Dear all, I am using a Fenics code in Python, with Docker, which solves a PDE with mixed finite-elements. The mesh is given by the region between two disks, both centered at the origin, with radii r=1 and R=2,. The code runs with no problem if the mesh resolution is >= 0.05, with the following outp…

[image] mekong: UMFPACK V5.7.1 (Oct 10, 2014): ERROR: out of memory The error is quite clear. Please search the forum for this message. [image] UMFPACK error: out of memory despite system having free memory Hey, this could indeed be a bug of umfpack. As far as I …

[image] dokken: The error is quite clear. Oh sorry I had not seen that! I will look into the links that you provided, thank you!

I tried to use the suggestion given in the first link by using mumps as a liner solver params = {'nonlinear_solver': 'newton', 'newton_solver': { 'linear_solver': 'mumps', } } solver.parameters.update( params ) solver.solve() but it…

First of all. When you say it doesn’t work, what is the traceback? Secondly, no need to install pyadjoint, as it has nothing to do with the problem you are trying to solve.

[image] dokken: First of all. When you say it doesn’t work, what is the traceback? Here it is Solving nonlinear variational problem. Newton iteration 0: r (abs) = 3.917e+02 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-09) Traceback (most recent call last): File "example.py", line …

Well, now you are getting another error, not out of memory. [image] mekong: *** Reason: Solution failed to converge in 0 iterations (PETSc reason DIVERGED_PC_FAILED, residual norm ||r|| = 0.000000e+00). You can again search for this error on the forum ,getting results like: [im…

Following the suggestions in the first link, I used params = {'nonlinear_solver': 'newton', 'newton_solver': { 'linear_solver' : 'umfpack', },} and we are back to an out-of-memory error: Solving nonlinear variational problem. Newton iteration 0: r (abs) = 3.917e+02 (tol = 1.000e-10) r (rel) =…

[image] mekong: Solving nonlinear variational problem. Newton iteration 0: r (abs) = 3.917e+02 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-09) UMFPACK V5.7.1 (Oct 10, 2014): ERROR: out of memory As you can see “lu”-> “umfpack”, so you need to use “mumps” or “superlu”. As you haven…

[image] dokken: so you need to use “mumps” or “superlu”. I tried mumps above and it gave the error I reported . With superlu I get Solving nonlinear variational problem. Newton iteration 0: r (abs) = 3.917e+02 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-09) Killed

[image] dokken: As you haven’t provided any code, I cannot give you any further help. I made a minimal working example for you, mwe.py: from __future__ import print_function from fenics import * from mshr import * import numpy as np import argparse import ufl as ufl #### function definition…

Cannot run Fenics code with fine mesh (memory issue?)

General

dokken October 28, 2024, 4:48pm 2

The error is quite clear.
Please search the forum for this message.

Topic		Replies	Views
PETSc Krylov Solver RuntimeError with fine mesh I/O	21	5678	October 28, 2022
UMFPACK error: out of memory despite system having free memory	14	6640	September 26, 2022
Large Mesh Error PETSc Error 76 - Adjoint Calculation dolfin-adjoint	34	6195	September 5, 2020
Dolfinx petsc error for large mesh Errors	44	1437	August 31, 2023
Residual error for big mess (\|\|r\|\| = 0.000000e+00) mesh	4	485	April 29, 2020

Cannot run Fenics code with fine mesh (memory issue?)

Related topics