Partial Derivatives

Subtopic: Multivariable

Partial derivatives measure how a function of several variables changes when you vary one variable while holding others fixed. They're the building blocks of multivariable calculus—combining to form gradients, Jacobians, and Hessians. Understanding partial derivatives is essential for optimization, physics, and machine learning.

Introduction

A function like ƒ⁡(x,y)=x2+x⁢y f with hook of open paren x comma y close paren equals x squared plus x y has two inputs. How does ƒ f with hook change when x x changes? When y y changes? These are different questions with different answers. Partial derivatives answer each question separately.

It's like asking how temperature varies with latitude vs longitude—each direction gives a different rate of change.

Definition

The partial derivative of ƒ⁡(x,y) f with hook of open paren x comma y close paren with respect to x x is:

∂ƒ∂x=lim⁡ƒ⁡(x+h,y)−ƒ⁡(x,y)h the fraction with numerator partial differential f with hook and denominator partial differential x equals lim of the fraction with numerator f with hook of open paren x plus h comma y close paren minus f with hook of open paren x comma y close paren and denominator h

Treat y as a constant and differentiate with respect to x using ordinary derivative rules.

Similarly for ∂ƒ∂y the fraction with numerator partial differential f with hook and denominator partial differential y: treat x x as constant, differentiate with respect to y y.

Worked Example

Find the partial derivatives of ƒ⁡(x,y)=x3⁢y+2⁢x⁢y2-y f with hook of open paren x comma y close paren equals x cubed y plus 2 x y squared minus y.

∂f/∂x (treat y as constant)

∂ƒ∂x=3⁢x2⁢y+2⁢y2 the fraction with numerator partial differential f with hook and denominator partial differential x equals 3 x squared y plus 2 y squared

The y y terms act like constants: (x3⁢y)′=3⁢y⁢x2 open paren x cubed y close paren prime equals 3 y x squared, (2⁢x⁢y2)′=2⁢y2 open paren 2 x y squared close paren prime equals 2 y squared, (-y)′=0 open paren negative y close paren prime equals 0.

∂f/∂y (treat x as constant)

∂ƒ∂y=x3+4⁢x⁢y−1 the fraction with numerator partial differential f with hook and denominator partial differential y equals x cubed plus 4 x y minus 1

Now x x is constant: (x3⁢y)′=x3 open paren x cubed y close paren prime equals x cubed, (2⁢x⁢y2)′=4⁢x⁢y open paren 2 x y squared close paren prime equals 4 x y, (-y)′=-1. open paren negative y close paren prime equals negative 1 period

Geometric Interpretation

For z=ƒ⁡(x,y) z equals f with hook of open paren x comma y close paren, the graph is a surface in 3 3D .

∂ƒ∂x the fraction with numerator partial differential f with hook and denominator partial differential x at (a,b) open paren a comma b close paren is the slope of the slice of the surface where y=b y equals b (walking in the x x-direction).

∂ƒ∂y the fraction with numerator partial differential f with hook and denominator partial differential y at (a,b) open paren a comma b close paren is the slope of the slice where x=a x equals a (walking in the y y-direction).

Each partial derivative measures steepness along one coordinate axis.

Higher-Order Partials

You can take partial derivatives of partial derivatives:

ƒxx=∂2ƒ∂x2 f with hook sub xx equals the fraction with numerator partial differential squared f with hook and denominator partial differential x squared, ƒxy=∂2ƒ∂(y)⁢∂x f with hook sub xy equals the fraction with numerator partial differential squared f with hook and denominator partial differential y partial differential x, ƒyy=∂2ƒ∂y2 f with hook sub yy equals the fraction with numerator partial differential squared f with hook and denominator partial differential y squared

Mixed partials: ƒxy f with hook sub xy means differentiate with respect to x x first, then y y.

Clairaut's Theorem

If mixed partials are continuous, they're equal:

ƒxy=ƒyx f with hook sub xy equals f with hook sub yx

Order doesn't matter (for "nice" functions).

The Gradient

The gradient collects all partial derivatives into a vector:

∇⁢ƒ=(∂ƒ∂x,∂ƒ∂y) nabla f with hook equals open paren the fraction with numerator partial differential f with hook and denominator partial differential x comma the fraction with numerator partial differential f with hook and denominator partial differential y close paren

It points in the direction of steepest increase; its magnitude is the steepest slope.

Chain Rule for Partials

If z=ƒ⁡(x,y) z equals f with hook of open paren x comma y close paren and x,y x comma y depend on t t:

dzdt=∂ƒ∂x⁢dxdt+ d z over d t equals the fraction with numerator partial differential f with hook and denominator partial differential x d x over d t plus ∂ƒ∂y⁢dydt the fraction with numerator partial differential f with hook and denominator partial differential y d y over d t

Changes propagate through all paths from t t to z z.

Applications

In physics, partial derivatives describe how physical quantities depend on multiple variables (pressure, temperature, volume).

In economics, marginal analysis uses partials to measure sensitivity to price, quantity, etc.

In machine learning, gradients (vectors of partials) drive optimization of loss functions.

Summary

Partial derivatives measure the rate of change with respect to one variable, holding others constant. Compute by treating other variables as constants. ∂ƒ∂x the fraction with numerator partial differential f with hook and denominator partial differential x is the slope along the x x-direction. The gradient ∇ƒ nabla f with hook collects all partials. Mixed partials are equal (ƒxy=ƒyx) open paren f with hook sub xy equals f with hook sub yx close paren when continuous. Chain rules sum contributions from all paths.