[R] Error in integrate(int.fn, lower = 0, upper = Inf) : evaluation of function gave a result of wrong length

[Berend Hasselman]

On 09-04-2012, at 14:54, Guaramy wrote:

> I read it believe me . The reason that a i post this is because i am making a
> thesis and a i am having this problem for over 2 weeks.
> I can´t solve it and its causing me real problems.


You didn't give us a reproducible example.
I'm assuming your original function is

GNL.pdf.fn <- function(x,mu,sigma,alpha,beta,rho) # missing in your original post
{
    y  <-  x-rho*mu

    cf.fn  <-  function(s){
        cplex = complex(1,0,1)
        temp1 = alpha*beta*exp(-sigma*s^2/2)
        temp2 = (alpha-cplex*s)*(beta+cplex*s)
        out = (temp1/temp2)^rho
        out
    }

    temp.fn  <-  function(s){ Mod(cf.fn(s))*cos(Arg(cf.fn(s))-s*y) }

    int.fn  <-  function(t){sapply(t,FUN=temp.fn)}
    te  <-  integrate(int.fn,lower=0,upper=Inf,rel.tol=1e-10,subdivisions=1000000)
    temp3  <-  ifelse(te$message == "OK",te$value/pi,NA)
    temp3
}

Question: why are you using ifelse when te$message is a scalar?

if( te$message == "OK" ) te$value/pi else  NA

would do what you want.

David meant that you should have a look at Vectorize. And then do some experimenting.
So I tried this

GNL.pdf.fn(1, .2, .3, 1, 1, .6)

GNL.vec <- Vectorize(GNL.pdf.fn, "x")

GNL.vec(c(.9,1,1.1), .2, .3, 1, 1, .6)

And now it's your turn for seq(-4,4,0.1).

Berend