Why doesn't the less than operator work inside a for loop for decimals

Im trying to get the following code to work. but every time the value of x goes below 0.6 even though it an if condition is set to prevent it..can someone help

 x=0.8;
for n=1:10

   if (x>0.6)

       x=x-0.1;

   end
   x
end

Answer 1

[James Tursa]

Floating point arithmetic effects. See this link for starters:

https://www.mathworks.com/matlabcentral/answers/57444-faq-why-is-0-3-0-2-0-1-not-equal-to-zero

For your specific example,

>> x = 0.8
x =
    0.8000
>> num2strexact(x)
ans =
0.8000000000000000444089209850062616169452667236328125
>> x = x - 0.1
x =
    0.7000
>> num2strexact(x)
ans =
0.70000000000000006661338147750939242541790008544921875
>> x = x - 0.1
x =
    0.6000
>> num2strexact(x)
ans =
0.600000000000000088817841970012523233890533447265625
>> num2strexact(0.6)
ans =
0.59999999999999997779553950749686919152736663818359375
>> x > 0.6
ans =
     1

So, the successive subtractions of 0.1 don't get you exactly to the value you were expecting. The decimal fractions you are using can't even be represented exactly in IEEE double arithmetic anyway. You need to use different logic in your looping. E.g.,

y = 8;  % 8 can be represented exactly in IEEE double
for n=1:10
   if( y > 6 )
       y = y - 1; % This subtraction will be done exactly in IEEE double
   end
   x = y / 10; % Now calculate the x value to use downstream
     :
end

[Shirosh Peiris]

Hi James

Thanks a lot for your answer..it makes sense now.