This example shows how to write a MATLAB function that uses structure arrays so that it is suitable for code generation. For code generation, you must first create a scalar template version of the structure before growing it into an array. The code generation inference engine uses the type of this scalar value as the base type of the array.
There are no prerequisites for this example.
The following code will create a folder in your current working folder (pwd). The new folder will only contain the files that are relevant for this example. If you do not want to affect the current folder (or if you cannot generate files in this folder), you should change your working folder.
The struct_array.m file uses a structure array.
% y = struct_array(n) % Take an input scalar number 'n' which will designate the size of the % structure array return. function y = struct_array(n) %#codegen % Copyright 2010-2013 The MathWorks, Inc. assert(isa(n,'double')); % Input is scalar double % To create a structure array you start to define the base scalar element % first. Typically, we initialize all the fields with "dummy" (or zero) % values so the type/shape of all its contents are well defined. s.x = 0; s.y = 0; s.vx = 0; s.vy = 0; % To create a structure array of fixed size you can do this in multiple % ways. One example is to use the library function 'repmat' which takes a % scalar element and repeats it to its desired size. arr1 = repmat(s, 3, 5); % Creates a 3x5 matrix of structure 's' % At this point you can now modify the fields of this structure array. arr1(2,3).x = 10; arr1(2,3).y = 20; arr1(2,4).x = 5; arr1(2,4).y = 7; % Another way of creating a structure array of fixed size is to use the % concatenation operator. arr2 = [s s s; s s s; s s s; s s s; s s s]; % If two variables agree on base type and shape you can copy one structure % array to the other using standard assignment. arr2 = arr1; % To create a structure array of variable size with a known upper bound can % be done in multiple ways as well. Again, we can use repmat for this, but % this time we will add a constraint to the (non constant) input variable. % This guarantees that the input 'n' of this function is less than or equal to 10. assert(n <= 10); % Create a row vector with at most 10 elements of structures based on 's' arr3 = repmat(s, 1, n); % Or we can use a for-loop with the concatenation operator. The compiler is % unable to analyze that 'arr4' will be at most 10 elements big, so we % add a hint on 'arr4' using coder.varsize. This will specify that the % dimensions of 'arr4' is exactly one row with at most 10 columns. Look at % the documentation for coder.varsize for further information. coder.varsize('arr4', [1 10]); arr4 = repmat(s, 1, 0); for i = 1:n arr4 = [arr4 s]; end % Let the top-level function return 'arr4'. y = arr4;
In MATLAB, when building up a structure array, you would typically just add fields as you go. For example, s(1).x = 10; s(2).y = 20; This "dynamic" style of building structures is not supported for code generation. One reason is that it is possible in MATLAB to have different structure fields for two different elements of a structure array, which conflicts with the more static approach of type inference. Therefore, you need to fully specify the base scalar element first, and then grow a structure array from this fully specified element. This method guarantees that two elements of a structure array always share the same type (fields).
Generate a MEX function using the command
codegen followed by the name of the MATLAB file to compile.
By default, 'codegen' generates a MEX function named 'struct_array_mex' in the current folder. This allows you to test the MATLAB code and MEX function and compare the results.
ans = 1x10 struct array with fields: x y vx vy
Remove files and return to original folder