CGAL users discussion list

[Adam Getchell <>]

Of course right after I sent this I thought about it and did:

n_1->set_neighbor(n_1->index(top), n_5);
n_2->set_neighbor(n_2->index(top), n_6);
n_3->set_neighbor(n_3->index(top), n_7);
n_4->set_neighbor(n_4->index(top), n_8);
n_5->set_neighbor(n_5->index(bottom), n_1);
n_6->set_neighbor(n_6->index(bottom), n_2);
n_7->set_neighbor(n_7->index(bottom), n_3);
n_8->set_neighbor(n_8->index(bottom), n_4);

But that doesn’t quite do the trick.

-- 
Adam Getchell
https://adamgetchell.org

On Jun 30, 2022, at 2:36 PM, Adam Getchell <> wrote:

Hi Marc,

On Jun 30, 2022, at 12:55 AM, Marc Alexa ( via cgal-discuss Mailing List) <> wrote:

Hi Adam,

I think you also need to update the neighbors of n_1,…. set_neighbors applied to after_1,… will only change the neighbor entries in after_1,…, but not in the associated neighbors. 

Okay, that makes sense (if a bit tedious).

However, for my test case I have just 6 vertices, plus the infinite vertex. All of the neighbors of my 4-cell are infinite. So, how would I specify the neighbors of infinite cells?

Or is this a complete end case and my test case needs to have at least two layers?

(The API is a bit inconsistent among the triangulations for different dimension. I know for dD triangulations there is a different set of 4 parameters, where you set one neighboring relationship between two adjacent full simplex with one call.)

That would be nice!

Best,

Marc

PS: I have a working (but dirty) implementation for bistellar flips in dD triangulations. I’d be interested in finding others who would like to work on this, improve the interface, clean it up a bit, and then make it available.

Interested, especially since my next stage is 4D triangulations, for which I’ll almost certainly need this.

Cheers,

Adam

On 29. Jun 2022, at 22:52, Adam Getchell ( via cgal-discuss Mailing List) <> wrote:

Hello all,

I’m working on a bistellar flip for 3D triangulations, briefly as follows:

Step 1: Find a candidate edge for which a circulator around that edge obtains 4 incident cells.

Step 2: Get the 2 vertices comprising the candidate edge.

Step 3: Get the remaining 4 vertices in the 4-cell complex (there are 6 vertices in the cell complex).

Step 4: Obtain a new pivot edge between 2 different vertices for which a circulator around that pivot edge would obtain 4 new incident cells comprised of the 6 vertices.

Step 5: Find the 8 neighboring cells of the original 4-cell complex.

Step 6: Delete the old cells.

Step 7: Create the new cells using the same 6 vertices.

Step 8: Assign the 8 neighboring cells to the cells in the new 4-cell complex.

Details here: https://github.com/acgetchell/bistellar-flip

The issue is at the conclusion of Step 8, all cells have been verified as valid, cell orientations have been verified/reoriented, but the triangulation itself is not valid.

The error is: “neighbor of c has not c as neighbor”

My question is, do I also need to track the 32 neighboring cells of the 8 neighbors of the 4-cell complex being flipped? (And in my test case, the 8 neighbors are all infinite.)

Is there some other mistake?

Thanks for your advice,

Adam

P.S. I have looked at the Tetrahedral Remeshing package at Sebastien Loriot’s suggestion. I’m presently unsure if any of those functions (e.g. construct_opposite_vector_3_object()) are relevant.

-- 
Adam Getchell
https://adamgetchell.org

--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://sympa.inria.fr/sympa/info/cgal-discuss

--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://sympa.inria.fr/sympa/info/cgal-discuss