The expression is_not_greater_equal[definitely = tol](x, y) where x and y must be floating point values, evaluates to true if and only if x is definitely not greater than y. This means that:
if tol is a floating value then \(x \le y - \mbox{tol}\cdot \max(|x|, |y|)\)
if tol is a positive integral value then \(x \le \mbox{prev}(y, \mbox{tol})\);
if tol is omitted then the tolerance tol default to 3*eps(as(x)).