dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Thu, Sep 24, 2009 at 04:13:25PM -0500, Robert Kirby wrote:
> Anders, two comments:
> 1.) You should use de Casteljau's algorithm for the Bernstein
> polynomials.

I'm not aware of that, but I'll make a note of it for when the time comes
to reimplement it in C++ (for Andre ;-).

> 2.) Using the recurrence relations in new version of FIAT will avoid
> the singular mapping and should allow evaluation outside the reference
> simplex.

Nice!

> But we need to make time to smooth this version out and
> hook it up to ffc.

Yes, I won't be able to do it just now. Maybe after the deadline for
the FEniCS book chapters.

--
Anders



> Rob
>
> On Wed, Sep 23, 2009 at 6:00 PM, Anders Logg <logg@xxxxxxxxx> wrote:
>
>     I just went through implementing Bernstein polynomials in Python
>     because Marie and I needed them for our experiments with adaptivity
>     and the FIAT basis functions could not be evaluated outside the
>     triangles (the mapping being singular along y = 1).
>
>     I've attached the code for the Bernstein polynomials. We will need to
>     implement this in C++ at some point when things are working.
>
>
>
>     On Wed, Sep 23, 2009 at 05:53:30PM -0500, Robert Kirby wrote:
>     > It probably should be added in dolfin/ufc.  Sometimes we want to play
>     around
>     > with things that aren't supported by ffc yet.  Having a full-featured c++
>     > library (and python bindings) would help.  An example would be Berstein
>     > polynomials, which use a different evaluation method than in ffc/fiat.
>     Another
>     > example is the transformation business I'm working that needs cell
>     jacobians
>     > and other geometric information.  I'd like to look at hand-done
>     implementations
>     > using the dolfin library (preferably PyDolfin) to make sure stuff works
>     before
>     > going through the labor of teaching ffc how to do it.
>     >
>     > Rob
>     >
>     > On Wed, Sep 23, 2009 at 5:45 PM, Garth N. Wells <gnw20@xxxxxxxxx> wrote:
>     >
>     >
>     >
>     >     Robert Kirby wrote:
>     >
>     >         Is there a method for getting the Jacobian of a mapping to a
>     reference
>     >         cell in dolfin, or is this always generated by ffc?
>     >
>     >
>     >
>     >     It's generated by FFC. Could be a suggestion for an addition to UFC.
>     >
>     >     Garth
>     >
>     >
>     >
>     >
>     >
>     ------------------------------------------------------------------------
>     >
>     >         _______________________________________________
>     >         DOLFIN-dev mailing list
>     >         DOLFIN-dev@xxxxxxxxxx
>     >         http://www.fenics.org/mailman/listinfo/dolfin-dev
>     >
>     >
>     >
>     >
>     >
>
>     > _______________________________________________
>     > DOLFIN-dev mailing list
>     > DOLFIN-dev@xxxxxxxxxx
>     > http://www.fenics.org/mailman/listinfo/dolfin-dev
>
>
>     -----BEGIN PGP SIGNATURE-----
>     Version: GnuPG v1.4.9 (GNU/Linux)
>
>     iEYEARECAAYFAkq6qI8ACgkQTuwUCDsYZdGEcACcD3FgXBQdSkw2n9/N0G2sVnxf
>     aMEAnRfNwekxL8KIxr3cihiwGCi749LC
>     =jVWS
>     -----END PGP SIGNATURE-----
>
>     _______________________________________________
>     DOLFIN-dev mailing list
>     DOLFIN-dev@xxxxxxxxxx
>     http://www.fenics.org/mailman/listinfo/dolfin-dev
>
>
>

> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev

Attachment: signature.asc
Description: Digital signature