Floating Point Comparisons in Matlab

A common problem in Matlab and every other programming language that uses floating point numbers is that calculations involving floats often do not yield the expected answers because of rounding, which can have undesirable effects on control statements. The immediate question is how to handle these rounding errors so that intuitively correct statements are recognized as true by a program. We would like to accomplish this while still retaining as much relevant information in the numbers as possible, thereby allowing the detection of minute differences that are not artifacts of rounding. In this post, some common errors in floating point comparisons will be discussed, as well as methods of handling these errors.

Continue reading