I think we should not insist on too rigorous a correspondence between
the names we use in the periodic table and the UFL element names, since
the needs are different. The former are intended to make an accurate
and visually appealing poster, and possibly to have some unifying effect
on usage by researchers, while UFL has many other needs.
Specifically, in my comment of 27 February at
https://bitbucket.org/fenics-project/ufl/pull-request/7/introduce-notation-for-the-periodic-table/diff
I included an image showing how the names will look in the periodic
table (including subscripts) and then a possible pure text version that
could be used in UFL. Based on the discussion so far, I would stick
with the image for the poster, but appreciate that the developers may
need to make changes to the UFL version (two such having been mentioned
so far: changing dP to DP to avoid possible confusion, and removing the
c from BDMce to avoid redundancy with the specification of the cell
type, etc.). Hopefully the differences can be made consistently. E.g.,
if you remove the "c" (meaning "cubic variant") from BDM and RT, it
should alo be removed from the Nedelec cubical elements and dPc.
-- Doug
On 02/28/2014 04:30 AM, Anders Logg wrote:
On Fri, Feb 28, 2014 at 10:55:05AM +0100, Marie E. Rognes wrote:
On 02/28/2014 09:41 AM, Anders Logg wrote:
Other opinions?
Would you consider dropping the c (or q/C/Q) for the BDMs/RTs on
cubes? In the UFL FiniteElement constructor, the cell is given
separately, so there is no need for the family name to indicate
this. Advantages would be: simpler names and greater possibility of
code independence wrt cell type.
That would also be consistent with the use of P for both triangles and
tetrahedra.
--
Anders
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics
