There is geometry in the humming of the strings, there is music in the spacing of the spheres (Pythagoras)

Spring is here and I will be on holiday next week. I cannot be more happy! It is time to celebrate so I have drawn another fractal. It is called the Pythagorean Tree:

Here you have the code. See you soon:

library("grid")
l=0.15 #Length of the square
grid.newpage()
gr <- rectGrob(width=l, height=l, name="gr") #Basic Square
pts <- data.frame(level=1, x=0.5, y=0.1, alfa=0) #Centers of the squares
for (i in 2:10) #10=Deep of the fractal. Feel free to change it
{
  df<-pts[pts$level==i-1,]
  for (j in 1:nrow(df))
  {
    pts <- rbind(pts, 
                 c(i, 
                   df[j,]$x-2*l*((1/sqrt(2))^(i-1))*sin(df[j,]$alfa+pi/4)-0.5*l*((1/sqrt(2))^(i-2))*sin(df[j,]$alfa+pi/4-3*pi/4), 
                   df[j,]$y+2*l*((1/sqrt(2))^(i-1))*cos(df[j,]$alfa+pi/4)+0.5*l*((1/sqrt(2))^(i-2))*cos(df[j,]$alfa+pi/4-3*pi/4),                  
                   df[j,]$alfa+pi/4))
    pts <- rbind(pts, 
                 c(i, 
                   df[j,]$x-2*l*((1/sqrt(2))^(i-1))*sin(df[j,]$alfa-pi/4)-0.5*l*((1/sqrt(2))^(i-2))*sin(df[j,]$alfa-pi/4+3*pi/4), 
                   df[j,]$y+2*l*((1/sqrt(2))^(i-1))*cos(df[j,]$alfa-pi/4)+0.5*l*((1/sqrt(2))^(i-2))*cos(df[j,]$alfa-pi/4+3*pi/4),                  
                   df[j,]$alfa-pi/4))
  }
}
for (i in 1:nrow(pts))
{
  grid.draw(editGrob(gr, vp=viewport(x=pts[i,]$x, y=pts[i,]$y, w=((1/sqrt(2))^(pts[i,]$level-1)), h=((1/sqrt(2))^(pts[i,]$level-1)), angle=pts[i,]$alfa*180/pi), 
            gp=gpar(col=0, lty="solid", fill=rgb(139*(nrow(pts)-i)/(nrow(pts)-1), 
                                                 (186*i+69*nrow(pts)-255)/(nrow(pts)-1), 
                                                  19*(nrow(pts)-i)/(nrow(pts)-1), 
                                                  alpha= (-110*i+200*nrow(pts)-90)/(nrow(pts)-1), max=255))))
}

Computer Science is no more about computers than astronomy is about telescopes (E. W. Dijkstra)

Some days ago I published a post about how to build fractals with R using Multiple Reduction Copt Machine (MRCM) algorithm. Is that case I used a feature of the grid package that allows you to locate objects easily into the viewPort avoiding to work with coordinates. It does not work well if you want to divide your seed image into five subimages located in the vertex of a regular pentagon. No problem: after refreshing some trigonometric formulas and after understanding how to work with coordinates I felt strong enough to program the Final-MRCM-Fractal-Builder. But here comes the harsh reality. My computer crashes when I try to go beyond five degrees of depth. Imposible. In the example of Sierpinski’s triangle, where every square in divided into three small ones, I reached seven degrees of depth. I am deeply frustrated. These are drawings for 1, 2, 3 and 5 degrees of depth.

Please, if someone modifies code to make it more efficient, let me know. I used circles in this case instead squares. Here you have it:

library(grid)
grid.newpage()
rm(list = ls())
ratio <- 0.4
pmax <- 5 # Depth
vp1 <- viewport(w=1, h=1)
vp2 <- viewport(w=ratio, h=ratio, just=c(0.75*sin(2*pi*1/5)+0.5, 0.75*cos(2*pi*1/5)+0.75*pi*1/5))
vp3 <- viewport(w=ratio, h=ratio, just=c(0.75*sin(2*pi*0/5)+0.5, 0.75*cos(2*pi*0/5)+0.75*pi*1/5))
vp4 <- viewport(w=ratio, h=ratio, just=c(0.75*sin(2*pi*2/5)+0.5, 0.75*cos(2*pi*2/5)+0.75*pi*1/5))
vp5 <- viewport(w=ratio, h=ratio, just=c(0.75*sin(2*pi*3/5)+0.5, 0.75*cos(2*pi*3/5)+0.75*pi*1/5))
vp6 <- viewport(w=ratio, h=ratio, just=c(0.75*sin(2*pi*4/5)+0.5, 0.75*cos(2*pi*4/5)+0.75*pi*1/5))
pushViewport(vp1)
grid.rect(gp=gpar(fill="white", col=NA))
m <- as.matrix(expand.grid(rep(list(2:6), pmax)))
for (j in 1:nrow(m))
{
for(k in 1:ncol(m)) {pushViewport(get(paste("vp",m[j,k],sep="")))}
grid.circle(gp=gpar(col="dark grey", lty="solid", fill=rgb(sample(0:255, 1),sample(0:255, 1),sample(0:255, 1), alpha= 95, max=255)))
upViewport(pmax)
}

Clouds are not spheres, mountains are not cones, coastlines are not circles and bark is not smooth, nor does lightning travel in a straight line (Benoit Maldelbrot)

Fractals are beautiful, hypnotics, mysterious. Cantor set has as many points as the real number line but has zero measure. After 100 steps, the Koch curve created from a 1 inch segment is long enough  to wrap around the Earth at the equator nearly four thousand times. The Peano Curve is a line that has the same dimension as a plane. Fractals are weird mathematical objects. Fractals are very cool.

One way to build fractals is using the Multiple Reduction Copy Machine (MRCM) algorithm which uses affine linear transformations over a seed image to build fractals. MRCM are iterative algorithms that perform some sort of copy+paste task. The idea is quite simple: take a seed image, transform it (clonation, scalation, rotation), obtain the new image and iterate.

To create the Sierpinsky Gasket Fractal you part from a square. Then you divide it into three smaller squares, locate them as a pyramid and iterate doing the same with avery new square created. Making these things is very easy with grid package. Defining the division (i.e. the affine transformation) properly using viewPort function and navigating between them is all you need. Here you have the Sierpinsky Gasket Fractal with 1, 3, 5 and 7 levels of depth. I filled in squares with random colours (I like giving some touch of randomness to pictures). Here you have pictures:

And here you have the code. Feel free to build your own fractals.

library(grid)
rm(list = ls())
grid.newpage()
pmax <- 5 # Depth of the fractal
vp1=viewport(x=0.5,y=0.5,w=1, h=1)
vp2=viewport(w=0.5, h=0.5, just=c("centre", "bottom"))
vp3=viewport(w=0.5, h=0.5, just=c("left", "top"))
vp4=viewport(w=0.5, h=0.5, just=c("right", "top"))
pushViewport(vp1)
m <- as.matrix(expand.grid(rep(list(2:4), pmax)))
for (j in 1:nrow(m))
{
for(k in 1:ncol(m)) {pushViewport(get(paste("vp",m[j,k],sep="")))}
grid.rect(gp=gpar(col="dark grey", lty="solid",
fill=rgb(sample(0:255, 1),sample(0:255, 1),sample(0:255, 1), alpha= 95, max=255)))
upViewport(pmax)
}