Apprendre Implementing Matrix Transformation in Python

Glissez pour afficher le menu

Solving a Linear System

We define a system of equations:

2x + y = 5 \\ x - y = 1

This is rewritten as:


              123456789
            
import numpy as np

A = np.array([[2, 1],   # Coefficient matrix
              [1, -1]])

b = np.array([5, 1])    # Constants (RHS)

x_solution = np.linalg.solve(A, b)
print(x_solution)

This finds the values of $x$ and $y$ that satisfy both equations.

Why this matters: solving systems of equations is foundational in data science - from fitting linear models to solving optimization constraints.

Applying Linear Transformations

We define a vector:

v = np.array([[2], [3]])

Then apply two transformations:

Scaling

We stretch $x$ by 2 and compress $y$ by 0.5:

S = np.array([[2, 0],
              [0, 0.5]])

scaled_v = S @ v

This performs:

S \cdot v = \begin{bmatrix} 2 & 0 \\ 0 & 0.5 \end{bmatrix} \begin{bmatrix} 2 \\ 3 \end{bmatrix} = \begin{bmatrix} 4 \\ 1.5 \end{bmatrix}

Rotation

We rotate the vector $90°$ counterclockwise using the rotation matrix:

theta = np.pi / 2
R = np.array([[np.cos(theta), -np.sin(theta)],
              [np.sin(theta),  np.cos(theta)]])

rotated_v = R @ v

This yields:

R \cdot v = \begin{bmatrix} 0 & -1 \\ 1 & 0 \end{bmatrix} \begin{bmatrix} 2 \\ 3 \end{bmatrix} = \begin{bmatrix} -3 \\ 2 \end{bmatrix}

Visualizing the Transformations

Using matplotlib, we plot each vector from the origin, with their coordinates labeled:


              1234567891011121314151617181920212223242526272829303132333435363738394041424344454647
            
import numpy as np
import matplotlib.pyplot as plt

# Define original vector
v = np.array([[2], [3]])

# Scaling
S = np.array([[2, 0],
              [0, 0.5]])
scaled_v = S @ v

# Rotation
theta = np.pi / 2
R = np.array([[np.cos(theta), -np.sin(theta)],
              [np.sin(theta),  np.cos(theta)]])
rotated_v = R @ v

# Vizualization
origin = np.zeros(2)
fig, ax = plt.subplots(figsize=(6, 6))

# Plot original
ax.quiver(*origin, v[0], v[1], angles='xy', scale_units='xy', scale=1, color='blue', label='Original')

# Plot scaled
ax.quiver(*origin, scaled_v[0], scaled_v[1], angles='xy', scale_units='xy', scale=1, color='green', label='Scaled')

# Plot rotated
ax.quiver(*origin, rotated_v[0], rotated_v[1], angles='xy', scale_units='xy', scale=1, color='red', label='Rotated')

# Label vector tips
ax.text(v[0][0]+0.2, v[1][0]+0.2, "(2, 3)", color='blue')
ax.text(scaled_v[0][0]+0.2, scaled_v[1][0]+0.2, "(4, 1.5)", color='green')
ax.text(rotated_v[0][0]-0.6, rotated_v[1][0]+0.2, "(-3, 2)", color='red')

# Draw coordinate axes
ax.annotate("", xy=(5, 0), xytext=(0, 0), arrowprops=dict(arrowstyle="->", linewidth=1.5))
ax.annotate("", xy=(0, 5), xytext=(0, 0), arrowprops=dict(arrowstyle="->", linewidth=1.5))
ax.text(5.2, 0, "X", fontsize=12)
ax.text(0, 5.2, "Y", fontsize=12)

ax.set_xlim(-5, 5)
ax.set_ylim(-5, 5)
ax.set_aspect('equal')
ax.grid(True)
ax.legend()
plt.show()

Why this matters: data science workflows often include transformations, for example:

Principal Component Analysis - rotates data;
Feature normalization - scales axes;
Dimensionality reduction - projections.

By visualizing vectors and their transformations, we see how matrices literally move and reshape data in space.

Tout était clair ?

Merci pour vos commentaires !

Section 1. Chapitre 33

Demandez à l'IA

Posez n'importe quelle question ou essayez l'une des questions suggérées pour commencer notre discussion

Section 1. Chapitre 33