How to debug the PETSC ERROR?

dolfinx
1

I am solving this PDE to get its steady state.
PNP
the variational form is lengthy. When I solve the problem with the NonlinearProblem of petsc, it just complains some unhelpful PETSC ERRORs. In this case, how to debug my code ?
Here are my codes,

# -*- coding:utf-8 -*-
import ufl 
import dolfinx
from dolfinx.fem.petsc import NonlinearProblem
from dolfinx.nls.petsc import NewtonSolver
from petsc4py import PETSc
from mpi4py import MPI
import numpy as np 

epsilon0  = 8.8541878128e-12 #vacuum permittivity, F/m 
e0        = 1.60217663e-19 #electron charge 
N_A       = 6.0221408e+23 #Avogadro constant. 
kB        = 1.380649e-23 #m^2 kg s^(-2) K^(-1) 
pot       = -1.200000 #electrode potential
T         = 2.980000000000e+02 #temperature
pzc       = 0.000000 #potential of zero charge
potential = (pot - pzc)* e0 / kB / T 
delta_RP  = 3.200000000000e-10 #the thickness of Helmholtz layer
epsilon_RP= 4.170000 #the dielectric constant of Helmholtz layer
epsilon_s = 78.500000 #the dielectric constant of bulk solution

n0        = N_A
D0        = 1.0e-9
lambda_d  = np.sqrt(epsilon_s * epsilon0 * kB * T / e0**2 / n0)
Ld        = lambda_d * 60 #the position of bulk
factor_a  = delta_RP*epsilon_s/epsilon_RP/lambda_d
factor_1  = Ld/lambda_d
factor_2  = Ld*lambda_d/D0
ns_ions   = 2

Length_limit = Ld / lambda_d

domain = dolfinx.mesh.create_interval(MPI.COMM_WORLD, 2048, points=(0.0, Length_limit))
V = dolfinx.fem.FunctionSpace(domain, ufl.VectorElement("Lagrange", domain.ufl_cell(), 1, dim=ns_ions+1))

fdim = domain.topology.dim - 1
right_boundary = dolfinx.mesh.locate_entities_boundary(domain, fdim, lambda x: np.isclose(Length_limit, x[0]))
left_boundary = dolfinx.mesh.locate_entities_boundary(domain, fdim, lambda x: np.isclose(0.0, x[0]))

bc_H = dolfinx.fem.dirichletbc(PETSc.ScalarType(1.000000000000e-04), 
                              dolfinx.fem.locate_dofs_topological(V.sub(0), fdim, right_boundary), 
                              V.sub(0))

bc_OH = dolfinx.fem.dirichletbc(PETSc.ScalarType(1.000000000000e-04), 
                              dolfinx.fem.locate_dofs_topological(V.sub(1), fdim, right_boundary), 
                              V.sub(1))

bc_phi = dolfinx.fem.dirichletbc(PETSc.ScalarType(0.000000000000e+00), 
                              dolfinx.fem.locate_dofs_topological(V.sub(2), fdim, right_boundary), 
                              V.sub(2))
u_n = dolfinx.fem.Function(V) #the solution of previous step.
u = dolfinx.fem.Function(V)   #the solution need to solve at the current step.
u_n_H, u_n_OH, u_n_phi = u_n.split()
u_H, u_OH, u_phi = u.split()
v_H, v_OH, v_phi = ufl.TestFunction(V)

#set the initial condition
u_n.sub(0).interpolate(lambda x: 1.000000000000e-04*np.ones(x[0].shape))
u_n.sub(1).interpolate(lambda x: 1.000000000000e-04*np.ones(x[0].shape))
#initial value for the potential 
u_n.sub(2).interpolate(lambda x: 7.944060176975e-03*x[0]+(-4.672608459440e+01))
u_n.x.scatter_forward()


#Define the constants for the flux.
flux_OH_at_left = dolfinx.fem.Constant(domain, 1.0e-5)
dt = dolfinx.fem.Constant(domain, 1.0e-10)

flux_H = 9.311000000000e+00*u_H*(1.0)*ufl.grad(u_n_phi)+9.311000000000e+00*ufl.grad(u_H)
flux_OH = 5.273000000000e+00*u_OH*(-1.0)*ufl.grad(u_n_phi)+5.273000000000e+00*ufl.grad(u_OH)

F_H = u_H*v_H*ufl.dx + dt*factor_1*ufl.dot(flux_H,ufl.grad(v_H))*ufl.dx -u_n_H*v_H*ufl.dx
F_OH = u_OH*v_OH*ufl.dx + dt*factor_1*ufl.dot(flux_OH,ufl.grad(v_OH))*ufl.dx -dt*factor_1*flux_OH_at_left*v_OH*ufl.ds -u_n_OH*v_OH*ufl.dx
F_phi = (u_phi - potential)/factor_a*v_phi*ufl.ds - ufl.dot(ufl.grad(u_phi), ufl.grad(v_phi))*ufl.dx + ((1.0)*u_H+(-1.0)*u_OH)*v_phi*ufl.dx

F = F_H + F_OH + F_phi

problem = NonlinearProblem(F, u, bcs=[bc_H, bc_OH, bc_phi])

solver = NewtonSolver(MPI.COMM_WORLD, problem)
solver.convergence_criterion = "residual"
solver.rtol = 1e-6

ksp = solver.krylov_solver
opts = PETSc.Options()
option_prefix = ksp.getOptionsPrefix()
opts[f"{option_prefix}ksp_type"] = "cg"
opts[f"{option_prefix}pc_type"] = "gamg"
opts[f"{option_prefix}pc_factor_mat_solver_type"] = "mumps"
ksp.setFromOptions()

n, converged = solver.solve(u)

The error is

[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 and https://petsc.org/release/faq/
[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.
Abort(59) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 59) - process 0
2

Use ufl.split(u_n)

3

Oh, sorry, I made the same error again. :sweat_smile: Thanks.

4

Hello, dokken, may I ask a technical question ? The PDE I am solving (as the pictures shows) is the so-called Poisson-Nernst-Planck equation (in 1-d domain). The left boundary condition is the flux of chemical species \text{flux}_i, the right boundary condition is the Dirichlet BC.

In the solving procedure, I have to get the concentrations of chemical species at the left boundary c_i|_{x=0}, then c_i|_{x=0} is passed to an ODE solver to calculate the reaction rates R_i, then \text{flux}_i = -R_i.

The problem is, \text{flux}_i is very large at the early stage. If the time step size is inappropriate, the solution will become negative at the left boundary. This is unphysical, because the concentration can’t be negative. As the reactions proceeds, some chemical species are run out, \text{flux}_i becomes a relatively small and steady state.

So how to determin the time step size in my problem, can you give me any suggestions ? Maybe the PETSc option I used is also a possible reason for the negative solution. When I use “cg” for ksp_type, the solver even does not converge.

ksp = solver.krylov_solver
opts = PETSc.Options()
option_prefix = ksp.getOptionsPrefix()
opts[f"{option_prefix}ksp_type"] = "bcgs"
#opts[f"{option_prefix}pc_type"] = "gamg"
#opts[f"{option_prefix}pc_factor_mat_solver_type"] = "mumps"
ksp.setFromOptions()