Partial Differentiation- A Comprehensive Tutorial

What Is Partial Differentiation?

Partial differentiation is a calculus technique used when you have functions with more than one variable. Instead of finding the derivative of the entire function, you find how the function changes when you vary one variable while keeping the others constant.

If you're studying multivariable calculus, engineering, physics, or economics, this is a foundational skill you need to master. It's not optional—it's the gateway to understanding how complex systems behave.

The Basic Idea

Imagine a function that depends on two variables:

f(x, y) = x² + 3xy + y³

A regular derivative assumes only one variable exists. Partial derivatives say "hold everything else still" and differentiate with respect to one variable at a time.

You treat y as a constant when differentiating with respect to x. You treat x as a constant when differentiating with respect to y.

Notation You'll Encounter

Different textbooks and contexts use different notation. Here's what you'll see:

• ∂f/∂x — the partial derivative of f with respect to x
• ∂f/∂y — the partial derivative of f with respect to y
• fₓ — shorthand for ∂f/∂x
• fᵧ — shorthand for ∂f/∂y
• z = f(x, y) — common when dealing with surfaces

The symbol ∂ is called "del" or "partial". It's not a typo—it's the standard notation for partial derivatives.

How To Find Partial Derivatives

Step 1: Identify Your Variables

Determine which variable you're differentiating with respect to. Everything else is treated as a constant.

Step 2: Apply Standard Differentiation Rules

You use the same rules as regular calculus:

• Power rule: d/dx(xⁿ) = nxⁿ⁻¹
• Product rule when needed
• Chain rule for composite functions

Step 3: Differentiate Term by Term

Take each term containing your variable and differentiate it. Drop terms that don't contain your variable (they become zero when treated as constants).

Worked Examples

Example 1: Simple Polynomial

f(x, y) = 4x³ + 2xy² + y

Finding ∂f/∂x:

• 4x³ → 12x²
• 2xy² → 2y² (x differentiates to 1, y² stays)
• y → 0 (no x, so it's a constant)

Result: ∂f/∂x = 12x² + 2y²

Finding ∂f/∂y:

• 4x³ → 0 (no y, constant)
• 2xy² → 4xy (y² differentiates to 2y)
• y → 1

Result: ∂f/∂y = 4xy + 1

Example 2: With Exponentials and Trig

f(x, y) = eˣʸ + sin(x) + ln(y)

∂f/∂x:

• eˣʸ → use chain rule → eˣʸ · y
• sin(x) → cos(x)
• ln(y) → 0

Result: ∂f/∂x = yeˣʸ + cos(x)

∂f/∂y:

• eˣʸ → eˣʸ · x
• sin(x) → 0
• ln(y) → 1/y

Result: ∂f/∂y = xeˣʸ + 1/y

Higher-Order Partial Derivatives

You can take partial derivatives multiple times. The notation gets more complex:

• fₓₓ = ∂²f/∂x² — second derivative with respect to x twice
• fᵧᵧ = ∂²f/∂y² — second derivative with respect to y twice
• fₓᵧ = ∂²f/∂y∂x — first x, then y
• fᵧₓ = ∂²f/∂x∂y — first y, then x

Clairaut's Theorem (Equality of Mixed Partials):

If the function is "nice enough" (continuous and smooth), then fₓᵧ = fᵧₓ. The order of mixed partials doesn't matter. Most functions you'll encounter satisfy this.

The Gradient Vector

The gradient combines all partial derivatives into one vector:

∇f = (∂f/∂x, ∂f/∂y, ...)

The gradient points in the direction of steepest ascent. It's fundamental to optimization and machine learning algorithms.

For f(x, y) = x² + y²:

∇f = (2x, 2y)

At the point (3, 4), the gradient is (6, 8).

Applications of Partial Differentiation

Physics

Partial derivatives appear constantly in thermodynamics, electromagnetism, and quantum mechanics. The Schrödinger equation? It uses partial derivatives. Heat equation? Same deal.

Economics

Utility functions, production functions, and demand functions often depend on multiple variables. Partial derivatives tell you how demand changes when price changes, holding income constant.

Machine Learning

Gradient descent algorithms use partial derivatives to minimize cost functions. Every neural network training run depends on understanding how changing each weight affects the final error.

Engineering

Stress analysis, fluid dynamics, and control systems all rely on partial differential equations—equations involving partial derivatives.

Common Mistakes to Avoid

• Treating all variables the same — Remember: you're differentiating with respect to ONE variable only
• Forgetting the chain rule — When variables appear in exponents, arguments of functions, or products, the chain rule applies
• Incorrect product rule application — If you have f(x)·g(x,y), the product rule still applies to the x terms
• Confusing notation — ∂f/∂x is NOT the same as df/dx. The partial symbol matters.

Partial vs. Total Derivative

This trips people up constantly. Here's the difference:

Example: z = f(x, y) where y = g(x)

The total derivative is:

dz/dx = ∂f/∂x + ∂f/∂y · dy/dx

The partial derivative ignores the relationship between x and y. The total derivative accounts for it.

Practice Problems to Try

Work through these to build fluency:

1. f(x, y) = x³y² + 5xy + 2 → Find fₓ and fᵧ
2. f(x, y) = eˣ�ʸ · cos(y) → Find fₓ
3. f(x, y, z) = xyz + x² + y² + z² → Find all first-order partials
4. Verify Clairaut's theorem for f(x, y) = x³y² + 2xy

Check your answers. The only way to get comfortable with partial derivatives is to practice until the process becomes automatic.

When You'll Use This Later

Partial derivatives lead directly to:

• Directional derivatives
• Tangent planes to surfaces
• Lagrange multipliers for constrained optimization
• Partial differential equations (PDEs)

If you don't nail partial derivatives now, you'll struggle with everything that follows. There's no way around it.

The Bottom Line

Partial differentiation is straightforward once you understand the core concept: differentiate with respect to one variable while treating everything else as constant. The notation looks intimidating, but the mechanics are just regular differentiation with extra steps.

Master the basics, practice the algebra, and don't skip the multivariable chain rule. Those three things will carry you through everything that comes next.