| lambertW {emdbook} | R Documentation |
Computes the Lambert W function, giving efficient solutions to the equation x*exp(x)==z
lambertW(z, b = 0, maxiter = 10, eps = .Machine$double.eps, min.imag = 1e-09)
z |
(complex) vector of values for which to compute the function |
b |
(integer) vector of branches: b=0 specifies the principal branch, 0 and -1 are the ones that can take non-complex values |
maxiter |
maximum numbers of iterations for convergence |
eps |
convergence tolerance |
min.imag |
maximum magnitude of imaginary part to chop when returning solutions |
Compute the Lambert W function of z. This function satisfies W(z)*exp(W(z)) = z, and can thus be used to express solutions of transcendental equations involving exponentials or logarithms. In ecology, the Lambert W can be used to solve the so-called "Rogers equation" for predator functional response with depletion.
Complex or real vector of solutions.
This implementation should return values within 2.5*eps of its counterpart in Maple V, release 3 or later. Please report any discrepancies to the author or translator.
Nici Schraudolph <schraudo@inf.ethz.ch> (original version (c) 1998), Ben Bolker (R translation)
Corless, Gonnet, Hare, Jeffrey, and Knuth (1996), "On the Lambert W Function", Advances in Computational Mathematics 5(4):329-359
?Lambert in the gsl package by Robin Hankin,
which uses Gnu Scientific Library code
curve(lambertW(x),from=0,to=10)
pvec = seq(-1,1,length=40)
m = outer(pvec,pvec,function(x,y)Re(lambertW(x+y*1i)))
persp(pvec,pvec,m,
theta=290,shade=0.5,zlab="lambertW")
num1 = uniroot(function(x) {x*exp(x)-1},lower=0,upper=1,tol=1e-9)
abs(lambertW(1)-num1$root)<1e-9
###
## Rogers random predator equation:
rogers.pred <- function(N0,a,h,T) {
N0 - lambertW(a*h*N0*exp(-a*(T-h*N0)))/(a*h)
}
holling2.pred <- function(N0,a,h) {
a*N0/(1+a*h*N0)
}
curve(rogers.pred(x,a=1,h=0.2,T=1),from=0,to=60,
ylab="Number eaten/unit time",xlab="Initial number",ylim=c(0,5),
main="Predation: a=1, h=0.2")
curve(rogers.pred(x,a=1,h=0.2,T=5)/5,add=TRUE,lty=2,from=0)
curve(rogers.pred(x,a=1,h=0.2,T=0.2)*5,add=TRUE,lty=3,from=0)
curve(rogers.pred(x,a=1,h=0.2,T=10)/10,add=TRUE,lty=4,from=0)
curve(holling2.pred(x,a=1,h=0.2),add=TRUE,lty=1,lwd=2,from=0)
abline(h=5)
legend(30,2,
c(paste("Rogers, T=",c(0.2,1,5,10),sep=""),
"Holling type II"),lwd=c(rep(1,4),2),lty=c(3,1,2,4,1))
## final size of an epidemic
finalsize = function(R0) {
1+1/R0*lambertW(-R0*exp(-R0))
}
curve(finalsize,from=1,to=10,xlab=expression(R[0]),ylab="Final size")