How to solve "double free or corruption (!prev)" error in mixed nonlinear problem?

[halfdome]

Hello,
I am attempting to solve a mixed nonlinear problem with conditions enforced by 6 Lagrange multipliers. I implemented the problem successfully in FEniCS legacy. However, after changing the code to FEniCSx I receive the error message shown below.

I am using FEniCSx via conda-forge on Ubuntu 22.04.2. The FEniCSx tutorials are executing without any problems.

This is a MWE that should converge after 0 iterations:

import ufl
from ufl import FiniteElement, MixedElement, TestFunctions, VectorElement, \
    outer, cross, diff, grad, det, tr, dot, dx, split

from petsc4py.PETSc import ScalarType
from mpi4py import MPI
from dolfinx import fem, mesh, nls

# Create mesh
domain = mesh.create_rectangle(MPI.COMM_WORLD, [[- 0.5, - 0.5], [0.5, 0.5]], [20, 20],
                               mesh.CellType.quadrilateral)
x = ufl.SpatialCoordinate(domain)

# Function spaces and functions
V = VectorElement("Lagrange", domain.ufl_cell(), degree=2, dim=3) # Deformation field
Q = FiniteElement("Real", domain.ufl_cell(), 0) # Single Lagrange multiplier
L = MixedElement([Q, Q, Q]) # Field of Lagrange multipliers
W = fem.FunctionSpace(domain, MixedElement([V, L, L]))

w = fem.Function(W)
(u, l, m) = split(w)
#(u, l, m) = w.split()
(d_x, d_l, d_m) = TestFunctions(W)

# Deformed configuration
x_cs = ufl.as_vector([x[0], x[1], 0]) + u

# Multiplicants of the Lagrange multipliers
A = ufl.as_tensor([
    [0, x_cs[2], x_cs[1]],
    [x_cs[2], 0, x_cs[0]],
    [x_cs[1], x_cs[0], 0]
]).T

B = ufl.as_vector([
    x_cs[1] * x_cs[2],
    x_cs[0] * x_cs[2],
    x_cs[0] * x_cs[1]
])

# Weak form functional
Func = (
    dot(l + A * m, d_x) + dot(x_cs, d_l) + dot(B, d_m)
) * dx

Jac = ufl.derivative(Func, w)

# Solving the problem
problem = fem.petsc.NonlinearProblem(Func, w, bcs=[], J=Jac)
solver = nls.petsc.NewtonSolver(MPI.COMM_WORLD, problem)
num_its, converged = solver.solve(w)

Previously I was splitting the Function with (u, l, m) = w.split(), which led to this error, which was reported before:

[0]PETSC ERROR: ------------------------------------------------------------------------
[0]PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation, probably memory access out of range
[0]PETSC ERROR: Try option -start_in_debugger or -on_error_attach_debugger
[0]PETSC ERROR: or see https://petsc.org/release/faq/#valgrind
[0]PETSC ERROR: or try http://valgrind.org on GNU/linux and Apple MacOS to find memory corruption errors
[0]PETSC ERROR: configure using --with-debugging=yes, recompile, link, and run 
[0]PETSC ERROR: to get more information on the crash.
[0]PETSC ERROR: Run with -malloc_debug to check if memory corruption is causing the crash.
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode 59.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------

Thanks to the advice in the post linked below I changed it to (u, l, m) = split(w), which is also how I am using it in my FEniCS legacy code.

Too Many values to unpack (expected 2)

I would suggest you have a look at: dolfinx/demo_stokes.py at main · FEniCS/dolfinx · GitHub which shows you how to set up the variational form for a coupled problem. You can combine this with: Implementation — FEniCSx tutorial or Custom Newton solvers — FEniCSx tutorial to create your fully coupled solver. When you solve non-linear problems, you cannot use TrialFunctions in your variational form. You have to use a function, and use u, p = ufl.split(w), as the residual F has to be differe…

However, now I obtain the following error message:

double free or corruption (!prev)
[mypc:248790] *** Process received signal ***
[mypc:248790] Signal: Aborted (6)
[mypc:248790] Signal code:  (-6)
[mypc:248790] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x42520)[0x7f2ad9642520]
[mypc:248790] [ 1] /lib/x86_64-linux-gnu/libc.so.6(pthread_kill+0x12c)[0x7f2ad9696a7c]
[mypc:248790] [ 2] /lib/x86_64-linux-gnu/libc.so.6(raise+0x16)[0x7f2ad9642476]
[mypc:248790] [ 3] /lib/x86_64-linux-gnu/libc.so.6(abort+0xd3)[0x7f2ad96287f3]
[mypc:248790] [ 4] /lib/x86_64-linux-gnu/libc.so.6(+0x896f6)[0x7f2ad96896f6]
[mypc:248790] [ 5] /lib/x86_64-linux-gnu/libc.so.6(+0xa0d7c)[0x7f2ad96a0d7c]
[mypc:248790] [ 6] /lib/x86_64-linux-gnu/libc.so.6(+0xa2efc)[0x7f2ad96a2efc]
[mypc:248790] [ 7] /lib/x86_64-linux-gnu/libc.so.6(free+0x73)[0x7f2ad96a54d3]
[mypc:248790] [ 8] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0x23b5c8)[0x7f2acce3b5c8]
[mypc:248790] [ 9] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0xda526)[0x7f2acccda526]
[mypc:248790] [10] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0x77ce0)[0x7f2accc77ce0]
[mypc:248790] [11] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x13d2d7)[0x55f8b077c2d7]
[mypc:248790] [12] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyObject_MakeTpCall+0x26b)[0x55f8b0775bdb]
[mypc:248790] [13] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyEval_EvalFrameDefault+0x559e)[0x55f8b0771afe]
[mypc:248790] [14] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyFunction_Vectorcall+0x6f)[0x55f8b077c73f]
[mypc:248790] [15] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyEval_EvalFrameDefault+0x332)[0x55f8b076c892]
[mypc:248790] [16] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x148fa1)[0x55f8b0787fa1]
[mypc:248790] [17] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0x1d1cc8)[0x7f2accdd1cc8]
[mypc:248790] [18] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/../../../libdolfinx.so.0.6(_ZN7dolfinx3nls5petsc12NewtonSolver5solveEP6_p_Vec+0x1a1)[0x7f2accb890f1]
[mypc:248790] [19] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0x1c7eee)[0x7f2accdc7eee]
[mypc:248790] [20] /home/myuser/anaconda3/envs/fenicsx-env/lib/python3.10/site-packages/dolfinx/cpp.cpython-310-x86_64-linux-gnu.so(+0x77ce0)[0x7f2accc77ce0]
[mypc:248790] [21] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x13d2d7)[0x55f8b077c2d7]
[mypc:248790] [22] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyObject_MakeTpCall+0x26b)[0x55f8b0775bdb]
[mypc:248790] [23] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x149134)[0x55f8b0788134]
[mypc:248790] [24] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyEval_EvalFrameDefault+0x4d1d)[0x55f8b077127d]
[mypc:248790] [25] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyFunction_Vectorcall+0x6f)[0x55f8b077c73f]
[mypc:248790] [26] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(_PyEval_EvalFrameDefault+0x735)[0x55f8b076cc95]
[mypc:248790] [27] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x1d5852)[0x55f8b0814852]
[mypc:248790] [28] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(PyEval_EvalCode+0x87)[0x55f8b0814797]
[mypc:248790] [29] /home/myuser/anaconda3/envs/fenicsx-env/bin/python(+0x206eec)[0x55f8b0845eec]
[mypc:248790] *** End of error message ***
Aborted (core dumped)

Do you have any idea what could be wrong with my code? I struggle to understand what causes this error in FEniCSx, while everything is working fine in FEniCS legacy.

Thank you very much in advance for your help!

[dokken]

As far as Im aware, real elements have not been implemented in DOLFINx yet, i.e.

import ufl
from petsc4py.PETSc import ScalarType
from mpi4py import MPI
from dolfinx import fem, mesh

# Create mesh
domain = mesh.create_rectangle(MPI.COMM_WORLD, [[- 0.5, - 0.5], [0.5, 0.5]], [20, 20],
                               mesh.CellType.quadrilateral)

# Function spaces and functions
Q = ufl.FiniteElement("Real", domain.ufl_cell(), 0) # Single Lagrange multiplier
W = fem.FunctionSpace(domain, Q)

will throw an error when trying to generate this function space