adding two numpy.float64 numbers get wrong result-CodePudding

I am new to numpy. I tried to add two numpy.float64 numbers, however, the numpy gave me a wrong answer.

Here's my code:(a and b are generated separately by the mean of one numpy array)

print(a)
print(type(a))
print(b)
print(type(b))
c = (a   b)/2
print(c)
print(type(c))

And here's what I got:

1.4617937201411304
<class 'numpy.float64'>
-1.4617937201411302
<class 'numpy.float64'>
1.1102230246251565e-16
<class 'numpy.float64'>

Clearly, the answer should be very close to 0, but what I got is far away from 0.

I guess the problem is that a b caused an overflow?

However, when I tried to double a, the answer was right:

c = (a * 2)/2

The answer is:

1.4617937201411304
<class 'numpy.float64'>
-1.4617937201411302
<class 'numpy.float64'>
1.4617937201411304
<class 'numpy.float64'>

CodePudding user response：

You missed the e-16 at the end of the print. It means the result is 1.1102230246251565 * 10^(-16), which is indeed almost zero. It's a convention to print numbers with e (the exponent) rather then 0.000000....1110