JIYIK CN >

NumPyis a Python package that contains a rich set of utilities for manipulating large multidimensional matrices and arrays and performing complex and straightforward mathematical operations on them.

These utilities are dynamic on input and are highly optimized and fast. NumPyThe package has a function exp()that computes the exponential of all elements of an input numpy array.

In other words, it computes e ^x , xfor each number in the input numpy array, eto be approximately equal 2.71828to the Euler number .

Because this calculation can produce huge numbers, some data types cannot handle such large values, so this function returns infand error instead of a valid floating point value.

For example, for numpy.exp(709), this function will return 8.21840746e+307, but for numpy.exp(710), it will return runtimeWarning：exp inf 中遇到溢出.

In this article, we will learn how to solve this problem.

Fix numpy.exp()overflow in Python NumPy function

We have to store the value in a data type that can hold such large values ​​to solve this problem.

For example, np.float128can hold numbers larger than float64and float32. All we have to do is typecast each value of the array to a larger data type and store it in a numpy array.

The following Python code describes this.

import numpy as np

a = np.array([1223, 2563, 3266, 709, 710], dtype=np.float128)
print(np.exp(a))

Output:

[1.38723925e+0531 1.24956001e+1113 2.54552810e+1418 8.21840746e+0307
 2.23399477e+0308]

Even though the Python code above runs seamlessly without any issues, we are still prone to the same mistakes.

The reason behind this is simple; even np.float128has a threshold of the number it can hold. Every data type has an upper limit, and if you exceed that upper limit, things start going wrong and your program starts giving overflow errors.

To understand the above mentioned point, refer to the following Python code. Although np.float128our problem is solved in the last Python code snippet, it does not work for larger values.

import numpy as np

a = np.array([1223324, 25636563, 32342266, 235350239, 27516346320], dtype=np.float128)
print(np.exp(a))

Output:

<string>:4: RuntimeWarning: overflow encountered in exp
[inf inf inf inf inf]

exp()The function returns an infinity for each value in the numpy array.