Gradient Descent from Scratch

This Python project demonstrates gradient descent on a univariate function using NumPy and Matplotlib. Two different starting points are used to show how the optimization paths converge to local or global minima.

It visualizes gradient descent on the function:

$$ f(x) = x^4 - 10x^2 + 9 $$

Two points with different initial values are optimized using gradient descent, and their paths are animated using Matplotlib.

Features

• Gradient descent visualization from two starting points
• Live visualization over 500 iterations
• Final positions highlighted on the graph

How to run ?

python gradient_descent_visualization_2d.py

Customizations

• Starting points: Change current_x0 and current_x1
• Learning rate: Modify learning_rate
• Iterations: Adjust the loop count
• More points: You can add as many points as you wish