Firedrake update issues/Error in step computations

[mim-GK]

Hi, I am currently working on a topology optimisation problem (code is shown)

from firedrake import *
from firedrake_adjoint import *
import matplotlib.pyplot as plt
import numpy as np
from firedrake import (
    inner,
    dx,
    ds,
    sym,
    Constant
)
from pyadjoint import Control, ReducedFunctional, MinimizationProblem, IPOPTSolver
from firedrake.adjoint import UFLInequalityConstraint

# Create a unit square mesh with 10x10 cells
mesh = UnitSquareMesh(300, 100)
X = SpatialCoordinate(mesh)
F = 2000 #####
traction = Constant([0, -F])
p = Constant(3)
eps = Constant(1e-3)
# Define a function space on the mesh
W = FunctionSpace(mesh, "CG", 1) ####
V = VectorFunctionSpace(mesh, "CG", 2)


# Structure material properties
# Elasticity parameters
E = Constant(1e5)
nu = Constant(0.3)
mu = (E / (2 * (1 + nu))) 
lmbda = (E * nu / ((1 + nu) * (1 - 2 * nu)))

# Define boundary conditions
# Fixed left edge (x = 0)
bc = DirichletBC(V, Constant((0, 0)), 1)  # Clamped at the left edge associated with vector values

def forward(rho):
    v = TestFunction(V)
    u = TrialFunction(V)
    # Define the traction force (applied in the middle of the right edge)

    
    L = inner(traction, v) * ds(2)

    # Variational problem setup
    a = 2*mu*inner(sym(grad(u)), sym(grad(v)))*dx + lmbda*div(u)*div(v)*dx

    # Define E_rho based on SIMP method
    E_rho = eps + (E - eps) * rho**p

    lmbda = nu * E_rho / ((1 + nu) * (1 - 2 * nu))
    mu = E_rho / (2 * (1 + nu))
    
    u = Function(V, name="Displacement")
    solve(a == L, u, bc, annotate=True)
    return u

# Set initial density and define the objective function
Vol = 0.5
rho = interpolate(Constant(Vol), W)
u = forward(rho)
J = assemble(inner(traction, u) * dx)####### could be 1 or dx
m = Control(rho)
Jhat = ReducedFunctional(J, m)
# Volume constraint
volume_constraint = UFLInequalityConstraint((Constant(Vol) - rho) * dx, m)

# Optimization setup
lb, ub = 0.0, 1.0
problem = MinimizationProblem(Jhat, bounds=(lb, ub), constraints=volume_constraint)
parameters = {"acceptable_tol": 1.0e-3, "maximum_iterations": 100}
solver = IPOPTSolver(problem, parameters=parameters)
rho_opt = solver.solve()

# PARAVIEW Results
results = File("newsolution.pvd")
results.parameters["flush_output"] = True
results.parameters["functions_share_mesh"] = True
results.write = (rho_opt, 0.)
results.write = (u, 0.)

where my code is resulting in an iteration but stops after the first one and results in an error as such:
[firedrake-update.log]

FutureWarning: The firedrake_adjoint module is deprecated.

Instead, use the firedrake.adjoint module and explicitly start taping
by calling continue_annotation().
warnings.warn("""The firedrake_adjoint module is deprecated.
/Users/fire/firedrake/src/firedrake/firedrake/interpolation.py:385: FutureWarning: The use of interpolate to perform the numerical interpolation is deprecated.
This feature will be removed very shortly.

Instead, import interpolate from the firedrake.__future__ module to update
the interpolation's behaviour to return the symbolic ufl.Interpolate object associated
with this interpolation.

You can then assemble the resulting object to get the interpolated quantity
of interest. For example,

from firedrake.__future__ import interpolate
...

assemble(interpolate(expr, V))

Alternatively, you can also perform other symbolic operations on the interpolation operator, such as taking
the derivative, and then assemble the resulting form.

warnings.warn("""The use of interpolate to perform the numerical interpolation is deprecated.
** ERROR RETURN ** FROM DMUMPS INFO(1)= -3
** INFO(2)= -1

List of options:

                                Name   Value                # times used
               hessian_approximation = limited-memory            7
                         print_level = 6                         2

---

This program contains Ipopt, a library for large-scale nonlinear optimization.
Ipopt is released as open source code under the Eclipse Public License (EPL).
For more information visit https://github.com/coin-or/Ipopt

---

This is Ipopt version 3.14.16, running with linear solver MUMPS 5.6.2.

Number of nonzeros in equality constraint Jacobian...: 0
Number of nonzeros in inequality constraint Jacobian.: 0
Number of nonzeros in Lagrangian Hessian.............: 0

Hessian approximation will be done in the space of all 36 x variables.

Scaling parameter for objective function = 1.000000e+00
objective scaling factor = 1
No x scaling provided
No c scaling provided
No d scaling provided
Initial values of x sufficiently inside the bounds.
Initial values of s sufficiently inside the bounds.
Total number of variables............................: 36
variables with only lower bounds: 0
variables with lower and upper bounds: 0
variables with only upper bounds: 0
Total number of equality constraints.................: 0
Total number of inequality constraints...............: 0
inequality constraints with only lower bounds: 0
inequality constraints with lower and upper bounds: 0
inequality constraints with only upper bounds: 0

---

*** Update HessianMatrix for Iteration 0:

---

Limited-Memory approximation started; store data at current iterate.

---

*** Summary of Iteration: 0:

---

iter objective inf_pr inf_du lg(mu) ||d|| lg(rg) alpha_du alpha_pr ls
0 3.4740741e-04 0.00e+00 3.27e-03 0.0 0.00e+00 - 0.00e+00 0.00e+00 0

---

*** Beginning Iteration 0 from the following point:

---

Current barrier parameter mu = 1.0000000000000000e+00
Current fraction-to-the-boundary parameter tau = 0.0000000000000000e+00

||curr_x||_inf = 5.0000000000000000e-01
||curr_s||_inf = 0.0000000000000000e+00
||curr_y_c||_inf = 0.0000000000000000e+00
||curr_y_d||_inf = 0.0000000000000000e+00
||curr_z_L||_inf = 0.0000000000000000e+00
||curr_z_U||_inf = 0.0000000000000000e+00
||curr_v_L||_inf = 0.0000000000000000e+00
||curr_v_U||_inf = 0.0000000000000000e+00

***Current NLP Values for Iteration 0:

                               (scaled)                 (unscaled)

Objective...............: 3.4740740740740750e-04 3.4740740740740750e-04
Dual infeasibility......: 3.2717037037037055e-03 3.2717037037037055e-03
Constraint violation....: 0.0000000000000000e+00 0.0000000000000000e+00
Complementarity.........: 0.0000000000000000e+00 0.0000000000000000e+00
Overall NLP error.......: 3.2717037037037055e-03 3.2717037037037055e-03

---

*** Update Barrier Parameter for Iteration 0:

---

Setting mu_max to 0.000000e+00.
Barrier Parameter: 1.000000e-11

---

*** Solving the Primal Dual System for Iteration 0:

---

** ERROR RETURN ** FROM DMUMPS INFO(1)= -3
** INFO(2)= 1
MUMPS used permuting_scaling 0 and pivot_order 0.
scaling will be 77.
Error=-3 returned from MUMPS in Factorization.
Factorization failed with retval = 4
LowRankAugSystemSolver: AugSystemSolver returned retval = 4 for right hand side.
*** Step could not be computed in iteration 0!
ERROR: Problem in step computation, but emergency mode cannot be activated.

Number of Iterations....: 0

                               (scaled)                 (unscaled)

Objective...............: 3.4740740740740750e-04 3.4740740740740750e-04
Dual infeasibility......: 3.2717037037037055e-03 3.2717037037037055e-03
Constraint violation....: 0.0000000000000000e+00 0.0000000000000000e+00
Variable bound violation: 0.0000000000000000e+00 0.0000000000000000e+00
Complementarity.........: 0.0000000000000000e+00 0.0000000000000000e+00
Overall NLP error.......: 3.2717037037037055e-03 3.2717037037037055e-03

Number of objective function evaluations = 1
Number of objective gradient evaluations = 1
Number of equality constraint evaluations = 0
Number of inequality constraint evaluations = 0
Number of equality constraint Jacobian evaluations = 0
Number of inequality constraint Jacobian evaluations = 0
Number of Lagrangian Hessian evaluations = 0
Total seconds in IPOPT = 4.648

EXIT: Error in step computation!
** ERROR RETURN ** FROM DMUMPS INFO(1)= -3
** INFO(2)= -2

I tried updating firedrake to ensure the installation is up-to-date and properly configured. However, this resulted in the following and I have attached the Update log along with it
firedrake-update.log
:

/Users/fire/firedrake/bin/firedrake-update:20: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
from pkg_resources.extern.packaging.version import Version, InvalidVersion
Running /Users/fire/firedrake/bin/firedrake-update
BLAS config is {'BLAS': '/Users/fire/firedrake/src/petsc/default', 'OPENBLAS': '/Users/fire/firedrake/src/petsc/default', 'LDFLAGS': '-L/usr/local/opt/gcc/lib/gcc/14'}
Configuration saved to /Users/fire/firedrake/.configuration.json
Updating the git repository for firedrake
Creating firedrake-update script.
Running ../bin/firedrake-update --no-update-script
Installing command line tools...
Checking brew doctor...
Installing required packages via homebrew...
Required package 'gcc' is already installed via Homebrew.
Required package 'autoconf' is already installed via Homebrew.
Required package 'make' is already installed via Homebrew.
Required package 'automake' is already installed via Homebrew.
Required package 'cmake' is already installed via Homebrew.
Installing required package 'ninja' via Homebrew.
Required package 'libtool' is already installed via Homebrew.
Required package 'boost' is already installed via Homebrew.
BLAS config is {'BLAS': '/Users/fire/firedrake/src/petsc/default', 'OPENBLAS': '/Users/fire/firedrake/src/petsc/default', 'LDFLAGS': '-L/usr/local/opt/gcc/lib/gcc/14'}
Configuration saved to /Users/fire/firedrake/.configuration.json
Finding dependencies of PyOP2

Finding dependencies of firedrake

Updating the git repository for petsc
Removing existing petsc installations

Updating the git repository for loopy
Updating the git repository for ufl
Updating the git repository for fiat
Updating the git repository for FInAT
Updating the git repository for tsfc
Updating the git repository for PyOP2
Updating the git repository for pyadjoint
Updating the git repository for pytest-mpi
Updating the git repository for firedrake
Cleaning petsc/
Installing petsc/
Depending on your platform, PETSc may take an hour or more to build!
Traceback (most recent call last):
File "/Users/fire/firedrake/src/../bin/firedrake-update", line 1960, in
install("petsc/")
File "/Users/fire/firedrake/src/../bin/firedrake-update", line 1043, in install
build_and_install_petsc()
File "/Users/fire/firedrake/src/../bin/firedrake-update", line 1155, in build_and_install_petsc
check_call([python, "./configure", "PETSC_DIR={}".format(petsc_dir), "PETSC_ARCH={}".format(petsc_arch)] + petsc_options)
File "/Users/fire/firedrake/src/../bin/firedrake-update", line 672, in check_call
log.debug(subprocess.check_output(arguments, stderr=subprocess.STDOUT, env=os.environ).decode())
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/Cellar/[email protected]/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/subprocess.py", line 466, in check_output
return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/Cellar/[email protected]/3.11.9/Frameworks/Python.framework/Versions/3.11/lib/python3.11/subprocess.py", line 571, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['/Users/fire/firedrake/bin/python', './configure', 'PETSC_DIR=/Users/fire/firedrake/src/petsc', 'PETSC_ARCH=default', '--download-mumps', '--download-netcdf', '--download-scalapack', '--with-fortran-bindings=0', '--with-c2html=0', '--with-zlib', '--download-hdf5', '--download-hwloc', '--download-pnetcdf', '--download-suitesparse', '--with-shared-libraries=1', '--with-x=0', '--download-pastix', '--CFLAGS=-Wno-implicit-function-declaration', '--download-superlu_dist', '--with-debugging=0', "--download-openblas-make-options='USE_THREAD=0 USE_LOCKING=1 USE_OPENMP=0'", '--download-mpich', '--download-ptscotch', '--download-hwloc-configure-arguments=--disable-opencl', '--download-hypre', '--download-mpich-configure-arguments=--disable-opencl', '--download-openblas', '--download-bison', '--download-metis', '--LDFLAGS=-Wl,-ld_classic']' returned non-zero exit status 1.

Update log saved in /Users/fire/firedrake/firedrake-update.log

As a result, my code is not working at all and is stating the following:

Traceback (most recent call last):
File "/Users/fire/canweend.py", line 1, in
from firedrake import *
File "/Users/fire/firedrake/src/firedrake/firedrake/init.py", line 1, in
import firedrake_configuration
File "/Users/fire/firedrake/src/firedrake/firedrake_configuration/init.py", line 37, in
options["with_parmetis"] = "parmetis" in petsc_packages()
^^^^^^^^^^^^^^^^
File "/Users/fire/firedrake/src/firedrake/firedrake_configuration/init.py", line 31, in petsc_packages
with open(os.path.join(conf["PETSC_DIR"], conf["PETSC_ARCH"], "include", "petscconf.h"), "r") as f:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: '/Users/fire/firedrake/src/petsc/default/include/petscconf.h'

[Ig-dolci]

Unfortunately, there have been issues with mumps when using parallel solvers on macOS. Could you please modify your solver by setting solve(a == L, u, bc, solver_parameters={"pc_factor_mat_solver_type": "superlu_dist"})? Note that you do not need to include annotate=True. To ensure your forward solver is annotated for the automated gradient computations, move from from firedrake_adjoint import * to from firedrake.adjoint import * and continue_annotation() right before your forward solver as shown in this notebook.

[connorjward]

This seems to be an issue with cyipopt. I expect that you could reproduce the issue just by running (with the venv activated) python -c "import cyipopt". I suggest contacting them to report the issue and find a fix.

[mim-GK]

Yes, as expected running python -c "import cyipopt" repeats the same errors unfortonutaly. In relation to my code are there any alternatives that would be compatible with the NumPy version within firedrake?

[Ig-dolci]

Sorry. Could you please clarify "NumPy version within firedrake"? Do you mean another package that solves optimisation problems with inequality constraints?

[mim-GK]

Yes @Ig-dolci, are there any alternative ipopt wrappers that I could use with this problem within firedrake or it has to be cyipopt since the previous outputs suggested only cyipopt?

[Ig-dolci]

I saw some users using the pyrol package. It looks like this pyrol repository is maintainable. Take a look at the examples there and how to install it.