July 2018 - firedrake - imperial.ac.uk

complex?
by Colin Cotter 01 Aug '18

01 Aug '18

Dear FD, Upon running the following code: from firedrake import * n = 20 mesh = UnitSquareMesh(n,n) V = FunctionSpace(mesh,"CG",1) I get the message at the bottom of this email. Does this mean I have incorrectly built a complex branch? (I built it by running the install script from the jenkinsfile in the complex branch, then changing to the complex branch.) all the best --cjc Traceback (most recent call last): File "complex_helmholtz.py", line 4, in <module> V = FunctionSpace(mesh,"CG",1) File "<decorator-gen-530>", line 2, in FunctionSpace File "/home/cjc1/firedrake-complex/firedrake/src/PyOP2/pyop2/profiling.py", line 60, in wrapper return f(*args, **kwargs) File "/home/cjc1/firedrake-complex/firedrake/src/firedrake/firedrake/functionspace.py", line 118, in FunctionSpace element = make_scalar_element(mesh, family, degree, vfamily, vdegree) File "/home/cjc1/firedrake-complex/firedrake/src/firedrake/firedrake/functionspace.py", line 40, in make_scalar_element mesh.init() File "/home/cjc1/firedrake-complex/firedrake/src/firedrake/firedrake/mesh.py", line 993, in init self._callback(self) File "/home/cjc1/firedrake-complex/firedrake/src/firedrake/firedrake/mesh.py", line 1335, in callback (self.num_vertices(), geometric_dim)) File "firedrake/dmplex.pyx", line 851, in firedrake.dmplex.reordered_coords ValueError: Buffer dtype mismatch, expected 'PetscReal' but got 'complex double' -- Colin Cotter Department of Mathematics Imperial College London I don’t expect a reply to this email from you outside of your normal working hours. http://www.imperial.ac.uk/people/colin.cotter www.cambridge.org/9781107663916

3 2

Re: [firedrake] view of underlying DMPlex?
by Ed Bueler 26 Jul '18

26 Jul '18

Lawrence and David-- The answer below was what I was missing. I use the ipython tab-completion so the _plex member was missing from my view of mesh, which is where I looked first. (Does that make sense?) And the firedrake mailing list digest feature was not helping me communicate, so I turned it off. (Sorry, I did not get Lawrence's answer before replying to David.) Thanks! Ed Hi Ed, > On 26/07/18 09:24, Ham, David A wrote: > > The DM is the one you found. The rest of this is more a PETSc4py > > question than a Firedrake question and I have to say that I?m not the > > PETSc4py expert. However I note that the DM you get from the > > FunctionSpace has a viewFromOptions method. I think that?s what you?re > > after. > As well as what David says here, you can also get at the DMPlex object. > A little explanation. > The DM attached to every FunctionSpace (W.dm in your code) is a > DMShell that describes the dof layout (as well as having a global > vector, it also has a default section (W.dm.getDefaultSection()). > It also has the relevant callback hooks attached to it to do > fieldsplitting, as well as coarsen/refine (if you built with a > MeshHierarchy). > As well as this DMShell, there is also a DMPlex hanging around, that > describes the mesh topology. But this is attached to the mesh object > you made, rather than the FunctionSpace. You can get at it via > mesh._plex: > So: > from firedrake import * > mesh = UnitSquareMesh(...) > V = FunctionSpace(mesh, "Lagrange", 2) > V.dm.view() # DMShell describing dof layout > topology = mesh._plex > # or > topology = V.ufl_domain()._plex > topology.view() # DMPlex describing topology > Thanks, > Lawrence -- Ed Bueler Dept of Math and Stat and Geophysical Institute University of Alaska Fairbanks Fairbanks, AK 99775-6660 301C Chapman

1 0

Re: [firedrake] view of underlying DMPlex?
by Ed Bueler 26 Jul '18

26 Jul '18

David -- > The DM is the one you found. The rest of this is more a PETSc4py question > than a Firedrake question and I have to say that I?m not the PETSc4py > expert. However I note that the DM you get from the FunctionSpace has a > viewFromOptions method. I think that?s what you?re after. Both view() and viewFromOptions() give the same view of a DMShell and the solution Vec. Here is output from W.dm.viewFromOptions('-dm_view') for a P^1 FunctionSpace on a UnitSquareMesh(2,1) in serial: DM Object: 1 MPI processes type: shell Vec Object: 1 MPI processes type: seq 0. 0. 0. 0. 0. 0. Once I find the DMPlex I will use view() or viewFromOptions() on it. So I'm afraid my question stands as is. It is a Firedrake question because I am trying to find the DMPlex, which I understand from Lange et al (2016) is foundational in Firedrake. This is a general PETSc4py pattern, by the way. PETSc is written in plain c > using an object oriented paradigm where the type is prepended to the > function name and the object is passed as the first argument. Since Python > has built in object oriented features, these functions become methods in > PETSc4py, without the type prefix and without explicitly passing the > object. Regrettably, PETSC4py is sparsely documented, so often you have to > ?just know? the conventions for turning the C interface into the Python one. Yup. The PETSc devs (e.g. Sanan) are working through their docs to make things better but docing petsc4py is a job for the future. Thanks! Ed > > Regards, > David > From: <firedrake-bounces(a)imperial.ac.uk> on behalf of Ed Bueler < > elbueler(a)alaska.edu> > Reply-To: firedrake <firedrake(a)imperial.ac.uk> > Date: Thursday, 26 July 2018 at 01:37 > To: firedrake <firedrake(a)imperial.ac.uk> > Subject: [firedrake] view of underlying DMPlex? > Dear Drake -- > Suppose I generate a mesh and FunctionSpace this way (or any other way): > mesh = UnitSquareMesh(2,1) > W = FunctionSpace(mesh, 'Lagrange', 1) > Is there a programmatic and/or option way to view the underlying DMPlex? > Actually, where is the underlying DMPlex? When I look at W.dm it is of > type DMShell. W.dm.view() gives an output including VecView on the > solution, but no real info on the mesh. > An example of what I have in mind is the output from DMViewFromOptions(). > (Is it somewhere in Firedrake/petsc4py?) For a certain 3-triangle mesh > (i.e. not the mesh from UnitSquareMesh() above) the output of > DMViewFromOptions() is: > DM Object: tiny mesh 1 MPI processes > type: plex > tiny mesh in 2 dimensions: > 0-cells: 5 > 1-cells: 7 > 2-cells: 3 > Labels: > depth: 3 strata with value/size (0 (5), 1 (7), 2 (3)) > So I know I have 3 cells with 7 edges and 5 vertices, and the > multi-process view tells me about the partition. > Thanks! I don't want to write my own viewer in Python until I have no > choice. ;-) > Ed -- Ed Bueler Dept of Math and Stat and Geophysical Institute University of Alaska Fairbanks Fairbanks, AK 99775-6660 301C Chapman

2 1

Reminder: postdoc position
by Ham, David A 26 Jul '18

26 Jul '18

Dear Firedrakers, Just a small reminder that the deadline for the Firedrake core developer position is Sunday! If you’d like to come to Imperial to conduct computational science research by working on Firedrake, see the advert here: https://www.imperial.ac.uk/jobs/description/NAT00209/research-associate-cod… Regards, David

1 0

view of underlying DMPlex?
by Ed Bueler 26 Jul '18

26 Jul '18

Dear Drake -- Suppose I generate a mesh and FunctionSpace this way (or any other way): mesh = UnitSquareMesh(2,1) W = FunctionSpace(mesh, 'Lagrange', 1) Is there a programmatic and/or option way to view the underlying DMPlex? Actually, where is the underlying DMPlex? When I look at W.dm it is of type DMShell. W.dm.view() gives an output including VecView on the solution, but no real info on the mesh. An example of what I have in mind is the output from DMViewFromOptions(). (Is it somewhere in Firedrake/petsc4py?) For a certain 3-triangle mesh (i.e. not the mesh from UnitSquareMesh() above) the output of DMViewFromOptions() is: DM Object: tiny mesh 1 MPI processes type: plex tiny mesh in 2 dimensions: 0-cells: 5 1-cells: 7 2-cells: 3 Labels: depth: 3 strata with value/size (0 (5), 1 (7), 2 (3)) So I know I have 3 cells with 7 edges and 5 vertices, and the multi-process view tells me about the partition. Thanks! I don't want to write my own viewer in Python until I have no choice. ;-) Ed -- Ed Bueler Dept of Math and Stat and Geophysical Institute University of Alaska Fairbanks Fairbanks, AK 99775-6660 301C Chapman

3 2

Error indicator
by Amireh Mousavi 22 Jul '18

22 Jul '18

Dear all, How can I use adaptive finite element with definable error indicator in Firedrake? I mean I want to define error indicator according to error analysis of my problem. I am a new user so I'm sorry if my questions are too simple. Thank you for always answering. All the best, Amireh

1 0

Argument comm in _from_cell_list
by Anna Kalogirou (MTH - Staff) 22 Jul '18

22 Jul '18

Dear all, I am trying to run an old code where the mesh is generated based on coordinates imported from file. I then had the following commands: from firedrake.mesh import _from_cell_list dm = _from_cell_list(2, cells, coords) mesh = Mesh(dm) It looks like _from_cell_list now requires a fourth argument ‘comm’. I read the relevant github page but it’s unclear how to define this communicator. Thank you for your help. Best wishes, Anna. Dr Anna Kalogirou Leverhulme Early Career Fellow School of Mathematics University of East Anglia annakalogirou.wixsite.com/home<http://annakalogirou.wixsite.com/home>

2 1

Firedrake failed to install - failed to clone petsc
by Terenin, Alexander 20 Jul '18

20 Jul '18

Hello, I've tried to install Firedrake on macomp001 in my home directory to use on the cluster but I am getting thw error message below. This is with `-v --disable-ssh`. Please let me know how to try to debug this. Thanks Alex Terenin Cloning petsc Failed to clone petsc branch firedrake. Traceback (most recent call last): File "firedrake-install", line 453, in git_clone raise InstallError("Skipping ssh clone because --disable-ssh") __main__.InstallError: Skipping ssh clone because --disable-ssh During handling of the above exception, another exception occurred: Traceback (most recent call last): File "firedrake-install", line 1056, in <module> packages = clone_dependencies("PyOP2") + packages File "firedrake-install", line 493, in clone_dependencies deps.append(git_clone(dep.strip())) File "firedrake-install", line 460, in git_clone check_call(["git", "clone", "-q", git_url(plain_url, "https")]) File "firedrake-install", line 366, in check_call log.debug(subprocess.check_output(arguments, stderr=subprocess.STDOUT, env=env).decode()) File "/usr/lib/python3.5/subprocess.py", line 626, in check_output **kwargs).stdout File "/usr/lib/python3.5/subprocess.py", line 708, in run output=stdout, stderr=stderr) subprocess.CalledProcessError: Command '['git', 'clone', '-q', 'https://github.com/firedrakeproject/petsc.git']' returned non-zero exit status 128

2 3

problems with install slepc
by Francis Poulin 20 Jul '18

20 Jul '18

Hello, On a different note, I needed to do a fresh install of firedrake and most seemed to go well but when I try and update to include slepc, I get the following error. Can anyone say if there is something badly set up on my end? Francis Installing SLEPc. Updating the git repository for slepc No need to rebuild SLEPc Installing slepc4py. Cloning slepc4py Failed to clone slepc4py using ssh, falling back to https. Successfully cloned repository slepc4py. Checking out branch firedrake Successfully checked out branch firedrake Installing slepc4py/ Traceback (most recent call last): File "../bin/firedrake-update", line 1180, in <module> build_and_install_slepc() File "../bin/firedrake-update", line 800, in build_and_install_slepc install("slepc4py/") File "../bin/firedrake-update", line 559, in install run_pip_install(["--ignore-installed", package]) File "../bin/firedrake-update", line 529, in run_pip_install check_call(pipinstall + pipargs) File "../bin/firedrake-update", line 366, in check_call log.debug(subprocess.check_output(arguments, stderr=subprocess.STDOUT, env=env).decode()) File "/usr/lib/python3.6/subprocess.py", line 336, in check_output **kwargs).stdout File "/usr/lib/python3.6/subprocess.py", line 418, in run output=stdout, stderr=stderr) subprocess.CalledProcessError: Command '['/home/fpoulin/software/firedrake/bin/python', '-m', 'pip', 'install', '--no-binary', 'mpi4py', '--no-deps', '-vvv', '--ignore-installed', 'slepc4py/']' returned non-zero exit status 1. ------------------ Francis Poulin Associate Dean, Undergraduate Studies 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

2 2

Errors running Firedrake on multiple nodes
by Aaron Matthew Baier-Reinio 20 Jul '18

20 Jul '18

Hello Firedrake team, I am trying to install Firedrake on a cluster and use mpi to run my Firedrake scripts on multiple nodes. I am running into some issues and was wondering if you would be able to help. The operating system on the cluster is CentOS 7.4. Both Python 3.5 and PETSc 3.9.0 are already loaded on the cluster via the environment modules tool. So, to install I am running the install script as follows: python3 firedrake-install --disable-ssh --no-package-manager --honour-petsc-dir --honour-pythonpath The install seems to go through successfully. The resultant installation seems to work when running on a single node, but I am getting errors when I run on multiple nodes. If I run my script as mpirun -np 8 python3 script.py and all 8 CPUs are located on the same node, then the script seems to run successfully. However, if the CPUs are distributed across multiple nodes, then I always get error messages. The general pattern to these error messages is that I will first get a PyOP2 caching error that looks like this: Traceback (most recent call last): File "/home/ambaierr/firedrake/src/PyOP2/pyop2/caching.py", line 197, in __new__ return cls._cache_lookup(key) File "/home/ambaierr/firedrake/src/PyOP2/pyop2/caching.py", line 205, in _cache_lookup return cls._cache[key] KeyError: ('9d856de341689e4f60b3a9709f8cabd1', False, False, False, <class 'pyop2.sequential.Arg'>, (3,), dtype('float64'), 20, (<class 'pyop2.base.IterationIndex'>, 0), None, Access('WRITE'), <class 'pyop2.sequential.Arg'>, (3,), dtype('float64'), 4, (<class 'pyop2.base.IterationIndex'>, 0), None, Access('READ'), <class 'pyop2.sequential.Arg'>, (1,), dtype('float64'), Access('READ')) This error will then be followed by a bunch of "During handling of the above exception, another exception occurred" errors. I have attached the full output from one of these error messages. Do you have any idea what could be going wrong here? Please let me know if you need more information. Thanks, Aaron

6 11