- firedrake - imperial.ac.uk

Profiling
by Floriane Gidel [RPG] 19 Oct '17

19 Oct '17

Dear all, How can I profile my firedrake script to known the allocation of the computational time? I tried python -m cProfile -o output.out my_script.py but the output file is not as expected (part of it is pasted below). Thank you, Floriane output.out: {(sG/Users/mmfg/firedrake/lib/python2.7/site-packages/sympy/core/numbers.pyi?t_eval_is_prime(iig???????>g?3??O?>{(sK/Users/mmfg/firedrake/lib/python2.7/site-packages/sympy/core/assumptions.pyi?t_ask(iig???????>g?3??O?>0(s2/Users/mmfg/firedrake/src/ufl/ufl/tensoralgebra.pyisCofactor(iig???????>g???????>{(s2/Users/mmfg/firedrake/src/ufl/ufl/tensoralgebra.pyi<module>(iig???????>g???????>0(sH/Users/mmfg/firedrake/lib/python2.7/site-packages/sympy/series/gruntz.pyivR(iig,C??6?gT?qs*9?{(sH/Users/mmfg/firedrake/lib/python2.7/site-packages/sympy/series/limits.pyiR(iig,C??6?gT?qs*9?0(sI/Users/mmfg/firedrake/lib/python2.7/site-packages/mpmath/ctx_mp_python.pyiHt _constant(iig?h㈵??>g?h㈵??>{(sI/Users/mmfg/firedrake/lib/python2.7/site-packages/mpmath/ctx_mp_python.pyiR(iig?h㈵??>g?h㈵??>0(s0/Users/mmfg/firedrake/src/PyOP2/pyop2/caching.pyi{t__new__(i/i6g1{?v?A?gV??6o???{(s-/Users/mmfg/firedrake/src/PyOP2/pyop2/base.pyi>tset(iigT?qs*?>g??̔???>(s-/Users/mmfg/firedrake/src/PyOP2/pyop2/base.pyi? __init__(iig?k$ ??gM0?k????(sB/Users/mmfg/firedrake/src/firedrake/firedrake/functionspaceimpl.pyi?texterior_facet_node_map(iig?h㈵??>g{?V???0?(s-/Users/mmfg/firedrake/src/PyOP2/pyop2/base.pyi8 ttoset(iig???????>g???????>(s9/Users/mmfg/firedrake/src/firedrake/firedrake/assemble.pyi?t _assemble(iig ?F??g"O???|??(sB/Users/mmfg/firedrake/src/firedrake/firedrake/functionspaceimpl.pyidof_dset(iig???????>gT?qs*?>(s5/Users/mmfg/firedrake/src/firedrake/firedrake/mesh.pyi?tlocal_facet_dat(iig?h㈵??>g?t><K??(s-/Users/mmfg/firedrake/src/PyOP2/pyop2/base.pyi?

2 2

Problem with PyOP2 on KNL machine
by Justin Chang 19 Oct '17

19 Oct '17

Hi everyone, I installed firedrake onto our local KNL box which runs on CentOS 7.4, so I managed to install all the necessary requirements using --no-package-manager. The installation returns no error but I am seeing this error: (firedrake) [root@caamhpc ~]# python -c "from firedrake import *" Traceback (most recent call last): File "<string>", line 1, in <module> File "/root/Software/firedrake/src/firedrake/firedrake/__init__.py", line 29, in <module> from pyop2 import op2 # noqa File "/root/Software/firedrake/src/PyOP2/pyop2/__init__.py", line 4, in <module> from pyop2.op2 import * # noqa File "/root/Software/firedrake/src/PyOP2/pyop2/op2.py", line 42, in <module> from pyop2.base import i # noqa: F401 File "/root/Software/firedrake/src/PyOP2/pyop2/base.py", line 57, in <module> from pyop2.sparsity import build_sparsity ImportError: /root/Software/firedrake/src/PyOP2/pyop2/ sparsity.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _intel_fast_memset I have never seen this before. Here's some information that may be helpful: (firedrake) [root@caamhpc ~]# ldd /root/Software/firedrake/src/PyOP2/pyop2/ sparsity.cpython-36m-x86_64-linux-gnu.so linux-vdso.so.1 => (0x00007fffc8f54000) libpetsc.so.3.07 => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libpetsc.so.3.07 (0x00007fbf5e1f0000) libpython3.6m.so.1.0 => /usr/local/lib/libpython3.6m.so.1.0 (0x00007fbf5dcb9000) libpthread.so.0 => /lib64/libpthread.so.0 (0x00007fbf5da86000) libc.so.6 => /lib64/libc.so.6 (0x00007fbf5d6c2000) liblapack.so.3 => /lib64/liblapack.so.3 (0x00007fbf5cf64000) libblas.so.3 => /lib64/libblas.so.3 (0x00007fbf5cd0b000) libparmetis.so => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libparmetis.so (0x00007fbf5cacf000) libmetis.so => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libmetis.so (0x00007fbf5c858000) libhdf5hl_fortran.so.10 => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libhdf5hl_fortran.so.10 (0x00007fbf5c645000) libhdf5_fortran.so.10 => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libhdf5_fortran.so.10 (0x00007fbf5c409000) libhdf5_hl.so.10 => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libhdf5_hl.so.10 (0x00007fbf5c1e7000) libhdf5.so.10 => /root/Software/firedrake/lib/python3.6/site-packages/petsc/lib/libhdf5.so.10 (0x00007fbf5bd08000) libm.so.6 => /lib64/libm.so.6 (0x00007fbf5ba06000) libgfortran.so.3 => /lib64/libgfortran.so.3 (0x00007fbf5b6e3000) libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007fbf5b4a7000) libmpicxx.so.12 => /opt/intel/compilers_and_libraries_2016.3.210/linux/mpi/intel64/lib/libmpicxx.so.12 (0x00007fbf5b287000) libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007fbf5af7e000) libdl.so.2 => /lib64/libdl.so.2 (0x00007fbf5ad7a000) libmpifort.so.12 => /opt/intel/compilers_and_libraries_2016.3.210/linux/mpi/intel64/lib/libmpifort.so.12 (0x00007fbf5a9dc000) libmpi.so.12 => /opt/intel/compilers_and_libraries_2016.3.210/linux/mpi/intel64/lib/libmpi.so.12 (0x00007fbf5a20c000) librt.so.1 => /lib64/librt.so.1 (0x00007fbf5a004000) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007fbf59dee000) libutil.so.1 => /lib64/libutil.so.1 (0x00007fbf59bea000) /lib64/ld-linux-x86-64.so.2 (0x00007fbf5f8b9000) (firedrake) [root@caamhpc ~]# which mpicc /opt/intel/compilers_and_libraries_2016.3.210/linux/mpi/intel64/bin/mpicc (firedrake) [root@caamhpc ~]# mpicc --version gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-16) Copyright (C) 2015 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. This system wants to use intel compilers but I told it to use GNU compilers. I suspect this error has something to do with intel. I can try and provide more information as necessary. Thanks, Justin

2 1

where is firedrake...
by Kara, Erdi 12 Oct '17

12 Oct '17

Hi I have recently installed Firedrake to my Ubuntu 16 and made all tests successfully. However it doest show up in any of Python3 directories. Therefore, the command " from firedrake import * " never works. If you have any suggestion, I would really appreciate. Have a good day.

3 18

CG2 giving me 4th order L2 convergence
by Justin Chang 05 Oct '17

05 Oct '17

CG2 for poisson equation should be 3rd order L2 convergence but for some reason I am getting near 4th order convergence. If I use DG2 elements, I get the 3rd order convergence as theory suggests. Why is this happening? Below is the snippet of my code mesh = UnitSquareMesh(20,20,quadrilateral=False) Q = FunctionSpace(mesh,CG,2) ... x = SpatialCoordinate(Q.mesh()) u_exact = interpolate(sin(2*x[0]**2*math.pi)*sin(2*x[1]**2*math.pi),Q) f = interpolate( 16*math.pi**2*sin(2*math.pi*x[0]**2)*sin(2*math.pi*x[1]**2 )*(x[0]**2+x[1]**2) - 4*math.pi*(cos(2*math.pi*x[0]**2)*sin(2*math.pi*x[1]**2) + sin(2 *math.pi*x[0]**2)*cos(2*math.pi*x[1]**2)),Q) ... L2_error = errornorm(u_exact,u_h) I never had this "problem" earlier this year when I ran CG2 experiments (though I didn't use the UFL format, but I don't think that's the problem here). Justin

3 3

netCDF4 error
by Burns, Paul 21 Sep '17

21 Sep '17

Dear Sir/Madam, I am getting the following netCDF4 error when I try to run Gusto: ---------------------- UFL:WARNING Discontinuous Lagrange element requested on interval * interval, creating DQ element. UFL:WARNING Discontinuous Lagrange element requested on interval * interval, creating DQ element. Traceback (most recent call last): File "examples/sk_nonlinear.py", line 112, in <module> stepper.run(t=0, tmax=tmax) File "/Users/pb412/firedrake/src/gusto/gusto/timeloop.py", line 169, in run state.dump(t, pickup=False) File "/Users/pb412/firedrake/src/gusto/gusto/state.py", line 345, in dump self.pointdata_output.dump(self.fields, t) File "/Users/pb412/firedrake/src/gusto/gusto/state.py", line 112, in dump var[idx, :] = vals File "netCDF4/_netCDF4.pyx", line 1908, in netCDF4._netCDF4.Dataset.__exit__ (netCDF4/_netCDF4.c:14289) File "netCDF4/_netCDF4.pyx", line 2012, in netCDF4._netCDF4.Dataset.close (netCDF4/_netCDF4.c:16381) File "netCDF4/_netCDF4.pyx", line 1996, in netCDF4._netCDF4.Dataset._close (netCDF4/_netCDF4.c:16289) File "netCDF4/_netCDF4.pyx", line 1581, in netCDF4._netCDF4._ensure_nc_success (netCDF4/_netCDF4.c:12601) RuntimeError: NetCDF: HDF error -------------------------------------------------------------------------- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode 1. ---------------------- There error occurs independent of the model I run (e.g. Rayleigh_Taylor, sk_nonlinear, etc) or the branch I am using (e.g. master, boussinesq_2d_lab, etc). So it seems there is something wrong with the way I have setup the netCDF4 software on my machine...? Please advise what the best way forward is. Regards, Paul

3 4

running in Parallel
by Floriane Gidel [RPG] 05 Sep '17

05 Sep '17

Dear all, I need to run my firedrake code in parallel but some of the commands (like appending an array) can't be executed in parallel. How can I force some lines to be executed by one core only? Thanks in advance, Floriane

3 4

Simplest way to access cell2vertex array from mesh
by Dodwell, Timothy 05 Sep '17

05 Sep '17

Hi all, Sorry for simple questions again, what is the simplest way to access cell to vertex data from a mesh. I am going around in circles! I am looking at the “pyop2” example of computing midpoints of cells I.e. vertices = op2.Set(num_vertices) cells = op2.Set(num_cells) cell2vertex = op2.Map(cells, vertices, 3, […]) So I need the data from the mesh to replace […] Thanks Tim Dr Tim Dodwell Senior Lecturer in Engineering Mathematics Rm 221 - Harrison Building College of Engineering, Mathematics & Physical Sciences University of Exeter Exeter Devon EX4 4PY mail: t.dodwell(at)exeter.ac.uk<http://exeter.ac.uk/> tel: +44 (0)1392 725899 mob: +44 (0)7745 622870 web: http://emps.exeter.ac.uk/engineering/staff/td336 Papers and Pre-prints: @Research-Gate<https://www.researchgate.net/profile/Timothy_Dodwell> Citations: @Google-Scholar<https://scholar.google.co.uk/citations?user=lPpjRfUAAAAJ&hl=en>

2 3

Normal to curved surface
by Dodwell, Timothy 25 Aug '17

25 Aug '17

Hi all, New to firedrake, I am building a test problem for a new PhD which is poro-elastic consolidation of a sphere, I want to apply a pressure boundary (Neumann) condition to a curved surface. I.e \int (\sigma \dot n) \dot v\; ds Where n is the normal vector to the surface ds. Is there a function normal() which returns it for me. A look a documentation doesn’t look promising! Cheers Tim Dr Tim Dodwell Senior Lecturer in Engineering Mathematics Rm 221 - Harrison Building College of Engineering, Mathematics & Physical Sciences University of Exeter Exeter Devon EX4 4PY mail: t.dodwell(at)exeter.ac.uk<http://exeter.ac.uk/> tel: +44 (0)1392 725899 mob: +44 (0)7745 622870 web: http://emps.exeter.ac.uk/engineering/staff/td336 Papers and Pre-prints: @Research-Gate<https://www.researchgate.net/profile/Timothy_Dodwell> Citations: @Google-Scholar<https://scholar.google.co.uk/citations?user=lPpjRfUAAAAJ&hl=en>

2 1

FEniCS'18 Conference 21-23 March, University of Oxford
by Patrick Farrell 24 Aug '17

24 Aug '17

We're pleased to announce that registration and talk/poster abstract submission is now open for the FEniCS Conference 2018, 21-23 March 2018, at the University of Oxford. For further details please see https://fenicsproject.org/fenics18/ . We are very much looking forward to seeing you all in Oxford! Best regards, Patrick

1 0

Using Intel's SDE and VTune for Firedrake/Python
by Justin Chang 24 Aug '17

24 Aug '17

Hi guys, I want to measure AI using the procedure described here on the NERSC website: http://www.nersc.gov/users/application-performance/measuring-arithmetic-int… Basically they obtain FLOPs from Intel's SDE and the bytes from VTune. And in their C code example, they inserted macros to indicate where to start and stop the measurements. Does anyone here know how to do something similar for Python-based programs? Or if there's a "hack" to allow me to insert these C macros into the firedrake code somewhere? Thanks, Justin

3 2