Dissect floating point numbers. IEEE logb().
y = LOG2(x) computes the logarithm base 2 of the elements of x.
[f,e] = LOG2(x) for a real matrix x, returns a matrix f of real
numbers, usually in the range 0.5 <= abs(f) < 1, and a matrix
e of integers, so that x = f .* 2.^e. Any zeros in x produce
f = 0 and e = 0. This corresponds to the ANSI C function frexp()
and the IEEE floating point standard function logb().
See also POW21BY81U2, NEXTPOW2 , REALMAX , REALMIN .