Infinite while loop when checking floats for equality [duplicate]

Question

I'm having an infinite loop when I run this code:

intensity = 0.50
while intensity != 0.65:
    print(intensity)
    intensity = intensity + 0.05

intensity values should be like 0.50 -> 0.55 -> 0.60 -> 0.65 then it should gets out of the loop. Why the program is doing an infinite loop instead?

Answer 1

Because of floating point imprecision, you may not end up with exactly 0.65.

To solve this, use < rather than != ^(a):

intensity = 0.50
while intensity < 0.65:
    print(intensity)
    intensity = intensity + 0.05

The output of that (showing what I mean by imprecision as well) is:

0.5
0.55
0.6000000000000001

If you kept going, you'd see:

0.6500000000000001
0.7000000000000002
0.7500000000000002
0.8000000000000003

which is why it's never equal to 0.65.

---

^(a) You may be wondering what the situation would be if the next value was 0.6499...9 rather than 0.650..1, and you would probably be right to be concerned.

While using < will fix the latter case, you'll almost certainly get one iteration too many for the former case.

You can fix that by a number of possible different strategies, some of which are:

• round the number to the correct number of decimals before comparing, such as round(intensity, 2).
• compare it with a "close enough" check, such as the absolute difference being less that 10-4 (for example) - something like if abs(intensity - 0.65) < 0.0001.
• use integral values exclusively and scale the value as needed (e.g., initialising intensity to 50, adding 5, comparing to 65, and printing round(intensity / 100, 2).

The Python documentation has an interesting article that you can read to further understand these limitations.

Answer 2

while True:
    if intensity>0.65:
        break
    print(intensity)
    intensity+=0.05

Answer 3

This is due to the way Python (and some other languages like C) handle floating-point provisions. See this or this. In general, you should avoid floating point loop counters.

If you still wanted to use one, you can round it off:

intensity = 0.50
while round(intensity,2) != 0.65:
    print(round(intensity,2))
    intensity = intensity + 0.05

Answer 4

Look at your output:

0.5
0.55
0.6000000000000001
0.6500000000000001
0.7000000000000002

0.05 cannot be exactly represented with a terminating binary float number.

See is floating point math broken?

If you want to check whether the value is close, then simply set your desired tolerance:

while abs(intensity - 0.65) < my_tolerance:

Better yet, use the built-in function, setting either relative or absolute tolerance:

isclose(intensity, 0.65, rel_tol=1e-9, abs_tol=0.0)