November 2017 - firedrake - imperial.ac.uk

Fw: REQTSK0145114 - IT request - Install software - Firedrake (Mac)
by Onno Bokhove 14 Nov '17

14 Nov '17

Hi, Can the firedrake install script be modified to use a different version of lbspatialindex ? Given the error meesage below. Some files from the gtest are missing. It can build by hand but that requires an modification with every update. Best wishes, Onno & Tim ________________________________ From: IT Service Desk <leeds(a)service-now.com> Sent: Tuesday, November 14, 2017 11:38 AM To: Onno Bokhove Subject: REQTSK0145114 - IT request - Install software - Firedrake (Mac) Dear Onno, The firedake-install script fails to complete as there appears to be an error in the configuration of the libspatialindex code which the script doesn't handle: libspatialindex/test/gtest/gtest-1.7.0:config.status:1140: error: cannot find input file: Makefile.in gtest doesn't appear to be used as libspatialindex compiles without it (outside of the firedrake-install script). Is there a way to get the firedake-install script to use a different version of libspatialindex rather than downloading the head from the git repository every time. Regards, Tim. IT Service Desk University of Leeds t: 0113 343 3333 [http://leedsdev.service-now.com/twitter.pngx]@ITServicesUoL<https://twitter.com/ITServicesUoL> For support and online information please visit: http://it.leeds.ac.uk<http://it.leeds.ac.uk/> We welcome any feedback about our service. Just visit https://leeds.onlinesurveys.ac.uk/it-service-desk-feedback Ref:MSG1606355

3 6

Dumbcheckpointing
by Floriane Gidel [RPG] 12 Nov '17

12 Nov '17

Dear all, I saved a function h using DumbCheckpoint through: res = 0.1 Nx = int(Lx/res) + 1 Ny = int(Ly/res) + 1 hor_mesh = RectangleMesh(Nx,Ny,Lx,Ly,quadrilateral=True) V = FunctionSpace(hor_mesh, "CG", 1) h_n0 = Function(V) ... [assign value to h_n0] ... h_data = DumbCheckpoint("h", mode=FILE_CREATE) h_data.store(h_n0) When trying to load it in another code, with: res = 0.1 Nx = int(Lx/res) + 1 Ny = int(Ly/res) + 1 hor_mesh = RectangleMesh(Nx,Ny,Lx,Ly,quadrilateral=True) V = FunctionSpace(hor_mesh, "CG", 1) h_n1 = Function(V) dumb_file = DumbCheckpoint("h", mode=FILE_READ) dumb_file.load(h_n1) I get the following error: File "TOC.py", line 130, in <module> dumb_file.load(h_n1) File "/Users/mmfg/firedrake/src/firedrake/firedrake/checkpointing.py", line 237, in load v.load(self.vwr) File "PETSc/Vec.pyx", line 447, in petsc4py.PETSc.Vec.load (src/petsc4py.PETSc.c:95805) petsc4py.PETSc.Error: error code 76 [0] VecLoad() line 925 in /private/var/folders/7h/wbj8xp7n3g5cfbr32ctcmwzcy3jf53/T/pip-yGXqPh-build/src/vec/vec/interface/vector.c [0] VecLoad_Default() line 397 in /private/var/folders/7h/wbj8xp7n3g5cfbr32ctcmwzcy3jf53/T/pip-yGXqPh-build/src/vec/vec/utils/vecio.c [0] VecLoad_HDF5() line 257 in /private/var/folders/7h/wbj8xp7n3g5cfbr32ctcmwzcy3jf53/T/pip-yGXqPh-build/src/vec/vec/utils/vecio.c [0] Error in external library [0] Error in HDF5 call H5Dopen2() Status -1 Exception RuntimeError: RuntimeError("Can't decrement id ref count (Can't close file, there are objects still open)",) in 'h5py._objects.ObjectID.__dealloc__' ignored Do you know what this error mean ? Thank you, Floriane

1 2

boundary extrema [was: problems with importing a mesh]
by G. D. McBain 10 Nov '17

10 Nov '17

> On 08/11/17 09:13, Lawrence Mitchell wrote: > >> On 03/11/17 20:15, Francis Poulin wrote: >> >>> I don't suppose >>> there is an easy way to find out the maximum of the boundary value in >>> the numerical solution? >> >> If you are imposing strong (Dirichlet) conditions, then the solution >> on the boundary nodes is, by definition, whatever you set it to. >> If you have some other means of enforcing boundary conditions and just >> want to know that the values are on the boundary, you could do: >> V = FunctionSpace(...) >> selector = DirichletBC(V, 0, "on_boundary") >> f = Function(V) >> >> solve into f >> >> boundary_values = f.dat.data_ro[selector.nodes, ...] >> >> Note this assumes that inspecting the point values of your space makes >> sense: this is fine for CG and DG, not so for other spaces. Hey, thanks very much! I didn't realize that the object returned by DirichletBC had a .nodes attribute. That's much more direct than the procedures in the other codes! (As usual for Firedrake.) This is very useful; in gratitude, I attach a simple example from Ostroumov, G. A. (1958). Free convection under the conditions of the internal problem. Technical Report [NACA-TM](https://wiki.memjet.local/display/lib/NASA+Technical+Memorandum)-1407, NACA, eq. 5.18, p. 34 lap w = -x in the first quadrant of the unit circle with w=0 on perimeter and y-axis and natural boundary conditions on x-axis, where the boundary maximum is to be sought. The exact solution is w=x(1-x**2-y**2)/8 with maximum w(1/sqrt(3), 0) = 1/12 sqrt(3). Using the same Gmsh-generated mesh as in the previous thread (again: -clscale 0.0625), but now also marking the x-axis as Physical Line 3, the nodal maximum is found as w=0.048122 at x=0.562500, the value being accurate to the fifth decimal place and the location as good as possible on the mesh.

1 0

problems with importing a mesh
by Francis Poulin 09 Nov '17

09 Nov '17

Hello, I have created a mesh of an oceanic basin that I have triangulated with gmsh. Finally! I am very excited to solve for the wind-driven gyre solutions with this new mesh (as I discussed back in March at the workshop), unfortunately, when I try to import the mesh I get the following: mesh = Mesh("test3.msh") ValueError: Provided mesh has some entities not reachable by traversing cells (maybe rogue vertices?) Any ideas as to what I might have done wrong and how I can fix this? Cheers, Francis ------------------ Francis Poulin Associate Dean, Undergraduate Studies Associate Professor Department of Applied Mathematics University of Waterloo email: fpoulin(a)uwaterloo.ca Web: https://uwaterloo.ca/poulin-research-group/ Telephone: +1 519 888 4567 x32637

4 17

SIG meeting December 11th-12th, Exeter
by Shipton, Jemma 08 Nov '17

08 Nov '17

Dear all, I am copying below an email from Martin Schreiber who is organising our next SIG meeting in Exeter on December 11th-12th. I thought it might be of interest to some on this list... The special interest group (SIG) "Next generation time-stepping strategies for computer simulations of multi-scale fluid flows", funded by the EPSRC UKFluids network, will have its 2nd meeting on December 11th and 12th 2017 in Exeter to discuss recent and future research in the field of numerical time stepping algorithms which also involves software engineering. Plenary speakers are: *Michael Klemm* (Intel, CEO of the OpenMP organization) Title: "OpenMP at Twenty - Past, Present, and Future" *Mario Acosta* (Barcelona Supercomputing Center) Title: "BSC performance activities on NEMO: Exploring the use of mixed precision in the NEMO model" *Chris Maynard* (MetOffice) Title: "Solving the solver problem: Can software enable productivity, portability and productivity?" *Andrew McRae* (Oxford) Title: "Reduced precision in weather and climate simulations" The main topic for this meeting will be *Mixed precision and how hardware impacts software abstractions* More information on our SIG is available at https://ngtssite.wordpress.com/ [https://ngtssite.files.wordpress.com/2017/06/homepage11.jpg?w=1200]<https://ngtssite.wordpress.com/> SIG Next Generation Time Stepping<https://ngtssite.wordpress.com/> ngtssite.wordpress.com Research on next generation time stepping methods for upcoming high performance computing architectures The meeting will also involve a joint dinner and a discussion about promising future research. SIG members are from various UK universities and industry partners such as the UK Met Office, the European Centre for Medium Range Weather Forecasts (ECMWF) and Cray UK. Please register by November 10th at https://goo.gl/forms/Dg8dreG59rJcn7u63 PhD and PostDoc travel grants: Interested PhD students or postdocs can apply for travel grants to attend the meeting. The grants will fund train tickets and hotel costs. To apply, please send in an application explaining (1) why you are interested in attending the meeting and (2) how the topic aligns with your research. Please send your application to d.ruprecht(a)leeds.ac.uk, j.shipton(a)imperial.ac.uk and m.schreiber(a)exeter.ac.uk before November 10th 2017. All the best, Martin

1 0

Firedrake installation on cluster
by Burns, Paul 02 Nov '17

02 Nov '17

Dear Sir/Madam, I am trying to get Firedrake up and running on the Isca cluster at Exeter. The standard install method (i.e., executing 'python3 firedrake-install') does not work, however, we have managed to install Firedrake by developing an install script that loads modules manually, that sets a number of petsc options, and that passes a number of extra options to firedrake-install. However, when I try and run a simulation on Gusto I get the following error: Traceback (most recent call last): File "./examples/boussinesq_2d_lab.py", line 1, in <module> from gusto import * File "/gpfs/ts0/home/pb412/firedrake-20171026/src/gusto/gusto/__init__.py", line 11, in <module> from gusto.state import * # noqa File "/gpfs/ts0/home/pb412/firedrake-20171026/src/gusto/gusto/state.py", line 3, in <module> from netCDF4 import Dataset File "/gpfs/ts0/home/pb412/firedrake-20171026/lib/python3.5/site-packages/netCDF4/__init__.py", line 3, in <module> from ._netCDF4 import * ImportError: /gpfs/ts0/home/pb412/firedrake-20171026/lib/python3.5/site-packages/netCDF4/_netCDF4.cpython-35m-x86_64-linux-gnu.so: undefined symbol: _PyThreadState_UncheckedGet So it looks like there is some problem with the netCDF4 library. Please advise. Regards, Paul

4 22

Install failing on PETSc build
by William Booker [RPG] 01 Nov '17

01 Nov '17

Hi, The install script keeps failing for me whenever it tries to build PETSc, what should I do to fix this? Thanks Will

3 3

notebooks and passwords
by Francis Poulin 01 Nov '17

01 Nov '17

Hello everyone, It's been a while since I've used notebooks in firedrake but I tried it today and found that it now requires a password (or token) like in fenics. I did a bit of searching around and read that I should see what the active notebooks are. I tried that with the following, (firedrake) fpoulin@domlt32:~/Documents/Teaching/2018/AMATH353/ComputerLabs$ jupyter notebook list Currently running servers: http://localhost:8888/ :: /home/fpoulin/Documents/Teaching/2018/AMATH353/ComputerLabs but it doesn't given me a token, as I believe I should. Can someone help me figure out how to start a notebook in firedrake? Cheers, Francis ------------------ Francis Poulin Associate Dean, Undergraduate Studies Associate Professor Department of Applied Mathematics University of Waterloo email: fpoulin(a)uwaterloo.ca Web: https://uwaterloo.ca/poulin-research-group/ Telephone: +1 519 888 4567 x32637

3 3