plotGrid: Plot Spatial Deformations from bdef

In guiludwig/bsplinedef: Spatial Deformation via Tensor Product of B-splines

Description

This function plot a spatial deformation on a grid obtained via the bdef function

Usage

plotGrid(model, nx = 20, ny = 20, colorO = "Black", colorD = "Red",
  plot = TRUE, margins = TRUE, persp = FALSE, F1 = NULL,
  F2 = NULL, chull = FALSE, offset = c(0, 0), ...)

Arguments

model	An object returned by the function bdef. Alternatively, a list with named components matching those returned by bdef.
nx	Number of ticks in x-axis for grid. Defaults to 10.
ny	Number of ticks in y-axis for grid. Defaults to 10.
colorO	Color of original data points and grid. Defaults to Black.
colorD	Color of deformed data points and grid. Defaults to Red.
plot	Whether a plot should be produced or not. Defaults to TRUE.
margins	Whether a layout should be produced, with f1, f2 estimates shown along the corresponding axes. Defaults to TRUE.
F1	Allows the user to supply a function that generates the deformed coordinate y1. Defaults to NULL, in which case the tensor product of B-splines is used instead.
F2	Allows the user to supply a function that generates the deformed coordinate y2. Defaults to NULL, in which case the tensor product of B-splines is used instead.
chull	Whether the deformation is displayed inside of the convex hull of the sampled sites, or a square window containing the sampled sites. Defaults to FALSE, which corresponds to the square window.
...	Extra parameters to be passed to perspective plot, in particular the azymutal direction and colatitude. See persp for a description.

Value

plotGrid invisibly returns the values of the deformation on the grid.

Author(s)

Guilherme Ludwig and Ronaldo Dias

References

To add.

See Also

bdef

Examples

# Example using artificially generated data
set.seed(1)
m <- 10
x1 <- (0:m)/m
x2 <- (0:m)/m
x <- as.matrix(expand.grid(x1,x2))
n <- nrow(x)
K <- 4
tempb1 <- bs(x[, 1], K, intercept = TRUE)
tempb2 <- bs(x[, 2], K, intercept = TRUE)
tempW <- matrix(0, n, K^2)
for(i in 1:n) tempW[i,] <- kronecker(tempb2[i,], tempb1[i,])
theta1 <- solve(crossprod(tempW), crossprod(tempW, x[,1] + .2*x[,2]))
theta2 <- solve(crossprod(tempW), crossprod(tempW, x[,2] - .2*sqrt(x[,1]*x[,2])))
def.x <- cbind(tempW%*%theta1, tempW%*%theta2)
fakeModel <- list(basis = list(B1 = bs(range(x[, 1]), K, intercept = TRUE),
                               B2 = bs(range(x[, 2]), K, intercept = TRUE)),
                  window = list(x = 0:1, y = 0:1),
                  x = x, def.x = def.x,
                  theta1 = theta1,
                  theta2 = theta2,
                  df1 = K,
                  df2 = K)
 plotGrid(fakeModel, margins = FALSE)
 plotGrid(fakeModel)
 plotGrid(fakeModel, persp = TRUE, theta = 15)
# Example with truth function
set.seed(1)
m <- 10
x1 <- (0:m)/m
x2 <- (0:m)/m
x <- as.matrix(expand.grid(x1,x2))
n <- nrow(x)
F1 <- function(x1,x2, a = 2.5, b = 1.0) {
  x <- x1 - 0.5; y <- x2 - 0.5
angle <- a*exp(-(x*x+y*y)/(b*b)) + 3*pi/2
return(cos(angle)*x + sin(angle)*y + 0.5)
}
F2 <- function(x1,x2, a = 2.5, b = 1.0) {
  x <- x1 - 0.5; y <- x2 - 0.5
angle <- a*exp(-(x*x+y*y)/(b*b)) + 3*pi/2
return(y + x^2)
}
plotGrid(list(window = list(x = range(x1), y = range(x2)),
x = x, def.x = cbind(F1(x[,1],x[,2]), 
                                  F2(x[,1],x[,2]))),
        F1 = F1, F2 = F2)

guiludwig/bsplinedef documentation built on May 16, 2020, 10:24 p.m.