Error while using variables inside "ismember"

The "ismember" function is not reading values when they are assigned to variables after simple arithmertic operations.

a=csvread('....\ATR.csv');  %it is a 230x2 array (acc. the data set I'm using)
a=unique(a,'rows','stable') %it is reduced to 176x2 array 
ismember([69.7 21.5],a,'rows') % returns true as the matrix specified is in a 
ismember([69.6+0.1 21.4+0.1],a,'rows') %returns false, but the data point after the arithmetic operations exists in a 
ax=69.6;
ay=21.6; 
ismember([ax ay],a,'rows') % returns true as the matrix specidied is in my data 
ax=ax+0.1; ay=ay-0.1; 
ismember([ax ay],a,'rows') % returns false but the matrix specified is in my data

[Alex Mcaulley]

See this:

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

[Adam Danz]

I bet it's a precision issue. Use ismembertol() instead.

[Sarthak De]

Thanks guys, both of your inputs really helped me understand and solve the problem !