Rounding error for simple array

I have noticed a very strange issue using R2012a installed on Windows 7.

I was creating a very simple array a = [0.2:.1:1.2]'

I then searched for the index where a == 0.6 and kept getting no hit.

Confused, I looked at 'a' in the command window (in format long e) and was shocked to see:

a =

     2.000000000000000e-01
     3.000000000000000e-01
     4.000000000000000e-01
     5.000000000000000e-01
     6.000000000000001e-01
     7.000000000000000e-01
     7.999999999999999e-01
     8.999999999999999e-01
     1.000000000000000e+00
     1.100000000000000e+00
     1.200000000000000e+00

0.6 , 0.8 and 0.9 are all off by the final machine error digit.

ANy idea what is causing this? Is this a known bug?

Update: To throw more fuel on the fire doing a = [2:1:12]./10 seems to work just fine.

Answer 1

[dpb]

See

http://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

or the Matlab FAQ

http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F