May 2017 - firedrake - imperial.ac.uk

Re: [firedrake] Solver error -- Provided residual is not a linear form
by Andrew McRae 15 May '17

15 May '17

Unless I'm being dumb, this really should work as you've written it (using assign/interpolate definitely *won't* do what you want). Can you send a fuller version of your code as I suspect the error is caused by something else? Andrew On 15 May 2017 at 09:48, Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > Hi Colin, > > Thanks, indeed I had forgotten a trial function in my residual. > > Yes I do need the the u-dependence in the maximum during the solve. If I > substitute the definition of maximum into the residual, i.e. write > F = v*(u + 0.5*(u + abs(u)))*dx > gives the error > ufl.log.UFLException: UFL conditions cannot be evaluated as bool in a > Python context. > > Will it work if I use assign instead of interpolate? > > Best, Anna. > > On 15 May 2017, at 07:59, Colin Cotter <colin.cotter(a)imperial.ac.uk> > wrote: > > Hi Anna, > Are you intending that the u-dependence in maximum should be maintained > during the solve? After doing the interpolation, there will be no symbolic > record of where maximum's values came from. > > That's probably not the cause of the error message though. It says that > the form is not a linear form. This might mean that you accidentally > included some TrialFunction (which turns it into a bilinear form) or > perhaps forgot a factor of dx or dS somewhere (which stops it from being a > form). > > all the best > --cjc > > On 14 May 2017 at 21:30, Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > >> Following up on my previous email, I have found a workaround to those >> issues, but now I have a problem with the solvers: >> >> Traceback (most recent call last): >> File "/Users/matak/Downloads/Inequality-constraint-Rockafellar/solvers.py", >> line 32, in solver_F >> F_problem = NonlinearVariationalProblem(F_solve, w)#, aP=Ap) >> File "/Users/matak/firedrake/src/firedrake/firedrake/variational_solver.py", >> line 47, in __init__ >> raise ValueError("Provided residual is not a linear form") >> ValueError: Provided residual is not a linear form >> >> There two issues here: >> 1. I cannot give a preconditioning matrix aP -- is this not possible for >> nonlinear problems? >> 2. I suspect the error above is because I have functions that depend on >> the solution, and I don’t know if that is acceptable in firedrake. I have >> something like the following: >> >> u = Function(V) >> v = TestFunction(V) >> maximum = interpolate(0.5*(u + abs(u)), V) >> F = v*(u + maximum)*dx >> >> Best wishes, >> >> Anna. >> > > > > -- > http://www.imperial.ac.uk/people/colin.cotter > > www.cambridge.org/9781107663916 > > > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > >

2 1

Re: [firedrake] Solver error -- Provided residual is not a linear form
by Colin Cotter 15 May '17

15 May '17

Hi Anna, Are you intending that the u-dependence in maximum should be maintained during the solve? After doing the interpolation, there will be no symbolic record of where maximum's values came from. That's probably not the cause of the error message though. It says that the form is not a linear form. This might mean that you accidentally included some TrialFunction (which turns it into a bilinear form) or perhaps forgot a factor of dx or dS somewhere (which stops it from being a form). all the best --cjc On 14 May 2017 at 21:30, Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > Following up on my previous email, I have found a workaround to those > issues, but now I have a problem with the solvers: > > Traceback (most recent call last): > File "/Users/matak/Downloads/Inequality-constraint-Rockafellar/solvers.py", > line 32, in solver_F > F_problem = NonlinearVariationalProblem(F_solve, w)#, aP=Ap) > File "/Users/matak/firedrake/src/firedrake/firedrake/variational_solver.py", > line 47, in __init__ > raise ValueError("Provided residual is not a linear form") > ValueError: Provided residual is not a linear form > > There two issues here: > 1. I cannot give a preconditioning matrix aP -- is this not possible for > nonlinear problems? > 2. I suspect the error above is because I have functions that depend on > the solution, and I don’t know if that is acceptable in firedrake. I have > something like the following: > > u = Function(V) > v = TestFunction(V) > maximum = interpolate(0.5*(u + abs(u)), V) > F = v*(u + maximum)*dx > > Best wishes, > > Anna. > -- http://www.imperial.ac.uk/people/colin.cotter www.cambridge.org/9781107663916

2 1

Solver error -- Provided residual is not a linear form
by Anna Kalogirou 15 May '17

15 May '17

Following up on my previous email, I have found a workaround to those issues, but now I have a problem with the solvers: Traceback (most recent call last): File "/Users/matak/Downloads/Inequality-constraint-Rockafellar/solvers.py", line 32, in solver_F F_problem = NonlinearVariationalProblem(F_solve, w)#, aP=Ap) File "/Users/matak/firedrake/src/firedrake/firedrake/variational_solver.py", line 47, in __init__ raise ValueError("Provided residual is not a linear form") ValueError: Provided residual is not a linear form There two issues here: 1. I cannot give a preconditioning matrix aP -- is this not possible for nonlinear problems? 2. I suspect the error above is because I have functions that depend on the solution, and I don’t know if that is acceptable in firedrake. I have something like the following: u = Function(V) v = TestFunction(V) maximum = interpolate(0.5*(u + abs(u)), V) F = v*(u + maximum)*dx Best wishes, Anna.

2 1

Re: [firedrake] Interpolation in RealFunctionSpace
by David Ham 15 May '17

15 May '17

On Sun, 14 May 2017 at 21:12 Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > Dear team, > > I am facing a few problems with the real function space, again. I can’t > seem to be able to interpolate on a real function, either with a C string > expression or UFL. The following sample code > > from firedrake import * > > mesh = IntervalMesh(10, 1.0) > > V = FunctionSpace(mesh, "CG", 1) > V_R = FunctionSpace(mesh, "R", 0) > > h = Function(V) > Z = Function(V_R) > > x = SpatialCoordinate(V.mesh()) > > h = interpolate(x[0], V) > Z = interpolate(Expression("1.0"), V_R) > Z = interpolate(1.0, V_R) > > gives the error > > Traceback (most recent call last): > File "interpolation_reals.py", line 17, in <module> > Z = interpolate(Expression("1.0"), V_R) > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 40, in interpolate > return Interpolator(expr, V, subset=subset).interpolate() > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 60, in __init__ > self.callable = make_interpolator(expr, V, subset) > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 129, in make_interpolator > subset)) > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 189, in _interpolator > args.append(dat(op2.WRITE, V.cell_node_map()[op2.i[0]])) > TypeError: 'NoneType' object has no attribute '__getitem__' > > I assume this functionality hasn’t been added for the real function space? > What does interpolation into R even mean? In general interpolating an expression into a FunctionSpace means evaluating the nodes of the FunctionSpace on the expression. Thus far we haven't actually even defined a Node for R (I guess integral over the domain normalised by domain volume might be it...) > Also, I noticed that when the UFL expression doesn’t include a coordinate > x, e.g. in h = interpolate(1.0, V), I get the error: > > Traceback (most recent call last): > File "interpolation_reals.py", line 23, in <module> > h = interpolate(1.0, V) > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 40, in interpolate > return Interpolator(expr, V, subset=subset).interpolate() > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 60, in __init__ > self.callable = make_interpolator(expr, V, subset) > File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", > line 94, in make_interpolator > assert isinstance(expr, ufl.classes.Expr) > AssertionError > > Actually, as the error says, this error is because you need to pass a UFL expression into interpolate and 1.0 isn't a UFL expression. interpolate(Constant(1.0), V) would work fine. > Furthermore, I’m having trouble the exponential UFL expression, e.g. > > exp_h = interpolate(exp(x[0]), V) > TypeError: nb_float should return float object > > I am unable to reproduce this problem. To be precise, ran the following script. It executed without error and printed what appears to be the right answer: from firedrake import * mesh = IntervalMesh(10, 1.0) V = FunctionSpace(mesh, "CG", 1) x = SpatialCoordinate(V.mesh()) exp_h = interpolate(exp(x[0]), V) print exp_h.dat.data_ro Thank you very much for your help, > > Anna. > > P.S. At the start of every simulation, the following sentence is printed > twice and delays the simulation for a couple of seconds: > [warn] kq_init: detected broken kqueue; not using.: No such file or > directory > What does it mean and can we get rid of it? > The error message is some Mac issue. I see it too but I have no idea what causes it. I am not convinced that it causes any delay at all. It's simply the case that loading firedrake is not completely instantaneous. Regards, David -- Dr David Ham Department of Mathematics Imperial College London

2 1

Interpolation in RealFunctionSpace
by Anna Kalogirou 14 May '17

14 May '17

Dear team, I am facing a few problems with the real function space, again. I can’t seem to be able to interpolate on a real function, either with a C string expression or UFL. The following sample code from firedrake import * mesh = IntervalMesh(10, 1.0) V = FunctionSpace(mesh, "CG", 1) V_R = FunctionSpace(mesh, "R", 0) h = Function(V) Z = Function(V_R) x = SpatialCoordinate(V.mesh()) h = interpolate(x[0], V) Z = interpolate(Expression("1.0"), V_R) Z = interpolate(1.0, V_R) gives the error Traceback (most recent call last): File "interpolation_reals.py", line 17, in <module> Z = interpolate(Expression("1.0"), V_R) File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 40, in interpolate return Interpolator(expr, V, subset=subset).interpolate() File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 60, in __init__ self.callable = make_interpolator(expr, V, subset) File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 129, in make_interpolator subset)) File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 189, in _interpolator args.append(dat(op2.WRITE, V.cell_node_map()[op2.i[0]])) TypeError: 'NoneType' object has no attribute '__getitem__' I assume this functionality hasn’t been added for the real function space? Also, I noticed that when the UFL expression doesn’t include a coordinate x, e.g. in h = interpolate(1.0, V), I get the error: Traceback (most recent call last): File "interpolation_reals.py", line 23, in <module> h = interpolate(1.0, V) File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 40, in interpolate return Interpolator(expr, V, subset=subset).interpolate() File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 60, in __init__ self.callable = make_interpolator(expr, V, subset) File "/Users/matak/firedrake/src/firedrake/firedrake/interpolation.py", line 94, in make_interpolator assert isinstance(expr, ufl.classes.Expr) AssertionError Furthermore, I’m having trouble the exponential UFL expression, e.g. exp_h = interpolate(exp(x[0]), V) TypeError: nb_float should return float object Thank you very much for your help, Anna. P.S. At the start of every simulation, the following sentence is printed twice and delays the simulation for a couple of seconds: [warn] kq_init: detected broken kqueue; not using.: No such file or directory What does it mean and can we get rid of it?

1 0

Simulation time sensitive to form of equations
by Anna Kalogirou 11 May '17

11 May '17

Dear all, I have noticed that when I divide one of my equations in a mixed system by the time step dt, the simulation time drops by more than half -- from 7m13s to 3m7s for the same simulation. Is there an explanation why this might be happening? Regards, Anna.

2 1

Re: [firedrake] Simulation time sensitive to form of equations
by David Ham 11 May '17

11 May '17

Yes, you're probably improving the conditioning of the operator by making the mass matrix in the d/dt term more important. Try switching on the monitors in the solver and see if the number of solver iterations has dropped. Regards, David On Thu, 11 May 2017 at 10:13 Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > Dear all, > > I have noticed that when I divide one of my equations in a mixed system by > the time step dt, the simulation time drops by more than half -- from 7m13s > to 3m7s for the same simulation. Is there an explanation why this might be > happening? > > Regards, > > Anna. > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > -- Dr David Ham Department of Mathematics Imperial College London

1 0

Re: [firedrake] Lagrange multipliers
by David Ham 10 May '17

10 May '17

Hi All, Real function spaces have now landed in master. Please attempt your simulations using current master Firedrake and report any bugs you find. Regards, David On Sat, 29 Apr 2017 at 21:19 Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > I still get the same "Mismatching function spaces” error, unfortunately. > The code I am running is basically the same as James’ but with a few more > lines added at the end. > > > > > On 29 Apr 2017, at 16:50, James Jackaman <wb816921(a)live.reading.ac.uk> > wrote: > > Thank you Anna, this was really useful. And thank you David (attached > message), the current Firedrake branch and real-function-space are now > working for me. > > On 25/04/17 21:58, Anna Kalogirou wrote: > > Dear James, > > I havenâ€™t updated firedrake on my old laptop for a few months (I think > since December). The version I have at the moment is: > > "fiat": "5bdea9eba05571791b05e763f69e318ccd57006eâ€ , > "COFFEE": "0a2a19a377aeba3b9bcdb15952fe6c0da99714a2â€ , > "ufl": "013e6929bc56fc03e019a4f0f3dfd9ceee2eb199â€ , > "firedrake": "e051f502a94edf68ec9e062d94d8a02fb5cc6624â€ , > "tsfc": "e05d750d9daeea53c7a4bc8c71e8d5fcc406b2bbâ€ , > "petsc4py": "7770e68fec1953a7315ed187986248af7b4dfb6aâ€ , > "petsc": "afae6a5591f0bed801d208622b499bf7ea2fde7eâ€ , > "PyOP2": "dd9d40ab7b28c07b7d9547dbb8830f5a87f0d9dfâ€ , > > Best, Anna. > > On 25 Apr 2017, at 16:44, James Jackaman <wb816921(a)live.reading.ac.uk> > wrote: > > Dear Anna, > > Thank you for looking into my issue further. While the error is being > investigated by the Firedrake team would you be able to let me know exactly > which version of Firedrake is working on your laptop (via firedrake-zenodo > )? > > Thanks again, > > James > On 24/04/17 11:34, Anna Kalogirou wrote: > > Dear James (and firedrake team), > > I get the same errors as you do when I run your neumann-poisson.py code on > my desktop machine, on which I installed firedrake last week. The fact that > it runs on my laptop which has a older firedrake version means that > something that changed recently is causing the error. I believe it is > related to COFFEE, please see .log and .err files attached. > > Best, Anna. > > On 23 Apr 2017, at 18:25, Anna Kalogirou <A.Kalogirou(a)leeds.ac.uk> wrote: > > Dear James, > > Your neumann-poisson.py code also works on my laptop. IÃ¢â‚¬â„¢ll leave it > to the firedrake developers to help you with your errors. > > > Best wishes, > > Anna. > > > On 21 Apr 2017, at 16:41, James Jackaman <wb816921(a)live.reading.ac.uk> > wrote: > > Hi Anna, > > Thanks :), this fixes the real_solver.py test, but unfortunately on this > branch of Firedrake the neumann-poisson.py test does not compile (output > attached). Is this a bug or am I doing something silly? > > All the best, > > James > On 21/04/17 13:47, Anna Kalogirou wrote: > > Hi James, > > Your test file for the RealFunctionSpace returns u=1 when I run it. You > need to install firedrake and use the correct branch of > real-function-space, i.e. > > python firedrake-install --package-branch firedrake real-function-space > > Regards, Anna. > > On 19 Apr 2017, at 15:09, James Jackaman <wb816921(a)live.reading.ac.uk> > wrote: > > Dear all, > > > I'm currently working on an Firedrake implementation which utilises > Lagrange multipliers, in the spirit of the Poisson equation with pure > Neumann boundary conditions where the average value of the solution is > constrained by a Lagrange multiplier. > > I have attempted to translate the Fenics demo, see > https://fenicsproject.org/olddocs/dolfin/1.6.0/python/demo/documented/neuma…, > into Firedrake with no success, see neumann-poisson.py in the attached > gist. > > I believe there may be a problem with the function space of real constants > *FunctionSpace(mesh,"R",0)*, see real_solver.py in the attached gist > which returns u=0 instead of u=1. > > How can I modify my example codes in the gist such that they behave as > expected, or alternatively how can I implement the Poisson equation with > pure Neumann boundary conditions using Lagrange multipliers in Firedrake? > > > Gist: https://gist.github.com/j130792/64d12e84cf5ede2b327d1859a36dc36d > > > Thanks for your help, > > James > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > > > _______________________________________________ > firedrake mailing listfiredrake@imperial.ac.ukhttps://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > <real-function-space-neumann-output.txt> > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > > _______________________________________________ > firedrake mailing listfiredrake@imperial.ac.ukhttps://mailman.ic.ac.uk/mailman/listinfo/firedrake > > _______________________________________________ > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > > _______________________________________________ > firedrake mailing listfiredrake@imperial.ac.ukhttps://mailman.ic.ac.uk/mailman/listinfo/firedrake > > > <Attached Message.eml>_______________________________________________ > > > firedrake mailing list > firedrake(a)imperial.ac.uk > https://mailman.ic.ac.uk/mailman/listinfo/firedrake > > -- Dr David Ham Department of Mathematics Imperial College London

3 3

imposing Dirichlet BCS
by Francis Poulin 10 May '17

10 May '17

Hello, We are trying to translate part of the mixed-poisson Fenics demo into Firedrake. I know that there is already a Firedrake demo that does this but for the special case of a square. I am hoping we can do this more generally. Below are the lines of code from Fenics. Can someone suggest how this could be implemented in Firedrake? Cheers, Francis https://fenics.readthedocs.io/projects/dolfin/en/latest/demos/mixed-poisson… # Define function G such that G \cdot n = g class BoundarySource(Expression): def __init__(self, mesh, **kwargs): self.mesh = mesh def eval_cell(self, values, x, ufc_cell): cell = Cell(self.mesh, ufc_cell.index) n = cell.normal(ufc_cell.local_facet) g = sin(5*x[0]) values[0] = g*n[0] values[1] = g*n[1] def value_shape(self): return (2,) G = BoundarySource(mesh, degree=2) # Define essential boundary def boundary(x): return x[1] < DOLFIN_EPS or x[1] > 1.0 - DOLFIN_EPS bc = DirichletBC(W.sub(0), G, boundary) ------------------ Francis Poulin 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

1 0

Compilation error
by Floriane Gidel [RPG] 05 May '17

05 May '17

Dear all, I try to solve a weak form with the NonlinearVariationalSolver but get the error below and couldn't figure out what causes this error. I get the same error when solving another weak form with the LinearVariationalSolver, but my third weak form (also solved with LinearVariationalSolver) works. I have updated firedrake but it didn't solve the issue. Does it ring a bell to someone ? If it is not an "obvious" error, I can also send my code. Thanks a lot, Floriane Traceback (most recent call last): File "main.py", line 366, in <module> h_solver.solve() File "/Users/mmfg/firedrake/src/firedrake/firedrake/variational_solver.py", line 220, in solve self.snes.solve(None, v) File "PETSc/SNES.pyx", line 537, in petsc4py.PETSc.SNES.solve (src/petsc4py.PETSc.c:172544) File "PETSc/petscsnes.pxi", line 261, in petsc4py.PETSc.SNES_Function (src/petsc4py.PETSc.c:33927) File "/Users/mmfg/firedrake/src/firedrake/firedrake/solving_utils.py", line 433, in form_function with ctx._F.dat.vec_ro as v: File "/usr/local/Cellar/python/2.7.13/Frameworks/Python.framework/Versions/2.7/lib/python2.7/contextlib.py", line 17, in __enter__ return self.gen.next() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/petsc_base.py", line 395, in vecscatter with acc(d) as v: File "/usr/local/Cellar/python/2.7.13/Frameworks/Python.framework/Versions/2.7/lib/python2.7/contextlib.py", line 17, in __enter__ return self.gen.next() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/petsc_base.py", line 328, in vec_context self._force_evaluation(read=True, write=not readonly) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/base.py", line 1681, in _force_evaluation _trace.evaluate(reads, writes) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/base.py", line 193, in evaluate comp._run() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/base.py", line 4085, in _run return self.compute() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/base.py", line 4127, in compute fun = self._jitmodule File "/Users/mmfg/firedrake/src/PyOP2/pyop2/utils.py", line 64, in __get__ obj.__dict__[self.__name__] = result = self.fget(obj) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/sequential.py", line 805, in _jitmodule pass_layer_arg=self._pass_layer_arg) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/caching.py", line 200, in __new__ obj = make_obj() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/caching.py", line 190, in make_obj obj.__init__(*args, **kwargs) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/sequential.py", line 635, in __init__ self.compile() File "/Users/mmfg/firedrake/src/PyOP2/pyop2/sequential.py", line 716, in compile comm=self.comm) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/compilation.py", line 385, in load dll = compiler.get_so(src, extension) File "/Users/mmfg/firedrake/src/PyOP2/pyop2/compilation.py", line 227, in get_so Compile errors in %s""" % (e.cmd, e.returncode, logfile, errfile)) pyop2.exceptions.CompilationError: Command "['mpicc', '-std=c99', '-fPIC', '-Wall', '-framework', 'Accelerate', '-march=native', '-O3', '-I/Users/mmfg/firedrake/lib/python2.7/site-packages/petsc/include', '-I/Users/mmfg/firedrake/src/PyOP2/pyop2', '-msse', '-o', '/tmp/pyop2-cache-uid1010350243/c993a8a607879888f3be0da8f3d4009d_p95228.so.tmp', '/tmp/pyop2-cache-uid1010350243/c993a8a607879888f3be0da8f3d4009d_p95228.c', '-dynamiclib', '-L/Users/mmfg/firedrake/lib/python2.7/site-packages/petsc/lib', '-Wl,-rpath,/Users/mmfg/firedrake/lib/python2.7/site-packages/petsc/lib', '-lpetsc', '-lm']" return error status 1. Unable to compile code Compile log in /tmp/pyop2-cache-uid1010350243/c993a8a607879888f3be0da8f3d4009d_p95228.log Compile errors in /tmp/pyop2-cache-uid1010350243/c993a8a607879888f3be0da8f3d4009d_p95228.err

2 4