Structures

A structure is a data type which contains several values, possibly of different types, referenced by name. The simplest way to create a structure is by simple assignment. For example, consider the function

The following m-file f.m computes the value, gradient, and Hessian of f at a point x, and returns them in a structure:

function fx = f(x)

fx.Value = (x(1)-1)^2+x(1)*x(2);
fx.Gradient = [2*(x(1)-1)+x(2);x(1)];
fx.Hessian = [2 1;1 0];

We can now use the function as follows:

>> x = [2;1]
x =
     2
     1
>> fx = f(x)
fx = 
       Value: 3
    Gradient: [2x1 double]
     Hessian: [2x2 double]
>> whos
  Name      Size         Bytes  Class

  fx        1x1            428  struct array
  x         2x1             16  double array

Grand total is 12 elements using 444 bytes

The potential of structures for organizing information in a program should be obvious.

Note that, in the previous example, Matlab reports fx as being a ``struct array''. We can have multi-dimensional arrays of structs, but in this case, each struct must have the same field names:

>> gx.Value = 12;
>> gx.Gradient = [2;1];
>> A(1,1) = fx;
>> A(2,1) = gx;
??? Subscripted assignment between dissimilar structures.

>> fieldnames(fx)
ans = 
    'Value'
    'Gradient'
    'Hessian'
>> fieldnames(gx)
ans = 
    'Value'
    'Gradient'

(Note the use of the command fieldnames, which lists the field names of a structure.)

Beyond simple assignment, there is a command struct for creating structures. For information on this and other commands for manipulating structures, see help struct.