Float types (float16, float32, float64) in NumPy

Introduction

Float types let us store decimal numbers with different levels of detail and memory use. They help balance speed, memory, and precision in calculations.

When you need to save memory by using fewer bytes for decimal numbers.

When you want more precise decimal calculations for scientific data.

When working with large datasets where speed and memory matter.

When you want to control the accuracy of your calculations.

When you need to match data types with other systems or files.

Syntax

NumPy

numpy.float16
numpy.float32
numpy.float64

These are data types in NumPy for floating-point numbers.

float16 uses 2 bytes, float32 uses 4 bytes, and float64 uses 8 bytes.

Examples

Create a NumPy array with float16 type and print values and type.

NumPy

import numpy as np

x = np.array([1.5, 2.5], dtype=np.float16)
print(x)
print(x.dtype)

Create a NumPy array with float32 type and print values and type.

NumPy

import numpy as np

y = np.array([1.5, 2.5], dtype=np.float32)
print(y)
print(y.dtype)

Create a NumPy array with float64 type and print values and type.

NumPy

import numpy as np

z = np.array([1.5, 2.5], dtype=np.float64)
print(z)
print(z.dtype)

Sample Program

This program creates three arrays with the same numbers but different float types. It prints each array and its type so you can see the difference.

NumPy

import numpy as np

# Create arrays with different float types
arr16 = np.array([0.1, 0.2, 0.3], dtype=np.float16)
arr32 = np.array([0.1, 0.2, 0.3], dtype=np.float32)
arr64 = np.array([0.1, 0.2, 0.3], dtype=np.float64)

# Print arrays and their data types
print('float16 array:', arr16)
print('float16 dtype:', arr16.dtype)

print('float32 array:', arr32)
print('float32 dtype:', arr32.dtype)

print('float64 array:', arr64)
print('float64 dtype:', arr64.dtype)

Important Notes

float16 uses less memory but can lose some precision in numbers.

float64 is the most precise and is the default float type in NumPy.

Choose the float type based on your need for precision and memory limits.

Summary

Float types store decimal numbers with different precision and memory use.

float16 uses 2 bytes, float32 uses 4 bytes, and float64 uses 8 bytes.

Pick the right float type to balance accuracy and memory for your data.