Find Command not working.

Hi everyone . I have a matrix with first row as T=[NaN 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9]. When I use the command now if i have a=0.6 and i try to find index of 'a' as V=find(T==a). it gives me an answer as :

1×0 empty double row vector

How to solve this ?

Answer 1

[Chunru]

use abs(difference) to comapare the float numbers.

T=[NaN 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9];
a=0.6;
V=find(abs(T-a)<1e-10)

V = 7

[Muhammad Owais]

Yes Thanks . This worked!

Answer 2

[Image Analyst]

It's probably not exactly 0.6. Try ismembertol().

Answer 3

[Walter Roberson]

T = [NaN 0.1:0.1:0.9]

T = 1×10

NaN 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000 0.9000

a = 0.3;

Visually, it looks like 0.3 is in T

find(T == a)

ans =

  1×0 empty double row vector

But find() says it is not there.

Is it there? Let's look at the bit representation

fprintf('a:\n'); arrayfun(@pX, a);

a:
3FD3333333333333

So for the literal, the bit representation ends in all hex 3's.

Is that bit representation present in T?

fprintf('\nT:\n'); arrayfun(@pX, T);

T:
FFF8000000000000
3FB999999999999A
3FC999999999999A
3FD3333333333334
3FD999999999999A
3FE0000000000000
3FE3333333333333
3FE6666666666666
3FE999999999999A
3FECCCCCCCCCCCCD

No. There is a very similar bit representation that ends in a 4 rather than a 3. The last 4 bits of the actual value are 0011 (decimal 3), but the closest that T has is something for which the last 4 bits are 0100 (decimal 4).

When you calculate something two different ways, the bits are not necessarily going to come out identical.

function pX(N)
    fprintf('%016X\n', typecast(N, 'uint64'));
end

[Image Analyst]

A thorough explanation. There is also an explanation in the FAQ:

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

This is stuff you would have learned in your numerical analysis/linear algebra class at the university if you had taken such a class.