Vejam o seguinte script: rm(list=ls(all=TRUE)) library(minpack.lm) dados<-read.table(header = TRUE, dec=",",text= "psi theta 0,01 0,5 0,5 0,48 1 0,42 3 0,4 10 0,39 30 0,35 100 0,3 300 0,28 1000 0,27 ") dados theta=dados$theta psi=dados$psi durner<-as.formula(theta~0.6*(thetar+((thetas-thetar)/(1+(alpha0*psi)^n0)^(1-1/n0)))+ 0.4*(thetar+((thetas-thetar)/(1+(alpha1*psi)^n1)^(1-1/n1)))) dur<-nlsLM(durner, start=list(thetas=0.5,thetar=0.25,alpha0=0.1,n0=2,alpha1=0.1,n1=1),data = dados, algorithm = "LM", trace = "F", control = list(maxiter = 500)) summary(dur) rsquared<-1-(sum(residuals(dur)^2)/sum((theta - mean(theta))^2)) rsquared thetas=coef(dur)[1] thetar=coef(dur)[2] alpha0=coef(dur)[3] n0=coef(dur)[4] alpha1=coef(dur)[5] n1=coef(dur)[6] par(mar=c(4,4,1,1)) par(mfrow = c(1, 1)) plot(psi,theta,xlim = c(min(psi),max(psi)),log = "x",las=1,xaxt="n",ylab=expression(theta~(m^{3}~m^{-3})),xlab=expression(abs(psi)~(kPa)),mgp=c(2.5, 1, 0)) labels <- 10^(-3:3) axis(1, at=labels, labels=labels,lwd.ticks = 2) #,tck=-0.05 axis(1, at=sort(sapply(1:9, function(x) x*10^(-3:3))), label=NA, lwd=0.5) curve(0.6*(thetar+((thetas-thetar)/(1+(alpha0*psi)^n0)^(1-1/n0)))+ 0.4*(thetar+((thetas-thetar)/(1+(alpha1*psi)^n1)^(1-1/n1))),add = T,xname = "psi") Gostaria de, ao invés de dar valores (0.6 e 0.4): durner<-as.formula(theta~0.6*(thetar+((thetas-thetar)/(1+(alpha0*psi)^n0)^(1-1/n0)))+ 0.4*(thetar+((thetas-thetar)/(1+(alpha1*psi)^n1)^(1-1/n1)))) adicionar dois parâmetros (w0 e w1): durner<-as.formula(theta~w0*(thetar+((thetas-thetar)/(1+(alpha0*psi)^n0)^(1-1/n0)))+ w1*(thetar+((thetas-thetar)/(1+(alpha1*psi)^n1)^(1-1/n1)))) mas de tal forma que: w0+w1==1. Alguém tem sabe como poderia fazer isso? Att., Delcio Bortolanza Doutorando em Agronomia-UPF