- firedrake - imperial.ac.uk

Calculate element-wise mass balance
by Mohammad Sarraf Joshaghani 27 May '20

27 May '20

Hello, I am trying to calculate the element-wise mass balance (i.e., \int_{\Omega_element} div(u) d\Omega) over every cell and project the output to DG0 space. Has this been done or is there a way (e.g., with par_loop() ) to implement this in firedrake? This is how I achieved it in fenics: ------------------------------------------------------------------- *cell_domains = CellFunction('size_t', mesh)cell_domains.set_all(0)* *dx = Measure('dx')[cell_domains]area = [cell.volume() for cell in cells(mesh)]* *for i in range(mesh.num_cells()): Error.vector()[i] = (1./area[i]) * assemble(div(solution)*dx(i))* *-------------------------------------------------------------------* Regards, Sarraf

2 1

Using SLATE to try and implement the MITC4 elements in firedrake.
by Subramanya G 27 May '20

27 May '20

Hi, I am trying to implement some MITC elements in firedrake using an approach similar to the one used by the fenics-shells project, using SLATE and static condensation. This is the symbolic expression that I am trying to compile, M_1[(range(0, 2), range(0, 2))]_0 + M_1[(range(0, 2), 3)]_2 * (M_1[(2, 3)]_3).inv * M_1[(2, 2)]_4 * (M_1[(3, 2)]_5).inv * M_1[(3, range(0, 2))]_6 However when I tried to assemble the condensed matrix, I got the following error, NotImplementedError: Compiling mixed slate expression Is there a way to work around this? If I explicitly write out each of the blocks for example? Thanks! Subramanya. सुब्रह्मण्य .

1 0

Reminder: reading group today at 2pm
by Lawrence Mitchell 25 May '20

25 May '20

Dear all, a reminder that we're kicking off our reading group today at 2pm UK time. It's a UK bank holiday, so if you are not there don't worry, we won't get too far! Recall that the thesis we're reading is: https://www.asc.tuwien.ac.at/~schoeberl/wiki/publications/diss.pdf Zoom link: https://durhamuniversity.zoom.us/j/95827899769?pwd=RU5MRW5IQkNxVHlLQXlFSXpI… Or by hand: Meeting ID: 958 2789 9769 Password: 982161 Cheers, Lawrence

1 1

Re: [firedrake] Firedrake compatible C compiling
by Lawrence Mitchell 25 May '20

25 May '20

Readding firedrake in cc. > On 25 May 2020, at 10:36, Cotter, Colin J <colin.cotter(a)imperial.ac.uk> wrote: > > The thing I was grappling with is that the mpi4py-fft pip package requires an FFTW installation. I've tried with various apt packages but they don't get put in the right place and I'm worried that they might not be compatible somehow. I saw that the Dedalus package requires to build your own FFTW, using the following: > > ./configure --prefix=$BUILD_HOME \ CC=mpiicc CFLAGS="-O3 -xCORE-AVX2" \ CXX=mpiicpc CPPFLAGS="-O3 -xCORE-AVX2" \ F77=mpiifort F90FLAGS="-O3 -xCORE-AVX2" \ MPICC=mpiicc MPICXX=mpiicpc \ LDFLAGS="-lmpi" \ --enable-shared \ --enable-mpi --enable-openmp --enable-threads > make -j > make install > > So the main thing to figure out is how to modify that to build something that will work with my mpi4py communicators. Do you reckon it is ok just to grab the compiler flags etc from the Firedrake PETSc build? Get petsc to build fftw, then point mpi4py-fft at it: export FFTW_INCLUDE_DIR=$PETSC_DIR/$PETSC_ARCH/include export FFTW_LIBRARY_DIR=$PETSC_DIR/$PETSC_ARCH/lib pip install mpi4py-fft ? Lawrence

2 1

Firedrake compatible C compiling
by Cotter, Colin J 25 May '20

25 May '20

Hi folks, I need to build a Firedrake-compatible FFTW. Where can I find all the correct C compiler libraries and mpich etc? all the best --cjc Professor Colin Cotter (he/him) Department of Mathematics 755, Huxley Building Imperial College London South Kensington Campus United Kingdom of Great Britain and Northern Ireland +44 2075943468

3 7

Online reading group
by Lawrence Mitchell 22 May '20

22 May '20

Dear Firedrakers, we have recently finished an online reading group going through Doug Arnold's finite element exterior calculus book. We're now planning the next thing to attack. My proposal, since it intersects with the interests of a number of people who have already been involved, is to read through Joachim Schoeberl's thesis on parameter robust multigrid methods (https://www.asc.tuwien.ac.at/~schoeberl/wiki/publications/diss.pdf) It provides an overview of additive Schwarz theory and multigrid methods, and analysis of convergence and robustness, and is generally a nice thesis on finite element methods and their analysis. Our present schedule has been at 14:00 BST (13:00 UTC) on Mondays, via Zoom. This works well for Europeans, but if enough west-pondians want to join we may consider moving (within the constraints of fitting the working day). The format is informal, and you don't have to have done any prep unless you're presenting (or want to). If you're interested, please let me know (and whether the proposed time is suitable). I'll then try and cook up a schedule and send out zoom links. Cheers, Lawrence

2 2

DG0 elements in 3D mesh ; TensorProductCell error
by Rodrigo Piccinini 18 May '20

18 May '20

Hi all, Could somebody help me with this? I'm having the following error msg when trying to assemble the bilinear form bellow. a = jump(u)*jump(v)*dS petsc_a = assemble(a).M.handle ValueError: interior_facet integral cannot be used with a > TensorProductCell; need to distinguish between vertical and horizontal > contributions. > I'm using a 3D mesh built as an extruded quadrilateral mesh. u and v are in DG0 function space - FunctionSpace(mesh, "DG", 0). Ultimately, I'd like to solve a heterogeneous diffusion equation in 3D. Thanks in advance -- Rodrigo

2 2

3D Elliptic Solvers
by Andrew Hicks 07 May '20

07 May '20

Hello all, I have a simple question. I'm trying to solve an elliptic PDE in 3D, but the regular solver in firedrake is just too slow. What is the best 3D solver in firedrake for this? Perhaps there is an algebraic multigrid solver or something along those lines? Any suggestions would be highly appreciated! Thanks, Andrew Hicks

3 7

Fw: PRISM workshop on best practices for software training and workshops - Thursday, 21st May at 14:00 – 16:00
by Cotter, Colin J 07 May '20

07 May '20

Dear all, Please see a virtual workshop on software training and workshops for finite element software being organised through the Platform for Research and Simulation at Imperial College. all the best --cjc ________________________________ From: Radomska, Anna <a.radomska(a)imperial.ac.uk> Sent: 07 May 2020 10:11 To: patrick.farrell <patrick.farrell(a)maths.ox.ac.uk>; Cotter, Colin J <colin.cotter(a)imperial.ac.uk>; Ham, David A <david.ham(a)imperial.ac.uk>; Kelly, Paul H J <p.kelly(a)imperial.ac.uk>; Cohen, Jeremy <jeremy.cohen(a)imperial.ac.uk>; Moxey, David <D.Moxey(a)exeter.ac.uk>; Gorman, Gerard J <g.gorman(a)imperial.ac.uk>; Piggott, Matthew D <m.d.piggott(a)imperial.ac.uk> Cc: Sherwin, Spencer J <s.sherwin(a)imperial.ac.uk> Subject: PRISM workshop on best practices for software training and workshops - Thursday, 21st May at 14:00 – 16:00 Dear All, Please join us for PRISM workshop on best practices for software training and workshops on Thursday, 21st May at 14:00 – 16:00. In this workshop we wish to share experiences and best practices on the use of workshops and tutorials for the PRISM related software. In light of the currently changes due to the Covid-19 pandemic it is also interesting to ask how our training experiences are likely to change at a broader level including all forms of remote teaching. The format of the on-line workshop will involve a series of 4 short 10-minutes talks followed by small group discussions and a summary session. The programme includes presentations by: * 2.00pm-2.15pm TBC on Firedrake * 2.15pm-2.30pm Spencer Sherwin / David Moxey on Nektar ++ * 2.30pm-2.45pm Matt Piggott / Gerard Gorman on Jupiter notebook * 2.45pm-3pm Katerina Michalickova on Software Carpentry * 3pm-4pm Brainstorming sessions on establishing best practices. Please cascade this information through your local network. To register please sign-up here<https://www.eventbrite.co.uk/e/prism-workshop-on-best-practices-for-softwar…> so you can receive a Microsoft Teams calendar invite which will be distributed closer to the date of the event. With best regards, Anna Anna Radomska, PhD | Programme Project Manager| Department of Mathematics, Imperial College London 732 Huxley Building, 180 Queen’s Gate, London, SW7 2AZ desk: +44 (0) 207 594 2770 | email: a.radomska(a)imperial.ac.uk<mailto:a.radomska@imperial.ac.uk>

1 0

very slow code when using a lot of functions belonging to Real spaces
by Roberto Federico Ausas 04 May '20

04 May '20

Hello, I have recently posted some questions related to the use of Real function spaces for some FSI problems I'm trying to solve. The problem I'm facing now is that the solving phase is consuming absurd amount of time. It does not seem to be the PETSc resolution itself, but the assembly phase, since, once the KSP iterations begin they proceed reasonably fast. The piece of code I believe is relevant is: #------------------------------------------------------------------------ . . . EU = VectorElement("Lagrange", 'triangle', orderU) EP = FiniteElement("Lagrange", 'triangle', orderP) EL = FiniteElement("Real", 'triangle', 0) ES = [] ndofs = 20 for i in range(ndofs): ES.append(FiniteElement("Real", 'triangle', 0)) EM = MixedElement([EU,EP,EL,ES]) W_f = FunctionSpace(mesh, EM) # mesh has a few hundred nodes sol = Function(W_f) auxs = split(sol) u = auxs[0] p = auxs[1] l = auxs[2] s = [] for i in range(ndofs): s.append(auxs[i+3]) auxT = TestFunctions(W_f) v = auxT[0] q = auxT[1] r = auxT[2] d = [] for i in range(ndofs): d.append(auxT[i+3]) uT = u for i in range(ndofs): uT = uT + s[i]*phifluid[i] # ( phifluid[:] are functions belonging to W_f that are zero everywhere except on some boundary nodes ) print(' |> - Fluid part') Res = momentum_residual(uT, p, l, s, v, mesh) # Stokes operator Res += mass_residual(uT, p, l, s, q, r, mesh) # Incompressibility restriction print(' |> - Beam part') aream1 = 1/assemble(Constant(1) * dx(mesh)) for i in range(ndofs): Res += momentum_residual(uT, p, l, s, d[i]*phifluid[i], mesh) force_b = Constant(force_beam(vphi, dphibeam[i])) Res -= d[i]*force_b*aream1*dx # Until this point the code runs reasonably fast # Here starts the bottleneck idsplit = '2' for i in range(ndofs): idsplit = idsplit+','+str(i+3) # idsplit ends up being 2,3,4,5,... sp = { "mat_type": "matfree", "snes_type": "newtonls", "snes_monitor": None, "snes_converged_reason": None, "snes_linesearch_type": "basic", "snes_max_it": 8, "snes_rtol": 1e-10, "ksp_type": "fgmres", "ksp_monitor_true_residual": None, "ksp_max_it": 10, "pc_type": "fieldsplit", "pc_fieldsplit_type": "schur", "pc_fieldsplit_schur_fact_type": "full", "pc_fieldsplit_0_fields": "0,1", "pc_fieldsplit_1_fields": idsplit, "fieldsplit_0_ksp_type": "preonly", "fieldsplit_0_pc_type": "python", "fieldsplit_0_ksp_monitor_true_residual": None, "fieldsplit_0_pc_python_type": "firedrake.AssembledPC", "fieldsplit_0_assembled_pc_type": "lu", "fieldsplit_0_assembled_pc_factor_mat_solver_type": "mumps", "fieldsplit_0_assembled_mat_mumps_icntl_14": 200, "mat_mumps_icntl_14": 200, "fieldsplit_1_ksp_type": "gmres", "fieldsplit_1_ksp_monitor_true_residual": None, "fieldsplit_1_ksp_max_it": ndofs+1, "fieldsplit_1_ksp_convergence_test": "skip", "fieldsplit_1_pc_type": "none" } solve(Res == 0, sol, bcs=bcs, J=Jac, solver_parameters=sp) #------------------------------------------------------------------------ Any tip on what could be going on would be welcome. Regards -- Roberto F. Ausas Instituto de Ciências Matemáticas e de Computação - ICMC/USP São Carlos, SP - Brasil *http://www.lmacc.icmc.usp.br/~ausas* <http://www.lmacc.icmc.usp.br/~ausas>

4 7