MSK Institute • Shikohabad
Loading Page Content...
Preparing courses, live labs, and learning resources.सामग्री लोड हो रही है, कृपया प्रतीक्षा करें...
Connecting serverPlease wait...
💻 Practical Labs🚀 Free Demo📜 Verified Certificates
Multi-Dimensional Array Slicing (Row, Column & Matrix Slices)0%
Multi-Dimensional Array Slicing (Row, Column & Matrix Slices)
Beginner12 min read•Updated: Jul 11, 2026
Multi-Dimensional Array Slicing (Row, Column & Matrix Slices)
In multi-dimensional arrays, slicing generalizes across multiple axes. Unlike standard Python (which requires clumsy chained brackets like matrix[row][col]), NumPy allows clean comma-separated slicing along every axis simultaneously: arr[row_slice, col_slice].
1. The 2D Coordinate Syntax: arr[rows, cols]
In a 2D matrix:
- Axis 0 represents Rows (vertical dimension).
- Axis 1 represents Columns (horizontal dimension).
import numpy as np
# Create a 4x5 matrix (4 rows, 5 columns)
grid = np.array([
[10, 11, 12, 13, 14],
[20, 21, 22, 23, 24],
[30, 31, 32, 33, 34],
[40, 41, 42, 43, 44]
])
# Extract a single element at Row 2, Column 3
print(grid[2, 3]) # 33
Click to Zoom2. Extracting Complete Rows and Columns
By using a lone colon : for an axis, you select all elements along that dimension:
# Extract entire Row 1 (all columns)
row_1 = grid[1, :]
print(row_1) # [20 21 22 23 24]
# Extract entire Column 2 (all rows)
col_2 = grid[:, 2]
print(col_2) # [12 22 32 42]
3. Sub-Matrix Slicing (2D Blocks)
You can slice rows and columns simultaneously to extract sub-grids:
# Extract Rows 1 to 3 (exclusive) and Columns 1 to 4 (exclusive)
sub_block = grid[1:3, 1:4]
print(sub_block)
# [[21 22 23]
# [31 32 33]]
4. Preserving Dimensions vs Reducing Dimensions
Notice the subtle difference between slicing with a single integer versus a slice range:
# Case A: Dimensionality Reduction (Returns 1D array)
r1 = grid[0, :]
print(r1.shape) # (5,) -> ndim reduced from 2 to 1!
# Case B: Dimensionality Preserved (Returns 2D array with 1 row)
r2 = grid[0:1, :]
print(r2.shape) # (1, 5) -> ndim remains 2!
In machine learning algorithms and neural networks, keeping rank-2 shapes (1, 5) is often mandatory for matrix multiplication!
5. The Ellipsis Operator (...)
When slicing arrays with 4, 5, or more dimensions (such as video frames: [batch, frames, height, width, channels]), writing colons for every axis becomes tedious. The Ellipsis (...) automatically expands to as many full-slice colons : as needed:
video = np.zeros((32, 100, 1080, 1920, 3))
# Select only the Red channel (channel 0) across all batches, frames, height, and width:
red_channel = video[..., 0] # Equivalent to video[:, :, :, :, 0]
print(red_channel.shape) # (32, 100, 1080, 1920)
Interactive Quiz
Multiple Choice Questions
Question 1
In a 2D NumPy array 'arr', how do you extract the entire third column across all rows?
Question 2
How does 'arr[0, :]' differ from 'arr[0:1, :]'?
Question 3
What does the Ellipsis operator '...' do in multidimensional slicing: 'tensor[..., 0]'?
Question 4
If matrix M has shape (6, 8), what will be the shape of 'M[2:5, 3:7]'?
Question 5
Why is 'matrix[1, 2]' preferred over 'matrix[1][2]' in NumPy?
Next Lesson
Memory Views vs Deep Copies (view() vs copy())
Continue learning with hands-on practice, examples, and exercises in the upcoming topic.
Related Lessons
| Previous Lesson | Next Lesson |
|---|---|
| 1D Array Indexing, Slicing & Step Strides | Memory Views vs Deep Copies (view() vs copy()) |
Practice Quiz
Test your understanding of this lesson with 5 questions. Each question has one correct answer.