A.Det=c(-0.595750,0.050250,0.270000,0.711500,1.608750,3.353500,4.937652,4.573750,4.955000)
luz=c(0,25,50,75,100,500,1000,1500,2000)
ajuste <- nls((A.Det~(a + 0.595750) * (1 - exp (-b * luz/(a-0.595750)))+0.595750),
start=list(a=500, b=5))
plot(luz,A.Det)
curve(3.9669 + 0.595750 * (1 - exp (-0.0061 * x/(3.9669-0.595750)))+0.595750,add=TRUE,col='blue')
curve(SSasympOff(x,4.8292,-5.7353,23.452),add=TRUE,col='red')
legend('bottomright', xjust=1,bty="n",
c('Alternativo','Gilson'),
lwd=2,col=c('red','blue'))
Sobre este modelo 'alternativo', consulte "Mixed Effects Models in S and S-Plus, Pinheiro & Bates (2000)"
OBS: Estes modelos não tem ponto de inflexão,
pois quanto x tende ao infinito, y tende a zero, veja:
D(expression(theta1*(1-exp(-exp(theta2)*(x-theta3)))),'x')
curve(4.8292 * (exp(-exp(-5.7353) * (x - 23.452)) * exp(-5.7353)),to=0,from=2000,ylab='y')
(S,f,P)
Allaman