Rounding using 'fix'... possibly a bug?

Could someone tell me if this is a bug, or is there some logical reason 'fix' would round:

.3/.1 = 2

whereas

.4/.1 = 4

K>> fix(.3/.1)
ans =
     2
K>> fix(.4/.1)
ans =
     4
K>> fix(.7/.1)
ans =
     6
K>> fix(.5/.1)
ans =
     5

[Jan]

It is such surprising that the well known floating point effects appear for beginners usually for the value 0.3 . Usually it is:

find((0:0.1:1) == 0.3)   % replies [] !!!

Answer 1

[James Tursa]

To see what is going on you can examine the exact floating point values. Note that none of your original numbers except 0.5 can be represented in IEEE double format exactly:

>> num2strexact(.7)
ans =
0.6999999999999999555910790149937383830547332763671875
>> num2strexact(.5)
ans =
0.5
>> num2strexact(.4)
ans =
0.40000000000000002220446049250313080847263336181640625
>> num2strexact(.3)
ans =
0.299999999999999988897769753748434595763683319091796875
>> num2strexact(.1)
ans =
0.1000000000000000055511151231257827021181583404541015625
>> num2strexact(.7/.1)
ans =
6.99999999999999911182158029987476766109466552734375
>> num2strexact(.5/.1)
ans =
5
>> num2strexact(.4/.1)
ans =
4
>> num2strexact(.3/.1)
ans =
2.999999999999999555910790149937383830547332763671875

You can find num2strexact here:

http://www.mathworks.com/matlabcentral/fileexchange/22239-num2strexact-exact-version-of-num2str

[Brad]

Thank you James, much appreciated!

Answer 2

[Image Analyst]

Looks like another case explained by the FAQ.