ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

Try adding the option -r quadrature when compiling.

-- 
Anders

On Mon, Mar 24, 2008 at 05:39:22PM -0700, Chong Luo wrote:
> Thank you for the hint. P. Le Tallec also suggested in his book to try those
> (V_h, P_h) that are stable for Stokes problem.
> However, for some reason, the form file with Taylor-Hood elements (P2-P1) does
> not compile.
> I got a "MemoryError".
> 
> The form file NeoHookean.form:
> vector = VectorElement("Lagrange", "tetrahedron", 2)
> scalar = FiniteElement("Lagrange", "tetrahedron", 1)
> mixed = vector+scalar
> v = TestFunction(mixed)
> u = TrialFunction(mixed)
> U0 = Function(mixed)
> ...
> a = aa(u,v)*dx+bb(pr(u),v)*dx+bb(pr(v),u)*dx
> def L1(v):
>     return aa(U0,v)*dx-2*c10*dot(defgrad(vel(U0)),grad(vel(v)))*dx+dot(vel
> (v),f)*dx+dot(vel(v),g)*ds
> def L2(v):
>     return bb(pr(v),U0)*dx+pr(v)*(detF(U0)-1)*dx
> L = L1(v)+L2(v)
> 
> 
> The error:
>     A0 = ReferenceTensor(m, facet0, facet1)
>   File "/home/grad/luoxx039/local/lib/python2.5/site-packages/ffc/compiler/
> representation/tensor/referencetensor.py", line 28, in __init__
>     self.A0 = integrate(monomial, facet0, facet1)
>   File "/home/grad/luoxx039/local/lib/python2.5/site-packages/ffc/compiler/
> representation/tensor/monomialintegration.py", line 59, in integrate
>     A0 = __compute_product(psis, monomial.numeric * weights)
>   File "/home/grad/luoxx039/local/lib/python2.5/site-packages/ffc/compiler/
> representation/tensor/monomialintegration.py", line 242, in __compute_product
>     B = numpy.multiply.outer(B, Psi[ tuple([q] + [b[i] for i in bpart])])
> MemoryError
> 
> On the other hand, the one with (P1-P1) does compile, but the computation is
> not stable. What a life ...
> 
> Also, maybe my NeoHookean.form is too complicated, it takes a lot of time (5-10
> minutes) to compile the form file.
> And the size of NeoHookean.h is 4.8M, so it also takes a lot of time to compile
> main.cpp. Is there any way to speed up the compiling?
> 
> Thank you!
> 
> 
> Best,
> Chong Luo
> 
> ----- Original Message ----
> From: Anders Logg <logg@xxxxxxxxx>
> To: ffc-dev@xxxxxxxxxx
> Sent: Monday, March 24, 2008 6:17:13 PM
> Subject: Re: [FFC-dev] stable finite element for incompressible nonlinear
> elasticity?
> 
> On Sun, Mar 23, 2008 at 03:44:31PM -0700, Chong Luo wrote:
> > Hi,
> >
> > I implemented 3D incompressible nonlinear elasticity (for neo-Hookean
> > materials) using FEniCS.
> > However, the iteration didn't converge.
> > The finite element spaces I chose are Lagrange element of degree 1 in
> > tetrahedron.
> >
> > I saw in P.Le Tallec's book (Handbook of Numerical Analysis, Vol III) that
> some
> > finite element spaces are stable (like Ruas finite element), while others are
> > not.
> > But it seems that Ruas finite element is not available in FEniCS.
> >
> > My question is, what are the possible choices of finite element spaces in
> > FEniCS
> > that would make 3D incompressible nonlinear elasticity stable?
> >
> > Thank you!
> >
> > Best,
> > Chong Luo
> 
> I have absolutely no idea, but why not try Taylor-Hood (P2-P1) or
> adding stabilization. See the Stokes demos in DOLFIN.
>