NumPy random.randint(): Generating Random Integers & Bounded Arrays

1. Syntax & Formal Parameter Specifications 🔝

The formal signature for numpy.random.randint() is:

# Formal signature
numpy.random.randint(low, high=None, size=None, dtype=int)

2. Single Argument: Sampling from [0, low) 🔝

When you pass only one integer argument, NumPy assigns it as the exclusive upper bound high, and sets the lower bound to 0:

import numpy as np

np.random.seed(42)

# Generates an integer from {0, 1, 2, 3}
val = np.random.randint(4)

print("Random integer [0, 4):", val)
print("Type:                  ", type(val))

Terminal Output:

Random integer [0, 4): 2
Type:                   <class 'int'>

3. Explicit Lower and Upper Bounds: [low, high) 🔝

Supplying two positional numbers sets both the inclusive start and exclusive ceiling:

import numpy as np

np.random.seed(10)

# Draws from integers: 4, 5, 6, 7, 8, 9
val = np.random.randint(4, 10)
print("Random integer in [4, 10):", val)

Terminal Output:

Random integer in [4, 10): 5

4. Specifying Array Shapes: 1D Vectors & 2D Matrices 🔝

Use the size parameter to return an ndarray rather than a single integer. Passing an integer creates a 1D vector; passing a tuple creates multidimensional matrices:

import numpy as np

np.random.seed(100)

# 1D vector of 5 integers from [4, 10)
vec = np.random.randint(4, 10, size=5)
print("1D Vector (5 items):
", vec)

# 2D matrix (3 rows, 2 columns)
mat = np.random.randint(4, high=10, size=(3, 2))
print("
2D Matrix (3 rows x 2 cols):
", mat)
print("Shape:", mat.shape)

Terminal Output:

1D Vector (5 items):
 [4 7 7 7 4]

2D Matrix (3 rows x 2 cols):
 [[6 8]
 [9 5]
 [9 4]]
Shape: (3, 2)

5. Controlling Precision & Memory with dtype 🔝

By default, randint() returns standard C-long integer arrays (int64 or int32 depending on OS architecture). Specify smaller types like 'int16', 'int8', or 'uint8' to conserve RAM in high-throughput pipelines:

import numpy as np

np.random.seed(20)

arr16 = np.random.randint(4, high=7, size=(3, 3), dtype='int16')
print("3x3 Matrix with dtype='int16':
", arr16)
print("Dtype:    ", arr16.dtype)
print("Item size:", arr16.itemsize, "bytes")

Terminal Output:

3x3 Matrix with dtype='int16':
 [[5 4 5]
 [5 5 5]
 [6 6 5]]
Dtype:     int16
Item size: 2 bytes

6. Practical Production Recipes 🔝

Recipe A: Random RGB Color Generator

Color channels in digital images are standard 8-bit unsigned integers from 0 to 255. Generate a random (Red, Green, Blue) color tuple using np.random.randint() with an exclusive upper bound of 256:

import numpy as np

# 256 is exclusive, so values are in {0, 1, ..., 255}
color_rgb = tuple(np.random.randint(0, 256, size=3))
print("Generated RGB Color:", color_rgb)

Recipe B: Simulating Fair Dice Rolls

Simulate rolling two 6-sided dice 1,000 times and compute sum frequencies:

import numpy as np

np.random.seed(42)
dice_rolls = np.random.randint(1, 7, size=(1000, 2))
totals = dice_rolls.sum(axis=1)

print("Average roll total (expected ~7.0):", np.mean(totals))
print("Minimum total:", totals.min(), "Maximum total:", totals.max())

7. Modern Replacement: default_rng().integers() 🔝

In modern NumPy (version 1.17 and later), the central default_rng() Generator provides integers(), which adds an explicit endpoint=True option to support inclusive upper bounds:

import numpy as np

rng = np.random.default_rng(seed=123)

# endpoint=True allows 6 to be included
die_rolls = rng.integers(low=1, high=6, endpoint=True, size=6)
print("Modern Generator die rolls [1, 6]:", die_rolls)

8. Common Developer Pitfalls 🔝

• Off-by-One Error on High: In np.random.randint(1, 10), the value 10 is never returned. If you need integers up to 10 inclusive, set high=11.
• Confusing with Python's random.randint(): Python's built-in random.randint(a, b) is inclusive of both endpoints, whereas NumPy's np.random.randint(a, b) is exclusive of b.
• Inverted Bounds: Supplying low >= high raises a ValueError: low >= high.

9. Interactive Practice Companion Notebook 🔝

Practice generating integer arrays, testing boundary behaviors, and running unit test verification suites in our companion notebook:

Interactive Practice: Colab & GitHub

Launch the notebook directly in Google Colab or view the source notebook on GitHub:

View on GitHub

10. Frequently Asked Questions 🔝

Subhendu Mohapatra

Author & Instructor at plus2net

I write and maintain practical tutorials on Python, PHP, SQL, JavaScript, HTML, jQuery, and web development at plus2net. The tutorials focus on clear explanations, working examples, and code that readers can test and adapt while learning.

About the author   Watch on YouTube

Connect: