Learn R Programming

TeachingDemos (version 2.13)

subplot: Embed a new plot within an existing plot

Description

Subplot will embed a new plot within an existing plot at the coordinates specified (in user units of the existing plot).

Usage

subplot(fun, x, y, size=c(1,1), vadj=0.5, hadj=0.5,
 inset=c(0,0), type=c('plt','fig'), pars=NULL)

Value

An invisible list with the graphical parameters that were in effect when the subplot was created. Passing this list to par will enable you to augment the embedded plot.

Arguments

fun

an expression defining the new plot to be embedded.

x

x-coordinate(s) of the new plot (in user coordinates of the existing plot), or a character string.

y

y-coordinate(s) of the new plot, x and y can be specified in any of the ways understood by xy.coords.

size

The size of the embedded plot in inches if x and y have length 1.

vadj

vertical adjustment of the plot when y is a scalar, the default is to center vertically, 0 means place the bottom of the plot at y, 1 places the top of the plot at y.

hadj

horizontal adjustment of the plot when x is a scalar, the default is to center horizontally, 0 means place the left edge of the plot at x, and 1 means place the right edge of the plot at x.

inset

1 or 2 numbers representing the proportion of the plot to inset the subplot from edges when x is a character string. The first element is the horizontal inset, the second is the vertical inset.

type

Character string, if 'plt' then the plotting region is defined by x, y, and size with axes, etc. outside that box; if 'fig' then all annotations are also inside the box.

pars

a list of parameters to be passed to par before running fun.

Author

Greg Snow 538280@gmail.com

Details

The coordinates x and y can be scalars or vectors of length 2. If vectors of length 2 then they determine the opposite corners of the rectangle for the embedded plot (and the parameters size, vadj, and hadj are all ignored).

If x and y are given as scalars then the plot position relative to the point and the size of the plot will be determined by the arguments size, vadj, and hadj. The default is to center a 1 inch by 1 inch plot at x,y. Setting vadj and hadj to (0,0) will position the lower left corner of the plot at (x,y).

If x is a character string, then it will be parsed for the strings "left", "right", "top", and "bottom" and x and y will be set appropriately (anything not specified will be set at the center in that dimension) using also the inset argument. This allows the position of the subplot to be specified as 'topleft' or 'bottom', etc. The inset argument is in proportion of the plot units, so 0.1 means inset 10% of the width/height of the plotting distance. If hadj/vadj are not specified, they will be set appropriately.

The rectangle defined by x, y, size, vadj, and hadj will be used as the plotting area of the new plot. Any tick marks, axis labels, main and sub titles will be outside of this rectangle if type is 'plt'. If type is 'fig' then the annotations will be inside the box.

Any graphical parameter settings that you would like to be in place before fun is evaluated can be specified in the pars argument (warning: specifying layout parameters here (plt, mfrow, etc.) may cause unexpected results).

After the function completes the graphical parameters will have been reset to what they were before calling the function (so you can continue to augment the original plot).

See Also

grconvertX, par, symbols, my.symbols, ms.image

Examples

Run this code
# make an original plot
plot( 11:20, sample(51:60) )

# add some histograms

subplot( hist(rnorm(100)), 15, 55)
subplot( hist(runif(100),main='',xlab='',ylab=''), 11, 51, hadj=0, vadj=0)
subplot( hist(rexp(100, 1/3)), 20, 60, hadj=1, vadj=1, size=c(0.5,2) )
subplot( hist(rt(100,3)), c(12,16), c(57,59), pars=list(lwd=3,ask=FALSE) )

### some of the following examples work fine in an interactive session,
### but loading the packages required does not work well in testing.

# augment a map
if( interactive() && require(spData) ){
	plot(state.vbm,fg=NULL)
	tmp <- cbind( state.vbm$center_x, state.vbm$center_y )
	for( i in 1:50 ){
		tmp2 <- as.matrix(USArrests[i,c(1,4)])
		tmp3 <- max(USArrests[,c(1,4)])
		subplot( barplot(tmp2, ylim=c(0,tmp3),names=c('',''),yaxt='n'),
			x=tmp[i,1], y=tmp[i,2], size=c(.1,.1))
	}
}

tmp <- rnorm(25)
qqnorm(tmp)
qqline(tmp)
tmp2 <- subplot( hist(tmp,xlab='',ylab='',main=''),
		grconvertX(0.1,from='npc'), grconvertY(0.9,from='npc'), 
		vadj=1, hadj=0 )
abline(v=0, col='red') # wrong way to add a reference line to histogram

# right way to add a reference line to histogram
op <- par(no.readonly=TRUE)
par(tmp2)
abline(v=0, col='green')
par(op)


# scatter-plot using images
if(interactive() && require(png)) {

  image.png <- function(x,...) {
    cols <- rgb( x[,,1], x[,,2], x[,,3], x[,,4] )
    z <- 1:length(cols)
    dim(z) <- dim(x[,,1])
    z <- t(z)
    z <- z[ ,rev(seq_len(ncol(z))) ]
    image(z, col=cols, axes=FALSE, ...)
  }

  logo <- readPNG(system.file("img", "Rlogo.png", package="png"))

  x <- runif(10)
  y <- runif(10)
  plot(x,y, type='n')
  for(i in 1:10) {
    subplot(image.png(logo), x[i], y[i], size=c(0.3,0.3))
  }
}

Run the code above in your browser using DataLab