Round to nearest integer towards zero in Python

To round to the nearest integer towards zero in Python, use the numpy.fix() method. It performs element-wise rounding of float arrays towards zero, which means it truncates the decimal part. For positive numbers, this rounds down, and for negative numbers, it rounds up towards zero.

Syntax

numpy.fix(x, out=None)

Parameters

The function accepts the following parameters:

• x ? Array of floats to be rounded
• out ? Optional output array where results are stored

Basic Example

Let's see how np.fix() rounds different types of numbers ?

import numpy as np

# Create an array with positive and negative floats
arr = np.array([120.6, -120.6, 200.7, -320.1, 320.1, 500.6])

print("Original array:")
print(arr)

print("\nRounded towards zero:")
print(np.fix(arr))

Original array:
[ 120.6 -120.6  200.7 -320.1  320.1  500.6]

Rounded towards zero:
[ 120. -120.  200. -320.  320.  500.]

How It Works

The numpy.fix() method truncates the fractional part, always rounding towards zero ?

import numpy as np

# Examples showing rounding behavior
examples = np.array([3.8, -3.8, 7.2, -7.2, 0.9, -0.9])

print("Original values:", examples)
print("Rounded towards zero:", np.fix(examples))
print("Regular rounding:", np.round(examples))

Original values: [ 3.8 -3.8  7.2 -7.2  0.9 -0.9]
Rounded towards zero: [ 3. -3.  7. -7.  0. -0.]
Regular rounding: [ 4. -4.  7. -7.  1. -1.]

Using Output Parameter

You can specify an output array to store results instead of creating a new array ?

import numpy as np

data = np.array([15.7, -22.3, 8.9])
result = np.zeros_like(data)

# Store result in existing array
np.fix(data, out=result)

print("Original:", data)
print("Result array:", result)
print("Data type:", result.dtype)

Original: [ 15.7 -22.3   8.9]
Result array: [ 15. -22.   8.]
Data type: float64

Comparison

Conclusion

Use numpy.fix() when you need to truncate decimal parts towards zero. This is particularly useful for converting floats to integers while preserving the sign and avoiding rounding effects.