Periodic background mesh

[zjwegert]

Hi,
Is there support for periodic background meshes? The MWE below does not work.

using Gridap
using GridapEmbedded,GridapEmbedded.LevelSetCutters
import Gridap.Geometry: get_node_coordinates,collect1d

model = CartesianDiscreteModel((0,1,0,1),(100,100),isperiodic=(true,true));
V_φ = TestFESpace(model,ReferenceFE(lagrangian,Float64,1))

φh = interpolate(x->-cos(4π*x[1])*cos(4*pi*x[2])/4-0.2/4,V_φ)
point_to_coords = model.grid_topology.vertex_coordinates
geo = DiscreteGeometry(get_free_dof_values(φh),point_to_coords,name="")
cutgeo = cut(model,geo)

Ω = Triangulation(cutgeo,PHYSICAL)

writevtk(Ω,"discrete_geo_periodic",cellfields=["φh"=>φh])

Based on some testing I've conducted, it seems that analytic geometries do respect periodicity of the background mesh though.

[zjwegert]

Similar to #89, it looks like this just requires that the FEFunction is evaluated at all model node coordinates. E.g.,

using Gridap
using GridapEmbedded,GridapEmbedded.LevelSetCutters
import Gridap.Geometry: get_node_coordinates,collect1d

model = CartesianDiscreteModel((0,1,0,1),(100,100),isperiodic=(true,true));
V_φ = TestFESpace(model,ReferenceFE(lagrangian,Float64,1))

φh = interpolate(x->-cos(4π*(x[1]))*cos(4*pi*x[2])/4-0.2/4,V_φ)
point_to_coords = collect1d(get_node_coordinates(model))
geo = DiscreteGeometry(φh(point_to_coords),point_to_coords,name="")
cutgeo = cut(model,geo)

Ω = Triangulation(cutgeo,PHYSICAL)

writevtk(Ω,"discrete_geo_periodic",cellfields=["φh"=>φh])

It might be nice to create a version of the constructor for DiscreteGeometry that takes an FEFunction/CellField and is optimised to avoid searching in φh(point_to_coords)

[zjwegert]

The changes in #89 allows for a periodic background mesh when generating a DiscreteGeometry. I have added a serial test of AgFEM with a periodic background mesh in #90. This works as expected.

I've also added a distributed test but this is currently disabled due to #87. I expect these will work once #87 has been fixed.

[ericneiva]

Hi, @zjwegert, concerning #87, I have already implemented the fix in
b70a256. Whenever I find some time, I will PR this into to GE. Hope I can do it soon.

PS: I only see your comment now (sorry, I was on holidays).

[ericneiva]

Hi, @zjwegert, FYI: #93