tessellation (version 2.3.0)

delaunay: Delaunay triangulation

Description

Delaunay triangulation (or tessellation) of a set of points.

Usage

delaunay(
  points,
  atinfinity = FALSE,
  degenerate = FALSE,
  exteriorEdges = FALSE,
  elevation = FALSE
)

Value

If the function performs an elevated Delaunay tessellation, then the returned value is a list with four fields: mesh, edges,

volume, and surface. The mesh field is an object of class mesh3d, ready for plotting with the rgl package. The

edges field is an integer matrix which provides the indices of the vertices of the edges, and an indicator of whether an edge is a border edge; this matrix is obtained with vcgGetEdge. The volume field provides the sum of the volumes under the Delaunay triangles, that is to say the total volume under the triangulated surface. Finally, the surface field provides the sum of the areas of the Delaunay triangles, thus this is an approximate value of the area of the surface that is triangulated. The elevated Delaunay tessellation is built with the help of the

interp package.

Otherwise, the function returns the Delaunay tessellation with many details, in a list. This list contains five fields:

vertices

the vertices (or sites) of the tessellation; these are the points passed to the function

tiles

the tiles of the tessellation (triangles in dimension 2, tetrahedra in dimension 3)

tilefacets

the facets of the tiles of the tessellation

mesh

a 'rgl' mesh (mesh3d object)

edges

a two-columns integer matrix representing the edges, each row represents an edge; the two integers of a row are the indices of the two points which form the edge.

In dimension 3, the list contains an additional field exteriorEdges

if you set exteriorEdges = TRUE. This is the list of the exterior edges, represented as Edge3 objects. This field is involved in the function plotDelaunay3D.

The vertices field is a list with the following fields:

id

the id of the vertex; this is nothing but the index of the corresponding point passed to the function

neighvertices

the ids of the vertices of the tessellation connected to this vertex by an edge

neightilefacets

the ids of the tile facets this vertex belongs to

neightiles

the ids of the tiles this vertex belongs to

The tiles field is a list with the following fields:

id

the id of the tile

simplex

a list describing the simplex (that is, the tile); this list contains four fields: vertices, a hash giving the simplex vertices and their id, circumcenter, the circumcenter of the simplex, circumradius, the circumradius of the simplex, and volume, the volume of the simplex

facets

the ids of the facets of this tile

neighbors

the ids of the tiles adjacent to this tile

family

two tiles have the same family if they share the same circumcenter; in this case the family is an integer, and the family is NA for tiles which do not share their circumcenter with any other tile

orientation

1 or -1, an indicator of the orientation of the tile

The tilefacets field is a list with the following fields:

id

the id of this tile facet

subsimplex

a list describing the subsimplex (that is, the tile facet); this list is similar to the simplex list of tiles

facetOf

one or two ids, the id(s) of the tile this facet belongs to

normal

a vector, the normal of the tile facet

offset

a number, the offset of the tile facet

Arguments

points

the points given as a matrix, one point per row

atinfinity

Boolean, whether to include a point at infinity; ignored if elevation=TRUE

degenerate

Boolean, whether to include degenerate tiles; ignored if elevation=TRUE

exteriorEdges

Boolean, for dimension 3 only, whether to return the exterior edges (see below)

elevation

Boolean, only for three-dimensional points; if TRUE, the function performs an elevated Delaunay triangulation (also called 2.5D Delaunay triangulation), using the third coordinate of a point as its elevation; see the example

See Also

getDelaunaySimplices

Examples

library(tessellation)
points <- rbind(
 c(0.5,0.5,0.5),
 c(0,0,0),
 c(0,0,1),
 c(0,1,0),
 c(0,1,1),
 c(1,0,0),
 c(1,0,1),
 c(1,1,0),
 c(1,1,1)
)
del <- delaunay(points)
del$vertices[[1]]
del$tiles[[1]]
del$tilefacets[[1]]

# an elevated Delaunay tessellation ####
f <- function(x, y){
  dnorm(x) * dnorm(y)
}
x <- y <- seq(-5, 5, length.out = 50)
grd <- expand.grid(x = x, y = y) # grid on the xy-plane
points <- as.matrix(transform( # data (x_i, y_i, z_i)
  grd, z = f(x, y)
))
del <- delaunay(points, elevation = TRUE)
del[["volume"]] # close to 1, as expected
# plotting
library(rgl)
mesh <- del[["mesh"]]
open3d(windowRect = c(100, 100, 612, 356), zoom = 0.6)
aspect3d(1, 1, 20)
shade3d(mesh, color = "limegreen", polygon_offset = 1)
wire3d(mesh)

# another elevated Delaunay triangulation, to check the correctness
#   of the calculated surface and the calculated volume ####
library(Rvcg)
library(rgl)
cap <- vcgSphericalCap(angleRad = pi/2, subdivision = 3)
open3d(windowRect = c(100, 100, 612, 356), zoom = 0.6)
shade3d(cap, color = "lawngreen", polygon_offset = 1)
wire3d(cap)
# exact value of the surface of the spherical cap:
R <- 1
h <- R * (1 - sin(pi/2/2))
2 * pi * R * h
# our approximation:
points <- t(cap$vb[-4, ]) # the points on the spherical cap
del <- delaunay(points, elevation = TRUE)
del[["surface"]]
# try to increase `subdivision` in `vcgSphericalCap` to get a
#   better approximation of the true value
# note that 'Rvcg' returns the same result as ours:
vcgArea(cap)
# let's check the volume as well:
pi * h^2 * (R - h/3) # true value
del[["volume"]]
# there's a warning with 'Rvcg':
tryCatch(vcgVolume(cap), warning = function(w) message(w))
suppressWarnings({vcgVolume(cap)})