implementing DG for convection dominated problem

Asked by Xujun Zhao

Hi there,

I am trying to solve a typical convection-dominated hyperbolic equation using upwind DG:
du/dt + b*grad(u) = 0
where b is velocity vector. I would like to use explicit time scheme, so define dt as the time step, phi0 is u(n) and phi is u(n+1)

My ufl script is as follows:

s_element = FiniteElement("Discontinuous Lagrange", "triangle", 2)
v_element = VectorElement("Lagrange", "triangle", 2)

# Trial and test function: phi and v
phi = TrialFunction(s_element)
v = TestFunction(s_element)

# velocity vector and facet normal;
phi0 = Coefficient(s_element)
b = Coefficient(v_element)
n = FacetNormal("triangle")
dt = Constant(triangle)

# jump and average quantities
jump_v = jump(v,n)
jump_phi0 = jump(phi0,n)
avg_phi0 = avg(phi0)

# variational form
bn = abs(dot(b('+'), n('+')))
F = (phi*v*dx - phi0*v*dx) \
    - dt*dot(b*phi0, grad(v))*dx \
    + dt*(dot(b('+'), jump_v)*avg_phi0 + 0.5*bn*dot(jump_phi0, jump_v))*dS \
    + dt*dot(b,jump_v)*phi0('-')*ds

a = lhs(F)
L = rhs(F)

// =========================================================
It always gives out some ffc compiling error:
Compiler stage 1 finished in 0.038763 seconds.

Compiler stage 2: Computing intermediate representation
-------------------------------------------------------
  Computing representation of 4 elements
  Computing representation of 4 dofmaps
  Computing representation of integrals
  Computing tensor representation
  Extracting monomial form representation from UFL form
  Transforming monomial form to reference element
  Precomputing integrals on reference element
  Computing quadrature representation
  Transforming cell integral
  Computing quadrature representation
  Transforming interior facet integral (0, 0)
Form argument must be restricted.

*** FFC: Form argument must be restricted.
*** FFC: To get more information about this error, rerun FFC with --verbose.

=====================================================
Seems there are some problems on interior facet integral (the 3rd line of the F definition). I can't figure out what is going wrong. Can anybody give me some hints? Thanks a lot.

XZ

Question information

Language:
English Edit question
Status:
Solved
For:
FEniCS Project Edit question
Assignee:
No assignee Edit question
Solved by:
Xujun Zhao
Solved:
Last query:
Last reply:
Revision history for this message
Marie Rognes (meg-simula) said :
#1

For interior facet integrals, all arguments must be restricted (via '+' or '-'). The error tells you that there is (at least) one argument that is not restricted. In particular, dt is not.

Revision history for this message
Xujun Zhao (xzhao99) said :
#2

Thank you, Marie. That solves my problem!

XZ