Shouldn't just
(*f)(x,z)
work? Of course z needs to be defined in NDint().
A weaker option is to use -external- declarations. That makes the code
messy and occasion-specific, which given your style you probably won't
like :))
On 6/16/09, Nathan Danneman <[email protected]> wrote:
> Hi all,
>
> I have written a function for a part of a larger program. This
> program generates a numeric integral for a function that it calls. I
> would like to allow the outter function to pass several optional
> arguments (between zero and seven) on to the called function which is
> being integrated, which could then pass those arguments on to another
> lower-level function...etc. The code for the integration function is
> below. In the pasted example, I integrate the function named "trial"
> at bottom. As a very simple example, how could I get the overall
> function, NDint, to pass the z parameter on to "trial"?
>
> Thanks,
> Nathan
>
>
> //An n-slice integration routine using simpson's approximation
>
> mata
> mata clear
> mata set matalnum off
>
> scalar NDint(
> // to perform integration via simpson'sapproximation
> pointer(real scalar function) scalar f,
> //points to the functionto be integrated
> real scalar lower,
> //lower bound of integration
> real scalar higher,
> //upper bound
> real scalar m) // where m = number of slices, must be an EVEN number
> {
>
> if ( (m/2) != (floor(m/2)) ) _error("m must be an EVEN number")
>
> real scalar width //I like declarations, your mileage may vary
> real colvector Y
> real rowvector X
> real scalar n //slices plus one
>
> n = m+1
>
> if (lower > higher) _error("limits of integration reversed")
> //make sure upper > lower NB: better to use _error() than
> return()
>
> width = (higher-lower) / m
> //calculates width of each "slice"
>
> Y = J(n,1,.)
> //creates an empty 15 x 1 column vector
> for (i=1; i<=n; i++) {
> x = lower + ((i-1)*width)
> Y[i,1] = (*f)(x)
> //places function evaluations directly into Y,
> avoiding all the copying you were doing previously
> }
>
> X = J(1,n,.)
> for (i=1;i<=n; i++) {
> if (i==1) X[1,i] = 1
> if (i==n) X[1,i] = 1
> if ((i/2) == (floor(i/2))) X[1,i] = 4
> if ( ((i+1)/2) == (floor( ((i+1)/2)) ) & (i!=1) & (i!=n) ) X[1,i] = 2
> }
>
>
> return(X*Y*width/3) //matrix multiplication makes this quicker
> and more readable
>
> }
>
> scalar trial(real scalar x) {
> //declares the output to be a scalar. otherwise
> you could run into problems in the main program
> return(z*exp(x))
> }
>
> f = &trial()
>
> NDint(f, 0, 4, 60)
>
> end
>
>
>
> --
> "Imagination is more important than knowledge..."
> -- Albert Einstein
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
