Classifying Critical Points: Local Extrema and Saddle Points
What You'll Learn
You'll learn to identify the three types of critical points in multivariable functions: local maxima, local minima, and saddle points.
The Three Types of Critical Points
Once you've found critical points (where the gradient is zero or undefined), the next step is figuring out what kind they are.
Local Maximum
At a local maximum, the function reaches a peak—the output is higher than all nearby points. Imagine standing at the top of a hill; every direction you walk leads downward.
Local Minimum
At a local minimum, the function reaches a valley—the output is lower than all nearby points. Think of standing at the bottom of a bowl; every direction you walk leads upward.
Saddle Point
At a saddle point, the function curves up in some directions and down in others—like sitting on a horse's saddle. If you walk north-south, you might be at a minimum, but if you walk east-west, you might be at a maximum. It's neither a true peak nor a true valley.
Why This Matters
Not all critical points are created equal! Just because the gradient vanishes doesn't tell you whether you've found a maximum (great for profit functions), a minimum (great for cost functions), or a saddle point (often not optimal for anything).
In the next lessons, you'll learn the Second Derivative Test (using the Hessian matrix) to systematically classify these critical points by examining how the function curves in all directions simultaneously.
Analogy
Think of a landscape: mountains have peaks (local max), valleys have bottoms (local min), and mountain passes have saddles (neither max nor min, just a transition point).
Key Takeaway: Critical points come in three flavors—local maxima (peaks), local minima (valleys), and saddle points (passes)—and identifying which type you have is essential for optimization problems.