if(Distributiony=="Normal"){
  yy <- y
  aux.formula <- paste("yy~s(x1,",DF3,") + s(x2,",DF4,")")
  right <- gam(aux.formula)
  wrong <- glm(yy~x1+x2) 
  sigmar <- sqrt(sum(right$residuals^2)/right$df.resid)
  sigmaw <- sqrt(sum(wrong$residuals^2)/wrong$df.resid)
  right <-predict.gam(right,newdata=data.frame(x1=Age,x2=CD4),type="response")
  wrong <-predict.gam(wrong,newdata=data.frame(x1=Age,x2=CD4),type="response")

  phi.r <- sigmar^2*alpha + right 
  phi.w <- sigmaw^2*alpha + wrong 
  weights.r <- rep(0,n); weights.r[r==1] <- 1/(1-p0);
  weights.w <- rep(0,n); weights.w[r==1] <- 1/(1-p0.linear)
}
if(Distributiony=="Lognormal"){
   yy <- log(y)
   aux.formula <- paste("yy~s(x1,",DF3,") + s(x2,",DF4,")")
   if(Linear){
     right <- gam(aux.formula)
     wrong <- glm(yy~x1+x2) ## wrogn actually right
     sigmar <- sqrt(sum(right$residuals^2)/right$df.resid)
     sigmaw <- sqrt(sum(wrong$residuals^2)/wrong$df.resid)
     right <-predict.gam(right,newdata=data.frame(x1=Age,x2=CD4),type="response")
     wrong <-predict.gam(wrong,newdata=data.frame(x1=Age,x2=CD4),type="response")
     phi.r <- exp(right + (2*alpha+1)*sigmar^2/2)
     phi.w <- exp(wrong + (2*alpha+1)*sigmaw^2/2)
   }
   else{ 
     right <- gam(aux.formula)
     wrong <- glm(y~x2,family="Gamma")
     shape <- 1/summary(wrong)$dispersion ###i think dispersion is 1/shape
     sigmar <- sqrt(sum(right$residuals^2)/right$df.resid)
     right <-predict.gam(right,newdata=data.frame(x1=Age,x2=CD4),type="response")
     wrong <-rep(wrong$fitted[1],n)
   
     phi.r <- exp(right + (2*alpha+1)*sigmar^2/2)
     beta <- wrong/shape
     phi.w <- (shape+alpha)*beta
   }
   weights.r <- rep(0,n); weights.r[r==1] <- 1/(1-p0);
  weights.w <- rep(0,n); weights.w[r==1] <- 1/(1-p0.linear)
 }