syfi team mailing list archive

[Martin Sandve Alnæs <martinal@xxxxxxxxx>]

On Fri, May 8, 2009 at 1:24 PM, Harish Narayanan
<harish.mlists@xxxxxxxxx> wrote:
> SyFi wrote:
>> One or more new changesets pushed to the primary SyFi repository.
>> A short summary of the last three changesets is included below.
>>
>> changeset:   515:bfc53fca84e60245a5f3996e80a61ab66eb36921
>> tag:         tip
>> user:        "Martin Sandve Alnæs <martinal@xxxxxxxxx>"
>> date:        Wed May 06 17:54:30 2009 +0200
>> files:       demo/HyperElasticityModels/cpp/HyperElasticity.ufl demo/HyperElasticityModels/cpp/Law.ufl demo/HyperElasticityModels/cpp/Makefile demo/HyperElasticityModels/cpp/main.cpp
>> description:
>> Implemented fiber function (identity) to test Fung. Seems correct!
>> Note that compiling all these Fung form variants from one form file
>> used almost all of my 2 Gb of memory. I guess some memory profiling
>> would be in order.
>
> With these recent AD changes in UFL, my toy (SVK) hyperelasticity
> attempts seem to be working too. I am using FFC and they also work in
> direct notation[1].

Great! I haven't got around to test the elasticity applications yet.

> Very cool Martin!

Thanks :)

> [1] i.e., they do not require the use of index notation, but I think you
> are using index notation in your examples here for debugging.

Yes, in the above I compute the same forms in three different ways,
without "derivative", with "derivative" used once, and with "derivative"
used twice, and get the same results for several material laws.

Martin