In some cases it may be advantageous to pass problem-specific data to a user-supplied function through the IMSL C Math Library interface. This ability can be useful if a user-supplied function requires data that is local to the user's calling function, and the user wants to avoid using global data to allow the user-supplied function to access the data. Functions in IMSL C Math Library that accept user-supplied functions have an optional argument(s) that will accept an alternative user-supplied function, along with a pointer to the data, that allows user-specified data to be passed to the function. The example below demonstrates this feature using the IMSL C Math Library function imsl_f_min_uncon and optional argument IMSL_FCN_W_DATA.
#include "imsl.h"
#include <math.h>
static float fcn_w_data(float x, void *data_ptr);
static float fcn(float);
void main()
{
float a = -100.0;
float b = 100.0;
float fx, x;
float usr_data[] = {5.0, 10.0};
x = imsl_f_min_uncon (fcn, a, b,
IMSL_FCN_W_DATA, fcn_w_data, usr_data,
0);
fx = fcn_w_data(x, (void*)usr_data);
printf ("The solution is: %8.4f\n", x);
printf ("The function evaluated at the solution is: %8.4f\n", fx);
}
/*
* User function that accepts additional data in a (void*) pointer.
* This (void*) pointer can be cast to any type and dereferenced to
* get at any sort of data-type or structure that is needed.
* For example, to get at the data in this example
* *((float*)data_ptr) contains the value 5.0
* *((float*)data_ptr+1) contains the value 10.0
*/
static float fcn_w_data(float x, void *data_ptr)
{
return exp(x) - (*((float*)data_ptr))*x + *((float*)data_ptr+1);
}
/* Dummy function to satisfy C prototypes. */
static float fcn(float x)
{
return;
}
|
Visual Numerics, Inc. PHONE: 713.784.3131 FAX:713.781.9260 |