Pure · Differentiation
1 / 11
The chain rule
Differentiating a function of a function — y = f(g(x)) — by multiplying the outside derivative by the inside derivative, with the dy/dx = (dy/du)(du/dx) substitution method and the reciprocal connection dx/dy = 1 / (dy/dx).
Pure · Differentiation
The chain rule
Differentiating a function of a function — y = f(g(x)) — by multiplying the outside derivative by the inside derivative, with the dy/dx = (dy/du)(du/dx) substitution method and the reciprocal connection dx/dy = 1 / (dy/dx).
Why it works
The power rule handles , but real functions are usually nested: a curve like is one function — "raise to the fifth" — wrapped around another — "triple and add one". You could expand the bracket and differentiate term by term, but for that is grim, and for it is impossible. The chain rule differentiates the nesting directly.Give the inside its own name. Write , so the outer function is . Now you have two easy derivatives: The chain rule says multiply them:
Why does multiplying rates work? Think of the derivatives as conversion factors. If changes times as fast as , and changes times as fast as , then changes times as fast as — the rates compound, exactly the way "miles per hour" times "hours per day" gives "miles per day". In Leibniz notation the 's look like they simply cancel, and that picture is a faithful guide even though is not really a fraction.
In practice you rarely write out. The working version is: differentiate the outside (leaving the inside alone), then multiply by the derivative of the inside. For : the outside gives , the inside differentiates to , so . The general pattern for a bracket to a power is worth memorising:
The reciprocal connection. Sometimes is given as a function of and you want without rearranging. Because the rates are reciprocals, If , then , so straight away. This is the same idea as the chain rule with the roles of and swapped, and it is the key to differentiating inverse functions.