Help using old version of Matlab Student

Hi, I periodically use an old version 5 student edition. I'm having a rather small err large problem.

This is what happens

a=[.95 .03 .05 .97]

a =

    0.9500    0.0300
    0.0500    0.9700

b=a-eye(2)

b =

   -0.0500    0.0300
    0.0500   -0.0300

det(b)

ans =

2.7756e-018

clear

b=[-.05 .03 .05 -.03]

b =

   -0.0500    0.0300
    0.0500   -0.0300

det(b) ans =

     0

What's happening here? Thanks.

[Manoj Kumar]

What's wrong ?

det(b) = 0 is equivalent to det(b) = 2.7756 x 10^*(-18) =0.

Answer 1

[Daniel Shub]

This is FAQ 6.1 wrapped up in a different way. FAQ 6.1 considers

>> 0.3 - 0.2 - 0.1
ans =
-2.7756e-17

The answer is the limited precision of floats. Using the same approach outlined in the FAQ gives:

>> sprintf('%20.18f\n', [.95 .03; .05 .97]-eye(2))
ans =
-0.050000000000000044
0.050000000000000003
0.029999999999999999
-0.030000000000000027

>> sprintf('%20.18f\n', [-.05 .03 .05 -.03])
ans =
-0.050000000000000003
0.029999999999999999
0.050000000000000003
-0.029999999999999999

Answer 2

[Manoj Kumar]

What's wrong ?

det(b) = 0 is equivalent to det(b) = 2.7756 x 10^*(-18) =0.