Putting It Together

Single-variable calculus from first principles

Now apply the sketch protocol to whole families of functions. The goal here isn't pinpoint accuracy; it's reading the qualitative shape: which way the ends go, how many bumps, where it blows up. A few quick checks usually reveal the silhouette.

For a polynomial, the highest-power term decides the ends. An odd degree with positive lead goes down-left, up-right (like x³); an even degree with positive lead goes up on both ends (like x²). The number of turning points is at most one less than the degree.

Sketching a function is like following a recipe from start to finish. You don't taste every grain of salt; you run through the same ordered steps you already learned — check the ends, find the turns, mark the roots — and the dish takes shape. Each step you practiced earlier is one line in the recipe, and reading them in order is what gives you the finished silhouette.

Where this lives in MLRecognising a function's silhouette at a glance is how you reason about activation and loss functions. The bump of 1/(x²+1) is the shape of a smooth attention/weighting kernel; the S-curve of a sigmoid, the ends-up bowl of a quadratic loss, the down-left/up-right of an odd nonlinearity: knowing the shape tells you how the function behaves at the extremes without plugging in numbers.

▶ Putting It Together

← Systematic Sketch Protocol Riemann Integration →