Separating plot rendering from analysis

separate_plots.R

# How to separate plotting from CPU intensive analysis
plotfuns = list()
Npts = 1000
Nsims = 12

# Visualising a 1D random walk, for example
makewalk = function(Npts){
    # Do some like real heavy computing man
    deltas = sample(c(-1,1),Npts,replace=TRUE)
    walk = cumsum(deltas)
    # Return a new function DEFINITION without executing plotwalk function
    return(list("plotter"=function() plotwalk(walk),"walk"=walk))  
}

plotwalk = function(walk) {
 mlab = paste0("Random Walk ",sprintf("%04d", i))
 plot(walk,type="s",xlab="Time (ms)",ylab="Position (mm)",cex.lab=1.54,main=mlab) 
}

for(i in 1:Nsims){
  # Without actually executing the plotwalk function, store its output inside another function in e.g. a list
  plotfuns[[i]] = makewalk(Npts)[["plotter"]]
}

# Now we have a list of functions, which we can execute to make the plots in whatever format we like
# One plot per page
pdf("RandomWalkPages.pdf")
 for(i in 1:Nsims) walk = plotfuns[[i]]()
dev.off()

# Four plots per page
pdf("RandomWalkArrays.pdf")
 op = par(mfrow=c(2,2),mai=c(0.75,0.75,0.5,0.5))
  for(i in 1:Nsims) walk = plotfuns[[i]]()
 par(op)
dev.off()