It means that there is a bug with assembly caching, and we should switch it off for now. Cjc On 25 Nov 2014 23:52, "Hiroe Yamazaki" <h.yamazaki@imperial.ac.uk> wrote:

Dear Colin and Lawrence,

switching off assembly caching worked for me too. Does this mean that

1. the error was all a matter of assemble() and nothing related to .assign() and/or .split()?

2. I had better use "parameters['assembly_cache']['enabled'] = False" as a default setting to make sure assemble() works as expected?

Thanks, Hiroe

2014-11-25 21:06 GMT+00:00 Cotter, Colin J <colin.cotter@imperial.ac.uk>:

...

Dear all,

Just to clarify that the file to execute the test is called "test_assign.py". For ease I've pasted it here: https://gist.github.com/colinjcotter/4b7eb6731a6cb388efee

Switching off assembly caching does indeed fix this for me, i.e. parameters['assembly_cache']['enabled'] = False

at the top of the file.

all the best --cjc

________________________________________ From: firedrake-bounces@imperial.ac.uk [firedrake-bounces@imperial.ac.uk] on behalf of Hiroe Yamazaki [h.yamazaki@imperial.ac.uk] Sent: 25 November 2014 19:03 To: firedrake Subject: [firedrake] .assign() with a function from a mixed function space

Dear all,

I got confused with .assign() with a function from a mixed function space and I would appreciate some suggestions. The test code is in bitbucket.org/colinjcotter/slicemodels/branch/tidy revision f87e865.

I expect a non-zero kinetic energy is printed out in line 54, but the result is still 0. To get an updated kinetic energy, I need line 57.

Am I misunderstanding what .assign() and/or .split() do?

All the best, Hiroe

_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake

_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake