GridFunction to numpy array to new GridFunction

Hi,

the fes can be a product space consisting of different component spaces.

the fesir can be a vectorial IRSpace, with all components having the same points.

When going from the polynomials to the points, you form one vector coefficient function as argument for the gfuir.Interpolate().
When going back to the polynomial space, you can extract components of the ir-gridfunction: gfuir[2:4]

Joachim

Hi Joachim,

Thank you for your response. Can you please elaborate on vectorial IRSpace? Do you mean:

fesir = IntegrationRuleSpace([fesir_u, fesir_p])?

If you want a vectorial IRSpace with vector dimension 17 you define it as

Dear Joachim,

Thank you for the response. This works. I do have an additional question.

Now that I’ve used IntegrationRuleSpace the data in gfuir.vec.data is equal to the actual u,p values as opposed to polynomial coefficients in the original gfu.vec.data.

However, I am not sure where the spatial data corresponding to these values is stored.

Normally with the original gfu I’d evaluate it like:

gfu(mesh(p1,p2)) or gfu.components[0](mesh(p1,p2))

For gfuir this won’t work.

Is there another way to obtain the corresponding spatial locations of the values?

you may create another GridFunction gfxyz in an IntegrationRuleSpace of dimension 3, and then you interpolate the coordinate coefficientfunctions as

then the vector entries are the coordinates at the points

Thank you, this worked.