January 2021 - firedrake - imperial.ac.uk

Restarting reading group
by Lawrence Mitchell 01 Feb '21

01 Feb '21

Dear all, as we kick off another term of online, if there's enthusiasm, I'd like to restart the reading group we were running last year. To that end, if you'd like to take part, can you select times that work in this doodle https://doodle.com/poll/b7bmt92uhgffaz7m?utm_source=poll&utm_medium=link I will try and pick something that suits most people. Assume we'll run approximately weekly on the selected slot (so don't just pick availability for next week). Times UK, so if you're left of the pond, subtract the appropriate amount. As usual, the choice of papers is up to the participants, you can see some of the things we covered last year here https://github.com/firedrakeproject/firedrake/wiki/Reading-group Cheers, Lawrence

3 4

Divergent nonlinear solve with MUMPS
by Artur Palha 28 Jan '21

28 Jan '21

Hello Everyone, I am trying to solve a (linear) system of equations I am constructing as a result of a specific discretisation of a PDE. As I try to solve with a direct solver (MUMPS) I notice that in some circumstances (mesh size and polynomial degree) I get a solve error: ConvergenceError: Nonlinear solve failed to converge after 0 nonlinear iterations. Reason: DIVERGED_LINEAR_SOLVE After much digging around I managed to (I think) reproduce the error on the mixed Poisson example. For that I simply changed the solver parameters. I add below the full code with the two options of solver parameters (if broken_parameters is set to True then I get the error, if set to False it works fine). The difference is that the broken one uses a direct solver (MUMPS) and the working one uses an iterative solver (gmres). Note that I changed the domain into a periodic domain and I solve for a purely Neumann problem, hence I add the nullspace in the solve step. I do this to make it more similar to what I have in the original system. I found here: https://lists.mcs.anl.gov/pipermail/petsc-users/2017-February/031699.html that MUMPS has issues with zeros along the diagonal. The resulting system contains zeros along the diagonal. Do you think this is the cause of the error? Has anyone experienced this issue? I wanted to start with a direct solver and then switch to an iterative solver. Nevertheless I also get a similar error on the step 0 of the iterative solver (again, not always). Thank you for your help. -artur palha from firedrake import * broken_parameters = True if broken_parameters: solver_parameters = {"ksp_type": "preonly", \ "pc_type": "lu", \ "mat_type": "aij", \ "pc_factor_mat_solver_type": "mumps", \ 'ksp_converged_reason': None, \ 'ksp_monitor_true_residual': None} else: solver_parameters = {"ksp_type": "gmres", \ "mat_type": "aij", \ "pc_factor_mat_solver_type": "mumps", \ 'ksp_converged_reason': None, \ 'ksp_monitor_true_residual': None} mesh = PeriodicUnitSquareMesh(16, 16) RT = FunctionSpace(mesh, "RT", 1) DG = FunctionSpace(mesh, "DG", 0) W = RT * DG sigma, u = TrialFunctions(W) tau, v = TestFunctions(W) x, y = SpatialCoordinate(mesh) f = Function(DG).interpolate(8.0*pi*pi*cos(2.0*pi*x)*cos(2.0*pi*y)) a = (dot(sigma, tau) + div(tau)*u + div(sigma)*v)*dx L = - f*v*dx w = Function(W) u_nullspace = firedrake.VectorSpaceBasis(constant=True) mixed_nullspace = firedrake.MixedVectorSpaceBasis(W, \ [W.sub(0), u_nullspace]) solve(a == L, w, solver_parameters=solver_parameters, nullspace=mixed_nullspace) sigma, u = w.split() File("poisson_mixed.pvd").write(u) try: import matplotlib.pyplot as plt except: warning("Matplotlib not imported") try: fig, axes = plt.subplots() colors = tripcolor(u, axes=axes) fig.colorbar(colors) except Exception as e: warning("Cannot plot figure. Error msg '%s'" % e) try: plt.show() except Exception as e: warning("Cannot show figure. Error msg '%s'" % e)

3 3

Divergent nonlinear solve
by Andrew Hicks 28 Jan '21

28 Jan '21

Dear Firedrakers, I've encountered a problem recently in trying to solve a particular PDE. This problem is elliptic in five dimensions, and I am working on a spherical shell mesh with outer radius 1. In Gmsh I've set the element size factor to be 0.05 and this works just fine, however when I refine the mesh to a size factor of 0.02, this results in the "nonlinear solver failing to converge after 0 nonlinear iterations" with the reason of DIVERGED_LINEAR_SOLVE. I find this a bit confusing especially since the problem I'm solving is linear. Any advice on how to solve this or why it might be happening? I'm using a ksp type of "cg" and a pc type of "gamg." Maybe I should try something different? Andrew Hicks

2 2

Re: [firedrake] High Performance Software Engineer in Extreme Scale Computing and Engineering
by Robert Sawko 27 Jan '21

27 Jan '21

Hello again, I have now obtained a link to job advert on careersportal. https://careersportal.co.uk/jobs/high-performance-software-engineer-in-extr… No login should be necessary for this one. Best wishes, Robert -- Dr Robert Manson-Sawko Research Staff Member, IBM Research Europe Daresbury Laboratory Keckwick Lane, Warrington WA4 4AD United Kingdom -- Email (IBM): RSawko(a)uk.ibm.com Email (STFC): robert.sawko(a)stfc.ac.uk Phone (office): +44 (0) 1925 60 3967 Phone (mobile): +44 759 301 0452 Profile page: http://researcher.watson.ibm.com/researcher/view.php?person=uk-RSawko -- Unless stated otherwise above: IBM United Kingdom Limited - Registered in England and Wales with number 741598. Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

1 0

High Performance Software Engineer in Extreme Scale Computing and Engineering
by Robert Sawko 21 Jan '21

21 Jan '21

Hello Firedrakers, This may be a bit off topic, but I wanted to advertise a job opportunity with our collaborators at the Hartree Centre. https://www.linkedin.com/jobs/view/2365533435/?lgTemp=jobs_jymbii_digest&eB… It would be great and very beneficial for the centre to have people with experience in Firedrake and PETSc there. Best wishes, Robert -- Dr Robert Manson-Sawko Research Staff Member, IBM Research Europe Daresbury Laboratory Keckwick Lane, Warrington WA4 4AD United Kingdom -- Email (IBM): RSawko(a)uk.ibm.com Email (STFC): robert.sawko(a)stfc.ac.uk Phone (office): +44 (0) 1925 60 3967 Phone (mobile): +44 759 301 0452 Profile page: http://researcher.watson.ibm.com/researcher/view.php?person=uk-RSawko --Unless stated otherwise above: IBM United Kingdom Limited - Registered in England and Wales with number 741598. Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

2 2

Fast assembly of time-dependent bilinear forms
by Artur Palha 21 Jan '21

21 Jan '21

Hello Everyone, I have an issue I have been struggling with for some time and I am unable to find a solution for it and I hope someone can help me with this. The issue is that I have a time dependent problem that results in the following weak form (already discretised in time): Find u^{n+1} \in V such that a_1(u^{n+1}, nu) + a_2(u^{n+1}, nu; w^{n+1/2)) = b(u^{n}), for all nu \in V - a_1 is a time-constant bilinear form, meaning that I can assemble it once for all time steps - a_2 is a time-dependent billinear form since it depends on w and this changes with the time step. My first approach was to assemble a_1 (using trial and test functions) in order to obtain a matrix. I could do this once, before the time stepping loop. Then, inside the time loop, I assembled a_2. As I tried to add the two resulting matrices I got an error stating addition is not possible for matrices. I then looked into LinearVariationalProblem and LinearVariationalSolver. My problem with this is that I am unable to understand if by defining the LinearVariational as the sum of a_1 and a_2, both will be kept constant or if only the a_2 part will be updated at each time step. So my question is: is there a standard way of solving this problem? Or do I need to recompute the whole system matrix for every time step (which is what I am doing now)? Thank you for your help. Kind regards, -artur palha

3 3

Difficulty installing firedrake
by Rajdip Nayek 18 Jan '21

18 Jan '21

Hi there, I am a Windows 10 user trying to install firedrake using WSL (Windows Subsystem for Linux) but have been unsuccessful. Could someone guide me on the installation? I keep getting this error message. [image: image.png] Or even better if someone can help me get it installed on google colab; I will share a google colab notebook for that purpose if you like. Thanks in advance. Rajdip -- Rajdip Nayek, PhD Research Associate The Department of Mechanical Engineering The University of Sheffield

2 4