[dune-functions] Bug in DiscreteGlobalBasisFunction

Simon Praetorius simon.praetorius at tu-dresden.de
Fri Aug 13 10:42:11 CEST 2021


Thanks Carsten for the info.

I have implemented an own DiscreteFunction representation, inspired by 
the DiscreteGlobalBasisFunction but with some additional functionality 
and something made differently. But, I will have a look whether I have 
adopted this part of the code. So far, I've not found a situation with 
strange results in some sub-components that I couldn't explain.

How to detect such errors in the future? An extended test is probably a 
good idea. Thanks for implementing this. Would it make sense to include 
valgrind in the testsuite, i.e., run the examples through a memory 
debugger? Maybe also a static code analyzer. There exist lost of tools, 
but they are not yet included in our automatic test framework.

Best regards,
Simon


Am 12.08.21 um 15:57 schrieb Carsten Gräser:
> Dear all,
> yesterday I fixed a bug that I introduced in the master branch of
> DiscreteGlobalBasisFunction some time ago. I'd like to point this out,
> because as a nasty consequence the function silently computed wrong 
> numbers
> in a special case. To be precise this happened if all of the following 
> apply:
>
> * You're using either the master or the releases/2.8 branch.
> * You use DiscreteGlobalBasis with a SubspaceBasis.
> * The tree path associated to the SubspaceBasis contains non-zero 
> entries.
>
> E.g. this is the case when writing the pressure component to vtk in the
> Stokes example coming with dune-functions. Unfortunately this case was 
> not
> covered by our tests so far. The bug is now fixed in both branches. For
> details see:
>
> https://gitlab.dune-project.org/staging/dune-functions/-/merge_requests/310
>
> I really hope that this did not effect anyone else despite me.
>
> Best regards,
> Carsten
>
-- 
Dr. Simon Praetorius
Technische Universität Dresden
Institute of Scientific Computing
phone: +49 351 463-34432
mail: simon.praetorius at tu-dresden.de
web: https://tu-dresden.de/mn/math/wir/das-institut/beschaeftigte/simon-praetorius


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5204 bytes
Desc: S/MIME Cryptographic Signature
URL: <https://lists.dune-project.org/pipermail/dune-functions/attachments/20210813/6fd5b481/attachment.bin>


More information about the dune-functions mailing list