ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

You need to do mean(uc)[i1] or better still something like

    ucbar = mean(uc)

and then use ucbar instead of uc.

This is because the mean is computed as a projection of the Function,
meaning that a new Function expanded in a new basis (piecewise
constants) is computed by a linear transformation of the coefficients
for the original Function, so you can't take the projection of
individual components (in general).

/Anders


On Wed, Dec 21, 2005 at 03:55:40PM +0100, Johan Hoffman wrote:
> I used it componentwise: mean(uc[i1])
> 
> Is that ok?
> 
> /Johan
> 
> 
> > Yes, it should be supported for vector-valued Functions, but it's not
> > supported for general expressions:
> >
> > 	  mean(f)       # ok
> > 	  mean(f + g)   # not ok
> > 	  mean(grad(f)) # not ok
> >
> > How have you used it in the form? If it's just mean(f) where f is a
> > vector-valued function, then it should work.
> >
> > /Anders
> >
> >
> > On Wed, Dec 21, 2005 at 01:28:46PM +0100, Johan Hoffman wrote:
> >> Anders,
> >>
> >> Is the mean() function in FFC not supported for vector functions? Or am
> >> I
> >> missing something?
> >>
> >> /Johan
> >>
> >>
> >>
> >>
> >> na41>ffc Poisson.form
> >> This is FFC, the FEniCS Form Compiler, version 0.2.4.
> >> For further information, go to http://www/fenics.org/ffc/.
> >> Parsing Poisson.form
> >> Output written to Poisson.py
> >>
> >> Compiling form: (dXa0/dx0)(dXa1/dx0) | ((d/dXa0)vi0)*((d/dXa1)vi1)*dX +
> >> (dXa0/dx 1)(dXa1/dx1) | ((d/dXa0)vi0)*((d/dXa1)vi1)*dX
> >> Finite element of test space:  Lagrange finite element of degree 1 on a
> >> Triangle  with 1 components
> >> Finite element of trial space: Lagrange finite element of degree 1 on a
> >> Triangle  with 1 components
> >> Compiling tensor representation for interior
> >> Hard signatures match for terms 0 and 1, factorizing
> >> Computing reference tensor, this may take some time...
> >> ................................................................................
> >> Reference tensor computed in 0.00739 seconds.
> >> Compiling tensor representation for boundary
> >> Hard signatures match for terms 0 and 1, factorizing
> >>
> >> Compiling form: w0_a0 | vi0*va0*dX
> >> Finite element of test space:  Lagrange finite element of degree 1 on a
> >> Triangle  with 1 components
> >> Finite elements for functions: [Lagrange finite element of degree 1 on a
> >> Triangl e with 1 components]
> >> Compiling tensor representation for interior
> >> Computing reference tensor, this may take some time...
> >> ................................................................................
> >> Reference tensor computed in 0.012 seconds.
> >> Compiling tensor representation for boundary
> >>
> >> Generating output for DOLFIN
> >> Output written to Poisson.h
> >> na41>cd
> >> na41>cd local/dolfin/src/modules/navierstokes/dolfin/
> >> na41>ffc NSEContinuity.form
> >> This is FFC, the FEniCS Form Compiler, version 0.2.4.
> >> For further information, go to http://www/fenics.org/ffc/.
> >> Parsing NSEContinuity.form
> >> Output written to NSEContinuity.py
> >>
> >> *** Error at w0_a69 | va69[a7]
> >> *** Mean values are only supported for Functions.
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> # Copyright (c) 2005 Johan Hoffman (hoffman@xxxxxxxxxxxx)
> >> # Licensed under the GNU GPL Version 2
> >> #
> >> # The contiuity equation for the incompressible
> >> # Navier-Stokes equations using cG(1)cG(1)
> >> #
> >> # Compile this form with FFC: ffc NSEContinuity.form.
> >>
> >> name = "NSEContinuity"
> >> #scalar = FiniteElement("Lagrange", "triangle", 1)
> >> #vector = FiniteElement("Vector Lagrange", "triangle", 1)
> >> #constant_scalar = FiniteElement("Discontinuous Lagrange", "triangle",
> >> 0)
> >> #constant_vector = FiniteElement("Discontinuous vector Lagrange",
> >> "triangle", 0)
> >> scalar = FiniteElement("Lagrange", "tetrahedron", 1)
> >> vector = FiniteElement("Vector Lagrange", "tetrahedron", 1)
> >> #constant_vector = FiniteElement("Discontinuous vector Lagrange",
> >> "tetrahedron", 0)
> >> constant_scalar = FiniteElement("Discontinuous Lagrange", "tetrahedron",
> >> 0)
> >>
> >> q            = BasisFunction(scalar)
> >> p            = BasisFunction(scalar)
> >> #um           = Function(constant_vector)
> >> uc           = Function(vector)
> >> f            = Function(vector)
> >>
> >> delta1       = Function(constant_scalar)        # stabilization
> >>
> >> i0 = Index()
> >> i1 = Index()
> >> i2 = Index()
> >>
> >> a = delta1*dot(grad(q), grad(p))*dx;
> >> L = delta1*dot(grad(q), f)*dx - q*uc[i0].dx(i0)*dx -
> >> delta1*q.dx(i0)*mean(uc[i1])*uc[i0].dx(i1)*dx
> >> #L = delta1*dot(grad(q), f)*dx - q*uc[i0].dx(i0)*dx -
> >> delta1*q.dx(i0)*um[i1]*uc[i0].dx(i1)*dx
> >>
> >> #a = delta1*q.dx(i0)*p.dx(i0)*dx
> >> #L = delta1*q.dx(i0)*f[i0]*dx - delta1*q.dx(i0)*uc[i1]*uc[i0].dx(i1)*dx
> >> -
> >> q*uc[i0].dx(i0)*dx
> >>
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> FFC-dev mailing list
> >> FFC-dev@xxxxxxxxxx
> >> http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> >>
> >
> >
> 
> 

-- 
Anders Logg
Research Assistant Professor
Toyota Technological Institute at Chicago
http://www.tti-c.org/logg/