• Introduction
• ANSI C vs. Non-ANSI C and IMSL.h
• Matrix Storage Modes for the MATH and STAT Libraries
• Memory Allocation for Output Arrays
• Finding the Right Function
• Organization of the Documentation
• Naming Conventions
• Error Handling, Underflow and Overflow
• Printing Results
• Missing Values
• Complex Arithmetic

All of the examples in this documentation conform to ANSI C. If you are not using ANSI C, you will need to modify your examples in functions that are declared or in those arrays that are initialized as type float.

Non-ANSI C does not allow for automatic aggregate initialization, and thus all auto arrays that are initialized as type float in ANSI C must be initialized as type static float in non-ANSI C. The following pro gram contains arrays that are initialized as type float and also a user-defined function:

1. #include <imsls.h>
   
2. 
   
3. float           fcn(int, float[], int, float[]);
   
4. 4
   
5. main()
   
6. {
   
7.      int         n_observations = 3,
   
8.                  n_parameters = 1,
   
9.                  n_independent = 1;
   
10.     float       *theta_hat;
    
11.     float       x[3] = {1.0, 2.0, 3.0};
    
12.     float       y[3] = {2.0, 4.0, 3.0};
    
13.                      /* Evaluate the integral */
    
14.     theta_hat = imsls_f_nonlinear_regression(fcn, n_parameters,
    
15.                 n_observations, n_independent, x, y, 0);
    
16.                      /* Print the result and the exact answer */
    
17.     imsls_f_write_matrix("estimated coefficient", 1, 1, theta_hat, 0);
    
18. }
    
19. float fcn(int n_independent, float x[], int n_parameters,
    
20.            float theta[])
    
21. {
    
22.     return exp(theta[0]*x[0]);
    
23. }
    If using non-ANSI C, you will need to modify lines 3, 11, 12, and on as follows:
    3  float          fcn(); /* Function is not prototyped */
         .
         .
         .
    11    static float       x[3] = {1.0, 2.0, 3.0};
    12    static float       y[3] = {2.0, 4.0, 3.0};
         .
         .
         .
    19  float fcn(n_independent, x, n_parameters,
    20            theta)     /*Declaration of variable names*/
    20a int n_independent;
    20b float x[];
    20c int n_parameters;
    20d float theta[];       /*Type definitions of variables*/