsu,k,i) means ith dof of species k)?
> Before I was using lfsu.getChild(k).
> Now it's deprecated.
> Also there's no longer a default constructor for the child space,
> and I can't declare an array of child spaces,
> just an array of pointers to them.
> What's the alternative?
> 
> best regards,
> Oleh Krehel
> 
> 
> 
> 
> 
> 
> 2011/8/9 Jö Fahlke <jorrit at jorrit.de>:
>> Am Tue,  9. Aug 2011, 14:12:37 +0200 schrieb Oleh Krehel:
>>> When switching from pdelab 2.1snapshot to trunk I got this problem
>>> when calling dunecontrol:
>>> 
>>> configure:36818: checking for dune-pdelab installation or source tree
>>> configure:36898: result: found in /localhome/krehel/DUNE2.1/dune-pdelab
>>> configure:36916: checking dune/pdelab/common/function.hh usability
>>> configure:36916: g++ -std=c++0x -c -g -O2
>>> -I/localhome/krehel/DUNE2.1/dune-common-2.1.0
>>> -I/localhome/krehel/DUNE2.1/dune-grid-2.1.0
>>> -I/localhome/krehel/DUNE2.1/dune-localfunctions-2.1.0
>>> -I/localhome/krehel/DUNE2.1/dune-istl-2.1.0
>>> -I/localhome/krehel/DUNE2.1/dune-pdelab conftest.cpp >&5
>> 
>> Well, you should be using the trunk of the dune-core modules as well if you're
>> using the trunk of dune-pdelab.
>> 
>> Bye,
>> Jö.
>> 
>> --
>> The irony is that Bill Gates claims to be making a stable operating
>> system and Linus Torvalds claims to be trying to take over the world.
>> -- seen on the net
>> 
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.4.10 (GNU/Linux)
>> 
>> iQIVAwUBTkEooumx9CKgc0OWAQjsxw/9EGNv4TbvOxVsXVVvBewoi6KAwqDaIAa0
>> ihgYZ/jZNBwKnQdCF3Ui0mFauGyeHa5kK0RHlBLmMwX4TzT+F4nz3atM5kH1B+t8
>> RBFD35iSwUna+5CR+n3tE2C/w7WbueZPwc+xEwfCaBXMX1A2BoRP0XUvbJSLA2Df
>> 2zAeaa3O8zJ4KJ1z0XsUOI39bDjqrm0SkrrsgyqdB/m4N0yTmWMdYBYz5IszhTcd
>> AufWaT/J2qJq/lIGDeVA2aewfIjkNNM1KnYOXuI8Ytdtpxpy3iNV3r8FVii/ePGD
>> HunGJQxGd0GSKF3X5LOKs1Egvrz0J2vCLWRlt4esAFPVA+Q9Vxmnt3v6faR6UpMd
>> UrfqyHjNpLCLbJytP0MQD5vnnyFmy/V2KPdM0PvUV/Ta9Jz/qUJdl4Z/6k1Q4EDO
>> Y5R0r95CX2kwbLRVkYzGkCBHqw2u8RbRIDPVlyGBHXQGOxGwZrTus7ta1poXS+kg
>> 0Mf6vTSideliXuN2GWWkM0iWBrWrX7bLHPGzZGCTrZ8xs1Ne59BQwF8Tub8l64JJ
>> eV1dWTG8pkxqeeF5xZjoi+Ro5vhOQCzt9lFsK2NhnR/zdAEznjbRWI6cxrDRuxDu
>> rV828S8ucytH8Yxo8ebWodjV0bGGh6CjaQTVs1H+3WGD9/3JKcwRd2bZGjD3I6mO
>> t/TXwJnqswo=
>> =/DrK
>> -----END PGP SIGNATURE-----
>> 
>> 
> 
> _______________________________________________
> dune-pdelab mailing list
> dune-pdelab at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-pdelab

Steffen Müthing
Universität Stuttgart
Institut für Parallele und Verteilte Systeme
Universitätsstr. 38
70569 Stuttgart
Tel: +49 711 685 88429
Fax: +49 711 685 88340
Email: steffen.muething at ipvs.uni-stuttgart.de