- firedrake - imperial.ac.uk

This may be of interest to some
by Colin Cotter 23 May '14

23 May '14

http://arxiv.org/abs/1405.5245

1 0

Firedrake benchmarks
by Florian Rathgeber 16 May '14

16 May '14

Hi firedrakers, I have set up a repository of Firedrake benchmarks [1] to assess performance for a varied set of applications and parameters. Implemented so far are: 1) matrix assembly for mass, laplace, helmholtz 2) stationary Poisson 3) advection-diffusion (split form) 4) Navier-Stokes (split form) 5) Cahn-Hilliard I have started compiling some preliminary results [2], which I plan to updated from time to time as I have new measurements. As of now there is assembly and advection-diffusion. Cheers, Florian [1]: https://github.com/firedrakeproject/firedrake-bench [2]: https://gist.github.com/kynan/116211fa75d66a66311e

1 0

UFL reset
by Lawrence Mitchell 13 May '14

13 May '14

Dear firedrakers, with latest Firedrake, you'll require a new UFL + FFC combo. This is merging upstream work such that we can start looking at implementing non-affine geometries. The UFL update is a reset of master, if you have a pip install of UFL, please just reinstall as before. If, however, you're maintaining a local clone of our UFL repository you'll need to do the following to update to latest master. ** Note ** The following sequence of instructions discards any local changes you have on the master branch. You'll need to move those to a new branch before carrying on. $ git checkout master $ git fetch $ git reset --hard origin/master Please do email if something goes amiss. Cheers, Lawrence

1 0

mesh
by Cotter, Colin J 13 May '14

13 May '14

Hi Firedrakers, What's the easiest way to get an unstructured mesh of the unit square into Firedrake? Is there a built in mesh? cheers --cjc

3 2

Serendipity elements
by Colin Cotter 28 Apr '14

28 Apr '14

The following looks interesting. There is now a family of FEEC spaces corresponding to serendipity elements on quads/hexes. http://arxiv.org/abs/1404.6275 cheers --cjc

1 0

Fwd: [atmos-seminars] Atmos seminar - David Jackson, Met Office - 29 April 11.30 Huxley 711C
by Andrew McRae 27 Apr '14

27 Apr '14

This might be of interest to the more application-driven members of the group. Protip: an abridged version of this talk used these slides<http://www.jodrellbank.manchester.ac.uk/meetings/nam2012/archive/MST3/Jacks…>(from 2012) ---------- Forwarded message ---------- From: Williams, Sian R <sian.williams07(a)imperial.ac.uk> Date: 22 April 2014 11:31 Subject: [atmos-seminars] Atmos seminar - David Jackson, Met Office - 29 April 11.30 Huxley 711C To: spat <spat(a)imperial.ac.uk>, atmos-seminars < atmos-seminars(a)imperial.ac.uk> Hi everyone, David Jackson from the Met Office will be giving the Atmospheric Physics seminar next Tuesday, which might be interesting for ‘space’ people too. As usual it will take place at 11.30am in Huxley 711C. The title of his talk is "The Met Office Unified Model and its extension to the thermosphere”. The abstract is below. If anyone would like to meet with David after his talk please let me know in the next few days. He is happy to discuss the Met Office’s space weather activities more informally in the afternoon if this is interesting to anyone. Best wishes, Sian *The Met Office Unified Model and its extension to the thermosphere* The Met Office Unified Model (UM) is a troposphere / stratosphere / mesosphere model which is used widely for both weather forecasting and climate studies. It includes a comprehensive representation of atmospheric dynamics, physics and chemistry, and currently has an upper boundary near 85 km. In order to develop an improved capability for space weather forecasts, it is desirable to develop a coupled system of models representing the domain from the Sun to the Earth's surface. Much space weather forecast research is focused on the thermosphere and ionosphere, since that is where many space weather impacts are seen. While the thermosphere and ionosphere are largely driven from above, recent research has shown that the coupling between this region and the lower atmosphere is also important. For example, non-migrating tides forced in the tropical troposphere have been linked to variations in the ionospheric F region. Therefore, representation of such coupling is important. An attractive way of doing this is by building a "whole atmosphere" model, which spans the neutral atmosphere from the Earth's surface to the exobase (around 600 km) and represents the coupling between different atmospheric levels in a self-consistent manner. In this presentation, the dynamical, physical and chemical developments required to make this extended model a reality are discussed, and some initial results are presented.

1 0

Reminder, meeting today 1pm
by Lawrence Mitchell 25 Apr '14

25 Apr '14

Dear firedrakers, for those of you who are around, a reminder that we have a meeting today at 1pm in WPL301. Lawrence

1 1

Re: [firedrake] Talk at FEniCS '14
by Fabio Luporini 25 Apr '14

25 Apr '14

Right, I should submit it to arxiv. How does this work usually? I guess I have to submit it, then they "review" it, and finally they just upload it, nothing more, right? And, as far as I know, this process shouldn't take much. -- Fabio 2014-04-25 12:13 GMT+01:00 Mitchell, Lawrence < lawrence.mitchell(a)imperial.ac.uk>: > On 25/04/14 11:59, Fabio Luporini wrote: > > Dear firedrakers, > > > > Attached you find the abstract for the talk I am planning to deliver at > > FEniCS '14. Any feedback would be appreciated. > > > > I think this looks fine. It would be nice to reference your submitted > SC paper, is it on the arxiv? > > Lawrence > > > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake >

2 1

Talk at FEniCS '14
by Fabio Luporini 25 Apr '14

25 Apr '14

Dear firedrakers, Attached you find the abstract for the talk I am planning to deliver at FEniCS '14. Any feedback would be appreciated. Regards -- Fabio

2 1

petsc/petsc4py incompatibilities
by Lawrence Mitchell 23 Apr '14

23 Apr '14

Dear firedrakers, we currently rely on petsc next, but you may have noticed, if installing petsc4py recently, that you can't build it with petsc next right now. Hopefully our dependencies will settle down a bit soon and you won't have to go through this upgrade dance. In the mean time, if you see a pip install petsc4py error like the below: src/petsc4py.PETSc.c:206419:43: error: ‘DMADDA’ undeclared (first use in this function) src/petsc4py.PETSc.c:206419:43: note: each undeclared identifier is reported only once for each function it appears in error: command '/usr/bin/mpicc' failed with exit status 1 Please do the following: $ git clone https://bitbucket.org/petsc/petsc4py Apply this patch: https://gist.github.com/wence-/11179460 and then in the petsc4py directory $ sudo python setup.py install (or no sudo if you haven't been using sudo to install things) Cheers, Lawrence

1 0