How do I make my code compatible with both single and double precision mode?

In order to generate single-precision code with the "-singleC" flag, I have to manually change some of my functions. For example, whenever I use the "realmax" function, I have to manually change every instance of "realmax('double')" to "realmax('single')".

How can I automate this process? What is the best workflow to easily convert back and forth between single and double precision?

Answer 1

[MathWorks Support Team]

You should utilize the "like" syntax supported by a number of functions. The documentation for the "zeros" function at the following link describes this workflow: https://www.mathworks.com/help/fixedpoint/ref/zeros.html#btuia59-1

In your case, this workflow could be applied as follows:

realmax('like', 0)

specifying that the "realmax" constant should have the same type as "0", i.e., single-precision when code is generated in single-precision mode (as with the "-singleC" flag), or double-precision otherwise.

This same workflow is also recommended for easy conversion between floating and fixed point types.