Polynomial Global Minimum Finder (Medium) — Practice with Code Visualizer

In optimization theory and machine learning, finding the global minimum of a function is one of the most fundamental challenges. Unlike local minima—which represent the lowest points within a neighborhood—the global minimum is the absolute lowest value a function can achieve across its entire domain.

Consider the following quartic polynomial function:

$$f(x) = x^4 - 3x^3 + 2$$

This function exhibits the characteristic behavior of higher-order polynomials: it can possess multiple critical points (where the derivative equals zero), including local minima, local maxima, and inflection points. The challenge lies in distinguishing between these critical points and identifying the true global minimum.

The Optimization Challenge:

Starting from an arbitrary initial position, you must navigate the function's landscape to find the x-coordinate where the global minimum occurs. This is analogous to:

A hiker seeking the lowest valley in a mountain range while starting from a random location
An algorithm finding the optimal parameters in a loss function
A system converging to its most stable equilibrium state

Mathematical Foundation:

To find critical points, we compute the first derivative and set it to zero: $$f'(x) = 4x^3 - 9x^2 = x^2(4x - 9) = 0$$

This yields critical points at $x = 0$ and $x = 2.25$. To determine their nature, we examine the second derivative: $$f''(x) = 12x^2 - 18x$$

At $x = 0$: $f''(0) = 0$ (inconclusive—requires further analysis) At $x = 2.25$: $f''(2.25) > 0$ (confirms local minimum)

Your Task:

Implement a Python function that finds the x-coordinate of the global minimum of $f(x) = x^4 - 3x^3 + 2$, starting from any given initial position. Your algorithm should reliably converge to the correct answer regardless of the starting point.

Note: Be cautious of local minima or saddle points that might trap naive optimization algorithms. Consider using techniques like gradient descent with appropriate step sizes, momentum, or analytical methods to ensure global convergence.

Starting from x = 0.0, the optimization algorithm encounters a saddle point where the gradient is zero (f'(0) = 0) but the second derivative is also zero. A robust optimization technique (like momentum or analytical checking) escapes this saddle point and successfully navigates to the true global minimum at x = 2.25, where f(2.25) ≈ -3.54.

Beginning at x = 1.0, which lies between the two critical points (x = 0 and x = 2.25), the gradient descent process computes f'(1) = 4(1)³ - 9(1)² = 4 - 9 = -5. The negative derivative indicates the function is decreasing, pushing the optimization toward larger x values. Following the descent direction, the algorithm converges to x = 2.25, the global minimum where f(2.25) = (2.25)⁴ - 3(2.25)³ + 2 ≈ -3.54.

Starting from x = -1.0 (in the negative domain), the steep negative gradient pushes the optimization toward increasing x values. The algorithm traverses past the saddle point at x = 0 and continues until reaching the global minimum at x = 2.25.