Thanks Colin.

What I was trying to set up is an image, so nx and ny are the number of pixels in each direction (I'm trying a pretty simplistic formulation as you can see...) The function g here will be things like the image gradient function, which I have only the numerical values but not the expression.

So I guess my question is how to get topological information about basis functions, in particular, for a rectangular mesh with degree 1 Lagrange element, are the basis functions arranged in a specific order? (so that I can use something like g.vector().set_local() to set the values).

Many thanks,
-TJ

On 05/05/16 18:34, Colin Cotter wrote:

Dear TJ,
If you want to initialise from a known function then you should do this by interpolating an Expression into a Function. Note that your mesh initialisation is probably not what you intended either, please check the doc-string for RectangleMesh.

all the best

--cjc

On 5 May 2016 at 18:31, Sun, Tianjiao <tianjiao.sun14@imperial.ac.uk> wrote:

Dear team,

I have a beginner question about function initialisation in Firedrake.

I have the follow simple mesh and function space, and I want to
initialise function g with values from numpy arrays (gx, gy). I'm not
sure which way should I align the data in the constructor, i.e. should I
concatenate gx, gy or interleave them?

mesh = RectangleMesh(nx-1,ny-1,nx,ny)
V = VectorFunctionSpace(mesh, "CG", 1)

# g = Function(V, np.concatenate((gx, gy), axis=0))) ???

Many thanks,
-TJ

_______________________________________________
firedrake mailing list
firedrake@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/firedrake

--

http://www.imperial.ac.uk/people/colin.cotter

www.cambridge.org/9781107663916
_______________________________________________
firedrake mailing list
firedrake@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/firedrake