PyTrilnos script that demonstrates memory leak even after calling `del`

exAztecOO.py

#! ${PYTHON_EXECUTABLE}
# -*- python -*-

# @HEADER
# ***********************************************************************
#
#          PyTrilinos: Python Interfaces to Trilinos Packages
#                 Copyright (2014) Sandia Corporation
#
# Under the terms of Contract DE-AC04-94AL85000 with Sandia
# Corporation, the U.S. Government retains certain rights in this
# software.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
# 1. Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
#
# 3. Neither the name of the Corporation nor the names of the
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
# LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
# NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# Questions? Contact William F. Spotz (wfspotz@sandia.gov)
#
# ***********************************************************************
# @HEADER

#
# System imports
from __future__ import print_function
from optparse   import *
import sys

#
# Parse the command-line arguments
parser = OptionParser()
parser.add_option("-t", "--testharness", action="store_true",
                  dest="testharness", default=False,
                  help="test local build modules; prevent loading system-installed modules")
parser.add_option("-v", "--verbosity", type="int", dest="verbosity", default=2,
                  help="set the verbosity level [default 2]")
options,args = parser.parse_args()

#
# Under normal usage, simply use 'from PyTrilinos import Epetra'.  For testing,
# we want to be able to control whether we import from the build directory or
# from a system-installed version of PyTrilinos.
# from testutil import fromPyTrilinosImport
# Epetra  = fromPyTrilinosImport('Epetra' , options.testharness)
# Galeri  = fromPyTrilinosImport('Galeri' , options.testharness)
# AztecOO = fromPyTrilinosImport('AztecOO', options.testharness)

from PyTrilinos import Epetra
from PyTrilinos import Galeri
from PyTrilinos import AztecOO

# Main routine, takes a communicator and a solution tolerance
from memory_profiler import profile

# fs = open("memory_main.log", "a+")
# @profile(stream=fs)
def main(comm, tol):

   # Obtain the example matrix from Galeri
   nx = 300
   ny = 300
   galeriList = {"n": nx * ny,  # for Linear map
                 "nx": nx,      # for Laplace2D, which requires nx
                 "ny": ny       # and ny
                 }
   map = Galeri.CreateMap("Linear", comm, galeriList)
   matrix = Galeri.CreateCrsMatrix("Laplace2D", map, galeriList)

   # Allocate and initialize the solution and RHS vector
   exact = Epetra.Vector(map)
   x = Epetra.Vector(map)
   b = Epetra.Vector(map)
   exact.Random()       # Fix exact solution
   x.PutScalar(0.0)     # Fix starting guess for solution
   matrix.Multiply(False, exact, b) # Fix rhs corresponding to exact

   # Build the solver object
   if 0:
      # this does not work on most installations
      problem = Epetra.LinearProblem(matrix, x, b)
      solver = AztecOO.AztecOO(problem)
   else:
      solver = AztecOO.AztecOO(matrix, x, b)

   # Set the solver parameters
   try:
      plist = {"Solver"          : "CG",
               "Precond"         : "Dom_Decomp",
               "Subdomain_Solve" : "ICC",
               "Output"          : 0
               }
      solver.SetParameters(plist,True)
   except AttributeError:
      # If AztecOO and its python wrappers have been compiled without Teuchos
      # support, then then the SetParameters method will not exist, thus raising
      # an AttributeError exception
      print("Teuchos ParameterLists not supported")
      solver.SetAztecOption(AztecOO.AZ_solver, AztecOO.AZ_cg)
      solver.SetAztecOption(AztecOO.AZ_precond, AztecOO.AZ_dom_decomp)
      solver.SetAztecOption(AztecOO.AZ_subdomain_solve, AztecOO.AZ_icc)
      solver.SetAztecOption(AztecOO.AZ_output, 16)

   # Solve the system
   solver.Iterate(galeriList["n"], tol)

   # Compute and return the solution error
   error = Epetra.Vector(map, (x - exact))
   
   del map
   del matrix
   del exact
   del x
   del b
   del solver
   
   return error

if __name__ == "__main__":

   # Call the main routine
   comm  = Epetra.PyComm()
   tol   = 1e-5
   for sweep in range(50):
       error = main(comm, tol)
       del error
       
   error = main(comm, tol)

   # Check the results
   myErrors = 0
   solnError = error.Norm2() / error.Map().NumGlobalElements()
   if solnError > tol:
      myErrors = 1
   totalErrors = comm.SumAll(myErrors)

   # Print results message
   if comm.MyPID() == 0:
      print("solution error norm =", solnError)
      if totalErrors == 0:
         print("End Result: TEST PASSED")
      else:
         print("End Result: TEST FAILED")

   # It is possible to get the following error message with this script:
   #
   # *** The MPI_comm_size() function was called after MPI_FINALIZE was invoked.
   # *** This is disallowed by the MPI standard.
   # *** Your MPI job will now abort.
   #
   # What this means is that the Epetra module (which calls MPI_Finalize() when
   # deleted) is getting deleted before the Epetra map created by the TriUtils
   # module (which lives on due to the error Vector). One workaround is to
   # delete this Vector object before the exit process.
   del error

   # Exit
   sys.exit(totalErrors)