Project on Quadrature function space in Fenics 2017.1.0

Question

Hi,

I used to be able to project a ufl expression into Quadrature functionspace in Fenics 2016.2.0 but that doesn't seems to be able to work in the 2017.1.0.

For example, the following code:

import dolfin as df

deg = 2
df.parameters["form_compiler"]["quadrature_degree"]=deg
df.parameters["form_compiler"]["representation"] = "uflacs"

mesh = df.UnitCubeMesh(1, 1, 1)
V = df.VectorFunctionSpace(mesh, "Lagrange", 2)
u = df.Function(V)
F = df.Identity(3) + df.grad(u)
C = F.T*F

Quadelem = df.FiniteElement("Quadrature", mesh.ufl_cell(), degree=deg,    quad_scheme="default")
Quad = df.FunctionSpace(mesh, Quadelem)

e1 = df.Function(V)

ls = df.inner(e1, C*e1)
lsprj = df.project(ls, Quad)

will produce the following error:

Moving new file over differing existing file:
src: /home/likchuan/Research/fenicsheartmodel2/strippeddown/jitfailure-ffc_form_c94e290f78247dd89872cf8388ce7a628cfa6fa6  /error.log.cf6b7283b08746b1be2d9e46180c9668
dst: /home/likchuan/Research/fenicsheartmodel2/strippeddown/jitfailure-     ffc_form_c94e290f78247dd89872cf8388ce7a628cfa6fa6/error.log
backup: /home/likchuan/Research/fenicsheartmodel2/strippeddown/jitfailure-ffc_form_c94e290f78247dd89872cf8388ce7a628cfa6fa6/error.log.old
Backup file exists, overwriting.
Compilation failed! Sources, command, and errors have been written to: /home/likchuan/Research/fenicsheartmodel2/strippeddown/jitfailure-   ffc_form_c94e290f78247dd89872cf8388ce7a628cfa6fa6
Traceback (most recent call last):
 File "test.py", line 38, in <module>
  lsprj = df.project(ls, Quad)
 File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/fem/projection.py", line 142, in project
   form_compiler_parameters=form_compiler_parameters)
 File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/fem/assembling.py", line 350, in assemble_system
A_dolfin_form = _create_dolfin_form(A_form, form_compiler_parameters)
 File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/fem/assembling.py", line 67, in _create_dolfin_form
function_spaces=function_spaces)
 File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/fem/form.py", line 89, in __init__
mpi_comm=mesh.mpi_comm())
 File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/compilemodules/jit.py", line 70, in mpi_jit
return local_jit(*args, **kwargs)
File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/compilemodules/jit.py", line 147, in jit
"ffc.jit failed with message:\n%s" % (tb_text,))
File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/cpp/common.py", line 2214, in dolfin_error
  return _common.dolfin_error(location, task, reason)
RuntimeError: 

*** -------------------------------------------------------------------------
*** DOLFIN encountered an error. If you are not able to resolve this issue
*** using the information listed below, you can ask for help at
***
***     fenics-support@googlegroups.com
***
*** Remember to include the error message listed below and, if possible,
*** include a *minimal* running example to reproduce the error.
***
*** -------------------------------------------------------------------------
*** Error:   Unable to perform just-in-time compilation of form.
*** Reason:  ffc.jit failed with message: 
Traceback (most recent call last):
  File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/dolfin/compilemodules/jit.py", line 142, in jit
    result = ffc.jit(ufl_object, parameters=p)
  File "/home/likchuan/.hashdist/bld/profile/rwlceu6wj3pg/lib/python2.7/site-packages/ffc/jitcompiler.py", line 210, in jit
    raise FFCJitError("A directory with files to reproduce the jit build failure has been created.") 
 FFCJitError: A directory with files to reproduce the jit build failure has been created.
 .
 *** Where:   This error was encountered inside jit.py.
 *** Process: 0
 *** 
 *** DOLFIN version: 2017.1.0
 *** Git changeset:  
 *** -------------------------------------------------------------------------

Any help is greatly appreciated!
Thanks!

Comments

I've experienced pretty much the same problem with the latest version. Here's another working example, in case it helps:

#!/usr/bin/python
from dolfin import *
from mshr import *

QUAD_DEG = 2
parameters["form_compiler"]["representation"] = "uflacs"
parameters["form_compiler"]["quadrature_degree"] = QUAD_DEG

r = Rectangle(Point(0.0,0.0), Point(1.0,1.0))
mesh = generate_mesh(r,64)

# used to work in 2016.2, but not in 2017.1
VHE = FiniteElement("Quadrature",mesh.ufl_cell(),\
                    degree=QUAD_DEG,quad_scheme="default")

# works in 2017.1
#VHE = FiniteElement("Lagrange",mesh.ufl_cell(),1)

VH = FunctionSpace(mesh,VHE)

# Hack based on mikael-mortensen's response in the following old thread:
#
# https://fenicsproject.org/qa/1425/derivatives-at-the-quadrature-points
#
# In the real application, Constant(1.0) is replaced with a complicated
# expression (not to be confused with the class Expression) involving
# derivatives of Functions, conditionals, etc. that I am trying to store
# as a history variable at quadrature points.
#
vH = TestFunction(VH)
uH = TrialFunction(VH)
Htemp = Function(VH)
H_proj_res = inner(uH-Constant(1.0),vH)*dx
aH_proj,LH_proj = lhs(H_proj_res), rhs(H_proj_res)
Hproblem = LinearVariationalProblem(aH_proj,LH_proj,Htemp)
Hsolver = LinearVariationalSolver(Hproblem)

Hsolver.solve()

---

Try to set

df.parameters["form_compiler"]["representation"] = "quadrature"

---

Thanks, that works !!

---

Thanks! However, does anyone know whether the issues with derivative() and conditional() discussed in these threads have been addressed in the latest release?

https://bitbucket.org/fenics-project/ufl/issues/70/derivative-of-conditional

https://bitbucket.org/fenics-project/ffc/issues/107/ffc-fails-with-conditionals-which-are

That is the reason I was using uflacs. (The "conditional workaround" didn't quite seem to solve the problems in my application.)

---

I find that using uflacs is much faster and take less memory during compilation. One way I use to get around is to directly specify uflacs during the solving i.e.,

solve(F == 0, w, bcs, J = Jac, form_compiler_parameters={"representation":"uflacs"})