inla.mesh.create: Low level function for high-quality triangulations

In inbo/INLA: Full Bayesian Analysis of Latent Gaussian Models using Integrated Nested Laplace Approximations

Description

Create a constrained refined Delaunay triangulation (CRDT) for a set of spatial locations.

Usage

inla.mesh.create(loc = NULL,
                 tv = NULL,
                 boundary = NULL, interior = NULL,
                 extend = (missing(tv) || is.null(tv)),
                 refine = FALSE,
                 lattice = NULL,
                 globe = NULL,
                 cutoff = 1e-12,
                 plot.delay = NULL,
                 data.dir,
                 keep = (!missing(data.dir) && !is.null(data.dir)),
                 timings = FALSE,
                 quality.spec = NULL,
                 crs=NULL)

inla.delaunay(loc, ...)

Arguments

loc	Matrix of point locations. Can alternatively be a SpatialPoints or SpatialPointsDataFrame object.
tv	A triangle-vertex index matrix, specifying an existing triangulation.
boundary	A list of inla.mesh.segment objects, generated by inla.mesh.segment, specifying boundary constraint segments.
interior	A list of inla.mesh.segment objects, generated by inla.mesh.segment, specifying interior constraint segments.
extend	logical or list specifying whether to extend the data region, with parameters n the number of edges in the extended boundary (default=8) offset the extension distance. If negative, interpreted as a factor relative to the approximate data diameter (default=-0.10) Setting to FALSE is only useful in combination lattice or boundary.
refine	logical or list specifying whether to refine the triangulation, with parameters min.angle the minimum allowed interior angle in any triangle. The algorithm is guaranteed to converge for min.angle at most 21 (default=21) max.edge the maximum allowed edge length in any triangle. If negative, interpreted as a relative factor in an ad hoc formula depending on the data density (default=Inf) max.n.strict the maximum number of vertices allowed, overriding min.angle and max.edge (default=-1, meaning no limit) max.n the maximum number of vertices allowed, overriding max.edge only (default=-1, meaning no limit)
lattice	An inla.mesh.lattice object, generated by inla.mesh.lattice, specifying points on a regular lattice.
globe	Subdivision resolution for a semi-regular spherical triangulation with equidistant points along equidistant latitude bands.
cutoff	The minimum allowed distance between points. Point at most as far apart as this are replaced by a single vertex prior to the mesh refinement step.
plot.delay	On Linux (and Mac if appropriate X11 libraries are installed), specifying a numeric value activates a rudimentary plotting system in the underlying fmesher program, showing the triangulation algorithm at work.
data.dir	Where to store the fmesher data files. Defaults to tempdir() if keep is FALSE, otherwise "inla.mesh.data".
keep	TRUE if the data files should be kept in data.dir or deleted afterwards. Defaults to true if data.dir is specified, otherwise false. Warning: If keep is false, data.dir and its contents will be deleted (unless set to tempdir()).
timings	If TRUE, obtain timings for the mesh construction.
quality.spec	List of vectors of per vertex max.edge target specification for each location in loc, boundary/interior (segm), and lattice. Only used if refining the mesh.
crs	An optional CRS or inla.CRS object
...	Optional parameters passed on to inla.mesh.create.

Details

inla.mesh.create generates triangular meshes on subsets of R^2 and S^2. Use the higher level wrapper function inla.mesh.2d for greater control over mesh resolution and coarser domain extensions.

inla.delaunay is a wrapper function for obtaining the convex hull of a point set and calling inla.mesh.create to generate the classical Delaunay tringulation.

Value

An inla.mesh object.

Author(s)

Finn Lindgren [e-mail]

See Also

inla.mesh.2d, inla.mesh.1d, inla.mesh.segment, inla.mesh.lattice, inla.mesh.query

Examples

loc = matrix(runif(10*2),10,2)

mesh = inla.delaunay(loc)
plot(mesh)

mesh = inla.mesh.create(loc,
                        interior=inla.mesh.segment(idx=1:2),
                        extend=TRUE,
                        refine=list(max.edge=0.1))
plot(mesh)

loc2 = matrix(c(0,1,1,0, 0,0,1,1), 4, 2);
mesh2 = inla.mesh.create(loc=loc,
                         boundary=inla.mesh.segment(loc2),
                         interior=inla.mesh.segment(idx=1:2),
                         quality.spec=list(segm=0.2, loc=0.05),
                         refine=list(min.angle=26))
plot(mesh2)

inbo/INLA documentation built on Dec. 6, 2019, 9:51 a.m.