ffc team mailing list archive

[Kristian Oelgaard <k.b.oelgaard@xxxxxxxxxx>]

Quoting Patrick Riesen <priesen@xxxxxxxxxxxxxxx>:

> hi
> 
> i'm under way of implementing a powerlaw stokes with newton iteration in 
> a mixed formulation (2d triangles).
> compiling my forms with ffc runs into the following:
> 
> a)
> using normal tensor representation the compile fails in 
> monomialintegration.py when computing the terms with:
> 
> ValueError: dimensions too large.

For very complex forms, this is not unusual for tensor representation.

> b)
> using quadrature representation the compile takes very long (>20min) and 
> cpu and memory are heavy and increasing, so i aborted.

What stage of the compilation process does the compiler reach?
Compile times up to 1 hour has been reported for some forms so you could try to
let it finish.
Alternatively, see if you can simplify your form somehow.
We're currently working on a transition to UFL which will hopefully speed up the
compile process although this is still at an early stage.

Kristian
 
> 
> i don't understand the error from (a), what does it mean?
> if (a) fails, does it necessarily mean that my forms have an error and 
> (b) will fail as well, or should i try running (b) to the end?
> 
> 
> thanks for your help
> 
> patrick
> 
> _______________________________________________
> FFC-dev mailing list
> FFC-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/ffc-dev
>