Friday, February 10, 2012

MATLAB - realmin Problem or BUG

A new problem/bug found in MATLAB

Real max of MATLAB is:
real max = 2 ^ 1024 - 1
That is the same as realmax of MATLAB!
To calculate you should get the series as 2 ^ 1023 + 2 ^ 1022 + 2 ^ 1021 + 2 ^ 1020 + ...

Real min of MATLAB is:
real min = 2 ^ -1074
BUT realmin of MATLAB = 2 ^ -1022
To prove the real min of MATLAB, you could test so many equation. You'll get that no change in the value!
Because the assigned bits to this floating point system is 53 = ( 1074 - 1022 ) + 1

By:
Diman Zad Tootaghaj
Farshid Farhat

No comments: