Basic Derivative Rules

Single-variable calculus from first principles

Computing a derivative from the limit definition every time would be exhausting. A handful of rules let you differentiate almost anything by inspection. Learn these and you'll never need the limit again for ordinary functions.

The workhorse. To differentiate a power, bring the exponent down in front and drop it by one:

Measuring a car's speed by hand, marking distances and timing them with a stopwatch, works but is painfully slow. A speedometer wired to the wheels gives you the same answer instantly. The derivative rules are that wired-in shortcut: instead of grinding through the limit definition every time, you read the slope straight off the formula.

Where this lives in MLEvery framework's autograd engine has these rules built in. When you call .backward(), it's applying the power rule, the sum rule, and the standard-function derivatives (plus the chain rule, next module) automatically across millions of operations. Knowing them by hand is how you sanity-check a gradient, debug a custom layer, or derive a loss's gradient on paper.
▶ Basic Derivative Rules
← DifferentiabilityProduct & Quotient Rules →