Why does comparing identical doubles result in a logical 0?

Hi there,

I have the problem that the comparison of doubles which I have exported from another software (COMSOL Multiphysics) with doubles of identical value result in a logical 0. How can I fix this? I've documented the problem above by copying the contents of my Command Window below:

==============

v =

     1.0e+03 *

      2.5000
      2.0000
      0.6000

>> v(2)

ans =

     2.0000e+03

>> v(2)==2000

ans =

       0

>> w=double(2000);
>> v(2)==w

ans =

       0

>> whos v
  Name      Size            Bytes  Class     Attributes

    v         3x1                24  double              

>> whos w
  Name      Size            Bytes  Class     Attributes

    w         1x1                 8  double              

>> v(2)==2.0000e+03

ans =

       0

>> 
==============

Any help would be highly appreciated.

Thanks a lot in advance, Joerg

Answer 1

[Jan]

When you observe, that v(2)==2000, it is a straight idea to ask Matlab for the difference between these numbers:

v(2) - 2000

This is an effect of the limited number of digits in the output in the Command Window, and the limited precision of floating point values, see: http://www.mathworks.com/matlabcentral/answers/57444-faq-why-is-0-3-0-2-0-1-not-equal-to-zero

[Joerg]

Thanks for the link. I didn't find that thread, because I was searching for "comparing doubles". I will read it straight away.

Answer 2

[Wayne King]

This is the well-known (and often treated in this forum) problem of floating-point precision. The point is that they are not identical.

You can use a tolerance to compare the numbers like abs(x-y)<lambda

where lambda is some very small number.

See the help for eps()

[Joerg]

Thank you very much. I didn't know about this function. I will try the tolerance approach.

Answer 3

[Shashank Prasanna]

If you want to see how they are different, try:

>> format hex
>> v(2),2000

change display format back to default

>> format