Hmax ignored in mmg2D when 1D boundary is "required"

[gbettega]

Hello forum
a question. I have a 2d boundary in which all the edges are defined as required, i.e. “fixed”. I attached it (“test.mesh”).

test.mesh (1.1 KB)

It seems that the mesh engine does not see the hmin and hmax values set by the user.

The command lines

mmg2d.exe test.mesh -hmax 0.1

and

mmg2d.exe test.mesh -hmax 0.01

give results having exactly the same number of nodes and elements (433 nodes, 830 triangles, in both the cases), with the result shown in the picture

Obviously the things change when the required edges are removed, not necessarily all, but also a bunch. See the following pictures (hmax = 0.1, hmax = 0.01)

Since I did not study the code in deepth, but I’m only a standard user, I cannot understand (I only can suppose) why this happens.

Grazie mille
Giovanni

[gbettega]

Hello
actually I verified that also the metric settings are not applied if the previous rectangle has a boundary defined as required.
I attact an input mesh with no required edges and a anisotropic metric

rectangle__no_requiredEdges.mesh (26.5 KB)
metric.sol (3.9 KB)

The results is in the picture: the metric is working

If the required edges (along the whole 1d boundary) are active, as in the following input file

rectangle__requiredEdges.mesh (26.5 KB)

the metric is not taken into account, as show in the picture:

Of course the “required edges” at boundary will impose a constraint on the metric application, but I thought that at least, “far” from the boundary; the metric would have been respected.
Any advice?
Grazie mille
Giovanni

[Johannes_Neumann]

I think you want to take a look a the hgrad option: https://www.mmgtools.org/mmg-remesher-try-mmg/mmg-remesher-options/mmg-remesher-option-hgrad

[gbettega]

Don you mean that I should use hgrad >> 1? The two grids have Benny obtained with the same (default) hgrad: the metric Is respected without required edges, and not with required edges.
Thank you
G

P.S @14:53 - tried with hgrad = 2.0 - 2.5, but the result is the same. No metric applied

[gbettega]

@gbettega

defined as required.
I attact an input mesh with no required edges and a anisotropic metric

It seems to work when hgrad = 1.0, which is reasonable. See picture:

Also hmax sizing seems working when hgrad = 1.0