Simulation of timeseries in R

Question

I need to make a simulation in R of the following function:

y[t] = 4 + (9*x[t-1])*y[y-1]+r[t]

I also need to make a 3D plot. I'm new to R, and I therefore find it extremely difficult and hope that someone in here can help me.

This is what I've tried so far:

library(rgl)
n <- 3000
x <- runif(n,-1,1)
r <- rnorm(n)
y <- rep(NA,n)
y[1] <- r[1]
for(t in 2:n){
y[t] <- 4 + (9*x[t-1])*y[t-1]+r[t]
}
D <- data.frame(y=y[-1], x1=x[-n],y1=y[-n])
open3d()
points3d(D$x1, D$y1, D$y, size=3, col="red")
aspect3d()
axes3d()

But I don't know if the simulation is done correctly, and nothing is shown when I plot the data points. I use the

Answer 1

you have y <- 4 + 9*x[t-1])*y[t-1] its rise like geometric progression 10^t (" 9 " coefficient) ... The problem in "9" and "4" coefficients. you would have " Inf " of " -Inf " most y. The rest 'y' would be huge...

Answer 2

Your simulation works fine, it's just that the values get too large too quickly. I've removed the infinite values and logged the large ones:

D <- D[!is.infinite(D$y) & !is.infinite(D$y1),]
D$y <- log(D$y)
D$y1 <- log(D$y1)
plot3d(D)

The above code successfully produces a 3D plot.