Skip to content

Chain Rule on a Curve

Theorem

Let I⊆RI\subseteq\mathbb{R} be an interval and let γ ⁣:I→Rn\boldsymbol{\gamma}\colon I\to\mathbb{R}^n be a differentiable curve. Its velocity at t0t_0 is

γ′(t0)=(γ1′(t0)⋮γn′(t0)).\boldsymbol{\gamma}'(t_0)= \begin{pmatrix} \gamma_1'(t_0)\\ \vdots\\ \gamma_n'(t_0) \end{pmatrix}.

If B⊆RnB\subseteq\mathbb{R}^n is an open set containing γ(I)\boldsymbol{\gamma}(I) and f ⁣:B→Rf\colon B\to\mathbb{R} is differentiable at γ(t0)∈B\boldsymbol{\gamma}(t_0)\in B, then

ddt(f∘γ)(t0)=Df(γ(t0))γ′(t0)=∇f(γ(t0))⊤γ′(t0).\frac{\mathrm{d}}{\mathrm{d}t}(f\circ\boldsymbol{\gamma})(t_0) =\boldsymbol{D}f(\boldsymbol{\gamma}(t_0))\boldsymbol{\gamma}'(t_0) =\boldsymbol{\nabla} f(\boldsymbol{\gamma}(t_0))^{\top}\boldsymbol{\gamma}'(t_0).

Interpretation

A curve describes motion through a multivariable state space. The chain rule says the value changes by combining the system's velocity with the local sensitivity of the function of several variables.

A curve gamma(t) crossing circular level sets of f. At gamma(t0) the velocity vector and the gradient of f point in partly the same direction, so f increases along the curve there.

The chain rule gives ddtf(γ(t))∣t=t0=∇f(γ(t0))⋅γ′(t0)\frac{\mathrm{d}}{\mathrm{d}t}f(\boldsymbol{\gamma}(t))\big|_{t=t_0}=\boldsymbol{\nabla} f(\boldsymbol{\gamma}(t_0))\cdot\boldsymbol{\gamma}'(t_0). In the figure, the velocity has a positive component along the gradient, so ff increases along the curve at t0t_0. The grey circles are level sets of ff.

A path assigns a position to each parameter value. Composing it with ff restricts ff to that motion; differentiating the composition pairs the derivative of ff with the path velocity.


Key Results

  • The derivative of f(γ(t))f(\boldsymbol{\gamma}(t)) is the total derivative of ff applied to the velocity of γ\boldsymbol{\gamma}.
  • If the velocity is a unit vector, the formula is a directional derivative.
  • The curve case is the one-dimensional input version of the multivariable chain rule.

Examples

  • If f(x,y)=x2+y2f(x,y)=x^2+y^2 and γ(t)=(t,2t)\boldsymbol{\gamma}(t)=(t,2t), then (f∘γ)′(t)=10t(f\circ\boldsymbol{\gamma})'(t)=10t.
  • If f(x,y)=xyf(x,y)=xy and γ(t)=(et,t2)\boldsymbol{\gamma}(t)=(\mathrm{e}^t,t^2), then (f∘γ)′(t)=et(t2+2t)(f\circ\boldsymbol{\gamma})'(t)=\mathrm{e}^t(t^2+2t).

Prerequisites

Sources

Aliases: chain rule along a curve, curve chain rule, chain rule for curves, chain rule along a path Keywords: velocity along a curve, derivative along a path Symbols: element of, gamma, nabla gradient, real numbers, subset

Last updated: 22 October 2026 19:04