Finding Maximum and Minimum Values in Multivariable Calculus

What You're Actually Looking For

When you work with functions of multiple variables, finding maximum and minimum values gets trickier. You've got more directions to check, more conditions to satisfy. This isn't rocket science, but you will trip up if you skip the fundamentals.

Here's what you need to know: extreme values in multivariable functions occur at critical points, on boundaries, or where the function isn't differentiable. Most textbook problems focus on finding interior extrema using calculus tools.

Critical Points: Where Everything Starts

A critical point is where the function's partial derivatives equal zero or don't exist. Finding these points is your first move.

How to Find Critical Points

Take each partial derivative and set them all equal to zero simultaneously. Solve the system of equations.

For a function f(x, y), you need:

f_x = 0
f_y = 0

Solve both equations together. The solutions are your critical points. This part is algebra—get fast at solving systems of equations.

The Second Derivative Test (Hessian Matrix)

Critical points exist. But you still need to classify them. A maximum? Minimum? Saddle point? The Hessian matrix tells you.

The Discriminant (D Test)

Calculate the determinant of the Hessian matrix:

D = f_xx(a,b) × f_yy(a,b) − [f_xy(a,b)]²

Then evaluate D at each critical point (a, b):

D > 0 and f_xx < 0: local maximum
D > 0 and f_xx > 0: local minimum
D < 0: saddle point
D = 0: test is inconclusive—use another method

That's it. Plug, calculate, classify.

Example Calculation

Let's classify f(x, y) = x² + y² − 2x − 4y + 8.

Partial derivatives:

f_x = 2x − 2
f_y = 2y − 4

Set equal to zero: x = 1, y = 2. Critical point is (1, 2).

Second derivatives: f_xx = 2, f_yy = 2, f_xy = 0

D = (2)(2) − (0)² = 4. Since D > 0 and f_xx > 0, this is a local minimum. The minimum value is f(1, 2) = 1.

Boundary Points: The Part Most Students Skip

Interior critical points aren't the whole story. Functions can achieve extreme values on boundaries of their domain.

You have three options for boundary analysis:

Parametrize the boundary and reduce to a single-variable problem
Use the method of Lagrange multipliers (constrained optimization)
Compare function values at all boundary candidates

For closed, bounded regions, absolute maxima and minima exist at either interior critical points or boundary points. Check both.

Lagrange Multipliers: Constrained Optimization

When your variables have constraints, Lagrange multipliers are the tool. You're optimizing f(x, y, z) subject to g(x, y, z) = 0.

The Setup

Set up the system:

∇f = λ∇g
g(x, y, z) = 0

Solve for x, y, z, and λ. The λ (lambda) is the Lagrange multiplier—don't ignore it, but don't overthink it either. It's just a placeholder that lets you solve the system.

Quick Example

Maximize f(x, y) = xy subject to x + 2y = 40.

∇f = (y, x), ∇g = (1, 2)

y = λ(1) → y = λ

x = λ(2) → x = 2λ

From constraint: 2λ + 2(λ) = 40 → λ = 10

So x = 20, y = 10. Maximum is f(20, 10) = 200.

Quick Reference: Classification Cheat Sheet

Condition	Classification
D > 0, f_xx > 0	Local minimum
D > 0, f_xx < 0	Local maximum
D < 0	Saddle point
D = 0	Inconclusive

Common Mistakes That Cost You Points

Forgetting boundary analysis—interior critical points don't guarantee absolute extrema
Solving only one partial derivative—you need all of them set to zero
Confusing local and absolute extrema—a local max isn't necessarily the global max
Miscomputing mixed partials—f_xy = f_yx if the function is well-behaved, but check your algebra
Dropping negative signs—this is elementary and will destroy your answer

Practical Workflow: Finding Extrema Step-by-Step

When you face a multivariable optimization problem, follow this order:

Find all partial derivatives
Set all partials equal to zero—solve the system
Calculate second derivatives and the Hessian determinant
Classify each critical point using the discriminant test
Evaluate the function at all critical points
Analyze boundary behavior if the domain has boundaries
Compare all candidates—the largest value is the absolute maximum, the smallest is the absolute minimum

When the Test Fails (D = 0)

The second derivative test gives no information when D = 0. You have two choices:

Use higher-order derivatives (tedious)
Analyze the function directly near the point (often easier)

Sometimes you can complete the square or rewrite the function to see what's happening. Don't get stuck on the Hessian when a different approach is faster.

Most problems you'll encounter either have D ≠ 0 or are constructed so the answer is obvious. If you get D = 0 on an exam, the problem is probably testing whether you recognize the limitation.