Function.fine() method works unexpectedly

[Sergey Alyaev]

I have slightly modified auto-adaptive Navier-Stokes example.
I wanted to get the desired quantity out and i was advised to use something like:

assemble(w_h.fine()[0]*dss(0))

which does not work with the error message:

print "goal eval fine", assemble(w_h.fine()[0]*dss(0))
TypeError: 'Function' object does not support indexing

while the indexing works well with the original "w_h" which also is an object of the Function class.

I attach the code where the error pops up.

[Marie Rognes (meg-simula)]

On 07/15/11 18:24, Sergey Alyaev wrote:
> Public bug reported:
>
> I have slightly modified auto-adaptive Navier-Stokes example.
> I wanted to get the desired quantity out and i was advised to use something like:
>
> assemble(w_h.fine()[0]*dss(0))
>
> which does not work with the error message:
>
> print "goal eval fine", assemble(w_h.fine()[0]*dss(0))
> TypeError: 'Function' object does not support indexing
>
> while the indexing works well with the original "w_h" which also is an
> object of the Function class.
>
> I attach the code where the error pops up.
>
> ** Affects: dolfin
> Importance: Undecided
> Status: New
>

Thanks!

Could you also please attach the mesh and subdomains .xml files used, or
modify the example so that it does not require additional files?

--
Marie

[Sergey Alyaev (cobxo3bot)]

Sorry,

There was no option to add to files.
Here it is.

[Garth Wells (garth-wells)]

The name of the function Hierarchical::fine should be changed. My first
reaction was that

  u.fine()

should mean 'Hello, are you OK?'

The description of the class Hierarchical doesn't make any reference to
'fineness'. It just describes a straight tree. A 'finest function' is
ambiguous, How can I define this on different meshes or with different
bases?

If Hierarchical is a tree, then an allowable operation would be to
return a vector of leaves (just one leaf for a straight tree). Likewise,
'Hierarchical::coarse' should be 'Hierarchical::root'.

Garth

On 01/09/11 05:10, Marie Rognes wrote:
> ** Changed in: dolfin
> Status: Confirmed => In Progress
>

[Marie Rognes (meg-simula)]

The original request has now been fixed, try for instance (at the end of demo/auto-adaptive-navier-stokes/python):

# Extract solutions on coarsest and finest mesh:
(u0, p0) = w.coarse().split()
(u1, p1) = w.fine().split()

# Operate on most refined objects:
ds = Measure("ds")[outflow_markers.fine()]
M = assemble(u1[0]*ds(0))

# Examine and extract data from solver
for data in solver.adaptive_data():
    info(data, True)
print data["functional_value"]

Note that the names coarse() and fine() may change at some point soon.