good triangulations and meshing
DESCRIPTION
Good Triangulations and Meshing. Lecturer: Ofer Rothschild. Problem with the borders. Meshes without borders. Given a set of points in [1/4, 3/4]X[1/4, 3/4] With diameter > ½ The points are vertices in the triangulation. Fat triangulation. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/1.jpg)
1
Good Triangulations and Meshing
Lecturer: Ofer Rothschild
![Page 2: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/2.jpg)
2
Problem with the borders
![Page 3: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/3.jpg)
3
Meshes without borders
• Given a set of points in [1/4, 3/4]X[1/4, 3/4]• With diameter > ½
• The points are vertices in the triangulation
![Page 4: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/4.jpg)
4
Fat triangulation
• Aspect ratio of a triangle: the longest side divided by its height
• We want to minimize the maximal aspect ration in the triangulation
• Alpha-fat triangulation
• The longest dimension to the shortest
![Page 5: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/5.jpg)
5
Balanced quadtrees
• Corner
• Split
• Balanced: No more than one split in each side– And no compressed nodes
![Page 6: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/6.jpg)
6
Extended cluster
• Extended cluster of a cell: 9X9 cells around it with the same size
![Page 7: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/7.jpg)
7
Well-balanced quadtree
• Balanced• Every none-empty cell has all its extended
cluster,• And doesn’t have any other point in it
![Page 8: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/8.jpg)
8
Points -> Well balanced quadtree
• Given a set of points• Time: O(n log n + m)
• Build a compressed quadtree• Uncompress it• Make it balanced• Make it well balanced
![Page 9: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/9.jpg)
9
Make it balanced
• Queue of all nodes• Hash table of id -> node• For each node side:• Does or exist?• Create and all its parents• O(m) time
![Page 10: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/10.jpg)
10
Make it well balanced
• For every point p and its leaf cell :• If the (theoretical) extended cluster contains
other points:• Split enough times and insert its extended
cluster• Rebalance the new nodes• Time: testing time + new nodes
![Page 11: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/11.jpg)
11
Emptiness of EC
• During the algorithm:• We remember for every node if it contains
points in its sub tree• For every cell in the EC: • If it exists, check the flag.• Otherwise, its parent exists as a leaf.• If the parent isn’t empty, test the point.• Add illustration
![Page 12: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/12.jpg)
12
Well balanced -> Triangulation
• For every point: warp the cells and triangulate:
• For other cells:
![Page 13: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/13.jpg)
13
![Page 14: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/14.jpg)
14
The aspect ratio is at most 4
• For cells without points: 2
![Page 15: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/15.jpg)
15
Warped cells: point inside cell
![Page 16: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/16.jpg)
16
Warped cells: point outside of cell
![Page 17: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/17.jpg)
17
Local feature size
• For any point p in the domain, lfsP(p): the distance from the second closest point in P.
• The size of the triangulation:
dpplfs
DPDp
P
2))((1),(#
![Page 18: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/18.jpg)
18
Upper bound
• The leaves are not much smaller than lfs:
![Page 19: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/19.jpg)
19
J’accuse: why the cell exists
• EC of a point• Original quadtree cell– Cell with a point– Empty cell
• Balance condition– Blame another cell
*
*
![Page 20: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/20.jpg)
20
m = O(#(P, D))
![Page 21: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/21.jpg)
21
Lower bound
• Assume a triangulation with aspect ratio .
![Page 22: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/22.jpg)
22
Aspect ratio
• Let phi be the smallest angle in a triangle.
• In particular,
![Page 23: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/23.jpg)
23
Edge ratio
![Page 24: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/24.jpg)
24
p
![Page 25: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/25.jpg)
25
Star and link
![Page 26: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/26.jpg)
26
Disk in star
• Let e be the longest edge of p.• The disk with radius:
centered at p, is contained in star(p).
p
![Page 27: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/27.jpg)
27
Empty buffer around the triangle
![Page 28: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/28.jpg)
28
• We prove that every triangle contributes a constant to the integral of #.
• So the whole integral is proportional to the number of triangles.
![Page 29: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/29.jpg)
29
![Page 30: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/30.jpg)
30
Supplementaries
• 5 X 5• 9 X 9
• Size of the triangulation
![Page 31: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/31.jpg)
31
Covering by disks
• Given a set of points P in a domain D:• Cover the domain with a small number of
disks, s.t.:– Every disk contains up to one point of P– Every two intersecting disks
have roughly the same radius
![Page 32: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/32.jpg)
32
A greedy algorithm
• For every uncovered point p in D, add the disk with size
• Clearly, the algorithm stops
![Page 33: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/33.jpg)
33
Lemma: lfs is 1-Lipschitz
![Page 34: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/34.jpg)
34
Intersecting disks are about the same size
pq
r'r
![Page 35: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/35.jpg)
35
No point in the plane is covered by more than O(1) disks
![Page 36: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/36.jpg)
36
How many disks do we expect?• If lfs(p) is , the radius is /2 and the area is
( 2).• If the lfs is the same everywhere:• Number of disks will be (area(D)/ 2 )
![Page 37: Good Triangulations and Meshing](https://reader036.vdocument.in/reader036/viewer/2022062302/568166e9550346895ddb2bb9/html5/thumbnails/37.jpg)
37
But the lfs changes