Smooth outer does nothing

[chlnddev]

Smooth outer does nothing

[krober10nd]

• Edge function is indeed smoothed but it appears not to be used inside the generation of the mesh.
  

[WPringle]

Interesting, I guess we only tested the result of the smooth edgefx in detail. I'll test your example out.

[krober10nd]

Yea the outer box doesn't get the correct initial points especially near the inner nest even despite its sizing function having this variation in it.

[WPringle]

It does work if you do enough iterations. There was a bug in dpoly #119 btw that caused it to crash when it did more iterations. At 10 iterations it does the delete close points and starts to work.

[krober10nd]

Yea but that’s relying on the mesh adaptation. Shouldn’t the initial point distribution follow the variations in the sizing function?

[krober10nd]

We should create several of these tests.

[WPringle]

Yea the outer box doesn't get the correct initial points especially near the inner nest even despite its sizing function having this variation in it.

Hmm yeah I see it should delete some of them initially based on the edgefx. I'll take a look at this too.

[WPringle]

Yea the outer box doesn't get the correct initial points especially near the inner nest even despite its sizing function having this variation in it.

Hmm yeah I see it should delete some of them initially based on the edgefx. I'll take a look at this too.

Oh now I get it. Yeah so the h0 of the outer is larger than the edgefx value in the smoothed region so it can't be distributed more finely.

[krober10nd]

Yea so then the h0 should be determined after smooth outer is called ay

[WPringle]

Yeah if I re-update the h0 for the outer to be equal to the minimum of the ef then I get this:

[krober10nd]

Yea and what happens if you shrink the inner grade?

[WPringle]

That wouldn't affect anything, the grade would go by the outer one

[krober10nd]

Shouldn't it go by the minimum grade?

[krober10nd]

In theory we could enforce variable grade only in their respective boxes

[krober10nd]

Okay, so I make the initial point formation/rejection use the minimum of all h0

        disp(['    for box #' num2str(box_num)]);
                    % checking if cell or not and applying local values
                    h0_l = min(obj.h0);
                    max_r0 = 1/h0_l^2;   

That wouldn't affect anything, the grade would go by the outer one

And my my observations contradict this.

• Inner box grade set to 0.15 (outer box grade 0.15)

• Inner box grade set to 0.05 (outer box grade 0.15)

.

[krober10nd]

^ This result is also not because the initial point distribution is correct either. The periodic mesh improvement causes it to smooth out.

[krober10nd]

Nevermind I got it. I'll add it to the dpoly_patch PR