R help - Oct 2014 - Bug in myintegrate in Elliptic package

Hello,

I was using my integrate today and spend a couple of hours trying to figure
out why I was getting some weird results in my code when using myintegrate
function to do complex integration, after some time decided just to change
the name of the integration dummy variable in the code from 'x' to
'y' and
the code worked perfectly, here is the outpus of what I was using:
>   #Find the CDF to find the call prices
>   CDF_Merton = function(y,psi_measure){
+     integrand_cdf = function(u_vec,y,psi_measure)
Im(exp(-1i*u_vec*y)*psi_measure(u_vec))/u_vec
+   return(1/2 -
1/pi*Re(myintegrate(integrand_cdf,0,100,y=y,psi_measure=psi_measure)))
+   }
>   CDF_Merton(y=0.1,psi_measure=psi)
[1] 0.9368239
>
>   CDF_Merton = function(x,psi_measure){
+     integrand_cdf = function(u_vec,x,psi_measure)
Im(exp(-1i*u_vec*x)*psi_measure(u_vec))/u_vec
+   return(1/2 -
1/pi*Re(myintegrate(integrand_cdf,0,100,x=x,psi_measure=psi_measure)))
+   }
>   CDF_Merton(x=0.1,psi_measure=psi)
[1] 59.03344

where you can see that both codes are exactly the same except for the dummy
variable used to integrate, does any body know what might be going on here?

Thanks

Felipe Parra


-- 

Este mensaje de correo electr?nico es enviado por Quantil S.A.S y puede
contener informaci?n confidencial o privilegiada.

This e-mail is sent by Quantil S.A.S and may contain confidential or
privileged information

	[[alternative HTML version deleted]]

(1) The example is not reproducible since we don't have "psi" or
know
what it is.

(2) The package is called "elliptic" *NOT* "Elliptic".  The
R language
(unlike Micro$oft crap) is case sensitive.

(3) The code was given in a chaotic manner, making it impossible to copy 
and paste it into an R console.

(4) That being said, it's pretty obvious what's going on.  ***Look at
the code*** for "myintegrate()".  It uses "x" as the dummy
variable for
the function "f(x,...)" that is its first argument.  If you use
"x" as
the name of one of the "..." arguments you wind up getting a call of
the form "f(x,x=0.1,psi_measure=psi)" --- no wonder the poor damned 
thing gets confused and gets it wrong.

Try:

    melvin <- function(x,...){x^2}
    melvin(2)
    melvin(2,x=5)

and you'll see the sort of thing that happens.

cheers,

Rolf Turner

-- 
Rolf Turner
Technical Editor ANZJS

On 10/10/14 03:30, Luis Felipe Parra wrote:
> Hello,
>
> I was using my integrate today and spend a couple of hours trying to figure
> out why I was getting some weird results in my code when using myintegrate
> function to do complex integration, after some time decided just to change
> the name of the integration dummy variable in the code from 'x' to
'y' and
> the code worked perfectly, here is the outpus of what I was using:
>
>>    #Find the CDF to find the call prices
>>    CDF_Merton = function(y,psi_measure){
> +     integrand_cdf = function(u_vec,y,psi_measure)
> Im(exp(-1i*u_vec*y)*psi_measure(u_vec))/u_vec
> +   return(1/2 -
> 1/pi*Re(myintegrate(integrand_cdf,0,100,y=y,psi_measure=psi_measure)))
> +   }
>>    CDF_Merton(y=0.1,psi_measure=psi)
> [1] 0.9368239
>
>>
>>    CDF_Merton = function(x,psi_measure){
> +     integrand_cdf = function(u_vec,x,psi_measure)
> Im(exp(-1i*u_vec*x)*psi_measure(u_vec))/u_vec
> +   return(1/2 -
> 1/pi*Re(myintegrate(integrand_cdf,0,100,x=x,psi_measure=psi_measure)))
> +   }
>>    CDF_Merton(x=0.1,psi_measure=psi)
> [1] 59.03344
>
> where you can see that both codes are exactly the same except for the dummy
> variable used to integrate, does any body know what might be going on here?