Multivariable Calculus Study Guide

A Structured Reference Guide on Gradients, Directional Derivatives, and Optimization

Introduction
This comprehensive study guide is compiled directly from raw lecture notes and tailored for students seeking a
rigorous yet accessible understanding of multivariable functions. It bridges theoretical multivariable formulas with
practical, fully explained exam-style applications.

The guide is split into two primary pedagogical modules: Module 1 breaks down spatial rates of change, partial
derivatives, and vectors in 3D space. Module 2 details optimization strategies for surfaces, focusing on critical
points and the second derivative discriminant test.

Module 1: The Gradient & Directional Derivatives

1.1 Theoretical Intuition & Core Formulas

In single-variable calculus, the derivative f'(x) represents the slope of a line in exactly one dimensional direction
(along the x-axis). In multivariable systems, a function f(x, y, z) defines a surface or a spatial field where you can
move in an infinite number of spatial directions. To compute rates of change under these conditions, we use two
interconnected tools:

The Gradient Vector (∇f): The gradient collects all the direct orthogonal rates of change (partial
derivatives) of a function into a single vector. Geometrically, the gradient vector points in the direction of the
steepest ascent on a surface, and its magnitude equals that maximum rate of increase.

∂f ∂f ∂f
∇f(x, y, z) = fxi + fyj + fzk = ⟨ , , ⟩
∂x ∂y ∂z

The Unit Direction Vector (u): To measure a rate of change cleanly, our measuring tape must have a
standard length of exactly 1. If a problem gives you a generic directional vector v, you must normalize it by
dividing by its geometric length (magnitude):

v v1i + v2j + v3k

u= =
|v| √(v12 + v22 + v32)

Multivariable Calculus: Student Study Guide Page 1 of 10

 The Directional Derivative (Duf): The directional derivative calculates the scalar rate of change of the
function at a specific point along the specific trajectory path of the unit vector. It is computed via the dot
product:

Duf = ∇f · u = (fx · u1) + (fy · u2) + (fz · u3)

Student Warning - Common Pitfall: Always check if the directional vector provided is a unit vector. If you
perform a dot product with a non-unit vector v (where |v| ≠ 1), your answer will be scaled incorrectly by a
factor of |v|.

Multivariable Calculus: Student Study Guide Page 2 of 10

 1.2 Step-by-Step Practical Applications

Multivariable Calculus: Student Study Guide Page 3 of 10

 Example 1 (Trigonometric Functions):
Given the function f(x, y, z) = x sin(yz), determine the gradient vector ∇f and compute the exact directional
derivative Duf at the coordinate point (1, 3, 0) in the specific direction of vector v = i + 2j - k.

Step 1: Compute the generic partial derivatives.

Treat variables not being differentiated as constants, applying the chain rule where appropriate:

• ∂
With respect to x: fx = [x sin(yz)] = sin(yz)
∂x

• ∂
With respect to y: fy = [x sin(yz)] = x · cos(yz) · z = xz cos(yz)
∂y

• ∂
With respect to z: fz = [x sin(yz)] = x · cos(yz) · y = xy cos(yz)
∂z

This provides the general spatial gradient field:

∇f(x, y, z) = sin(yz)i + xz cos(yz)j + xy cos(yz)k

Step 2: Evaluate the gradient at the point (1, 3, 0).

Substitute x = 1, y = 3, and z = 0 into each expression:

• f (1, 3, 0) = sin(3 · 0) = sin(0) = 0

x
• f (1, 3, 0) = 1 · 0 · cos(3 · 0) = 0
y
• f (1, 3, 0) = 1 · 3 · cos(3 · 0) = 3 · cos(0) = 3 · 1 = 3
z

So, the local gradient vector is: ∇f(1, 3, 0) = 0i + 0j + 3k = ⟨0, 0, 3⟩.

Step 3: Normalize the direction vector into a unit vector.

Calculate the magnitude of the given vector v = \langle1, 2, -1 angle:

|v| = √(12 + 22 + (-1)2) = √(1 + 4 + 1) = √6

Divide the original components by this magnitude to construct the unit vector u:

1 2 1 1 2 1
u= i+ j- k=⟨ , ,- ⟩
√6 √6 √6 √6 √6 √6

Step 4: Execute the dot product.

1 2 1
Duf = ∇f · u = (0 · ) + (0 · ) + (3 · (- ))
√6 √6 √6

3 √6
Final Answer: Duf = - (or simplified as - ≈ -1.225)
√6 2

Multivariable Calculus: Student Study Guide Page 4 of 10

 Example 2 (Power Functions & Multi-Variable Chain Rule):
Given the square root radical function f(x, y, z) = √(xyz) = (xyz)1/2, evaluate the directional derivative Duf
at the point (3, 2, 6) in the direction of vector v = -i - 2j + k.

Step 1: Determine the partial derivatives using the general power chain rule.
d 1
Recall that [X1/2] = . Multiplying by the inner derivative gives:
dX 2√X

• 1 yz
fx = · ∂x(xyz) =
2(xyz)1/2 2√(xyz)

• 1 xz
fy = · ∂y(xyz) =
2(xyz)1/2 2√(xyz)

• 1 xy
fz = · ∂z(xyz) =
2(xyz)1/2 2√(xyz)

Step 2: Substitute values at the coordinate point (3, 2, 6).

First calculate the value inside the radical: xyz = 3 · 2 · 6 = 36. Therefore, √(xyz) = √36 = 6.

• 2·6 12
fx(3, 2, 6) = = =1
2·6 12

• 3·6 18 3
fy(3, 2, 6) = = =
2·6 12 2

• 3·2 6 1
fz(3, 2, 6) = = =
2·6 12 2

3 1 3 1
This constructs our localized gradient vector: ∇f(3, 2, 6) = 1i + j+ k = ⟨1, , ⟩.
2 2 2 2

Step 3: Normalize the direction vector.

Find the length of vector v = ⟨-1, -2, 1⟩:

|v| = √((-1)2 + (-2)2 + 12) = √(1 + 4 + 1) = √6

1 2 1
The normalized unit vector is: u = ⟨ - ,- , ⟩
√6 √6 √6

Step 4: Compute the dot product for the final rate of change.
1 3 2 1 1
Duf = (1 · (- )) + ( · (- )) + ( · )
√6 2 √6 2 √6

1 3 1 -2 - 6 + 1
Duf = - - + =
√6 √6 2√6 2√6

Multivariable Calculus: Student Study Guide Page 5 of 10

 7
Final Answer: Duf = - ≈ -1.429
2√6

Multivariable Calculus: Student Study Guide Page 6 of 10

Module 2: Maxima, Minima, & Critical Points

2.1 The Second Derivative Test Workflow

To optimize a surface defined by a function f(x, y), we must seek spatial locations where the tangent plane becomes
completely horizontal. This process mimics the single-variable strategy but utilizes higher-dimensional curvature
tests.

Phase 1: Isolate Critical Points

Find coordinates where both partial first derivatives simultaneously vanish to zero. This signifies a flat slope
profile:

fx(x, y) = 0 \quad ext{and} \quad fy(x, y) = 0

Solving this algebraic system reveals one or more critical coordinate pairs (x0, y0).

Phase 2: Calculate the Hessian Discriminant (D)

The discriminant evaluates a localized product of direct curvatures subtracted by the squared cross-curvature
to check for stability:

D = fxx(x0, y0) · fyy(x0, y0) - [fxy(x0, y0)]2

2.2 Critical Point Classification Matrix

Once you calculate the numeric values for D and fxx, apply this definitive rubric:

Multivariable Calculus: Student Study Guide Page 7 of 10

 Discriminant Direct Curvature Geometric
Physical Interpretation
Value (fxx) Classification

Bowl-shaped valley structure opening

D>0 fxx > 0 Local Minimum
upwards.

Dome-shaped mountain peak structure

D>0 fxx < 0 Local Maximum
opening downwards.

Pass structure (goes up in one direction,

D<0 Any Value Saddle Point
down in another).

Requires advanced higher-order

D=0 Any Value Inconclusive / Fails
polynomial tests.

Multivariable Calculus: Student Study Guide Page 8 of 10

 2.3 Step-by-Step Practical Applications

Example 3 (Locating a Saddle Point):

Perform a thorough optimization analysis on the following algebraic surface:

f(x, y) = 2x2 - 2xy + 2y

Step 1: Compute first-order partials and set to zero.

• ∂
fx = [2x2 - 2xy + 2y] = 4x - 2y = 0 \Rightarrow 2y = 4x \Rightarrow y = 2x
∂x

• ∂
fy = [2x2 - 2xy + 2y] = -2x + 2 = 0 \Rightarrow 2x = 2 \Rightarrow x = 1
∂y

Substitute x = 1 into the first equation: y = 2(1) = 2.

We have found a single critical point located at (1, 2).

Step 2: Evaluate second-order partial derivatives.

• ∂
fxx = (4x - 2y) = 4
∂x

• ∂
fyy = (-2x + 2) = 0
∂y

• ∂
fxy = (4x - 2y) = -2
∂y

Step 3: Apply the discriminant equation.

D = fxx · fyy - (fxy)2 = (4 · 0) - (-2)2 = 0 - 4 = -4

Conclusion: Since D = -4 < 0, the point (1, 2) is a Saddle Point. This surface contains no local maxima and
no local minima.

Multivariable Calculus: Student Study Guide Page 9 of 10

 Example 4 (Locating a Local Minimum):
Perform a thorough optimization analysis on the following algebraic surface:

f(x, y) = x2 + xy + y2 - 3x

Step 1: Compute first-order partials and solve the linear system.

• f = 2x + y - 3 = 0 \quad ext{(Eq. 1)}
x
• f = x + 2y = 0 \Rightarrow x = -2y \quad ext{(Eq. 2)}
y

Substitute Eq. 2 into Eq. 1:

2(-2y) + y - 3 = 0 \Rightarrow -4y + y = 3 \Rightarrow -3y = 3 \Rightarrow y = -1

Find x via Eq. 2: x = -2(-1) = 2.

We have found a single critical point located at (2, -1).

Step 2: Evaluate second-order partial derivatives.

• ∂
fxx = (2x + y - 3) = 2
∂x

• ∂
fyy = (x + 2y) = 2
∂y

• ∂
fxy = (x + 2y) = 1
∂x

Step 3: Apply the discriminant equation.

D = fxx · fyy - (fxy)2 = (2 · 2) - (1)2 = 4 - 1 = 3

Conclusion: Since D = 3 > 0, the point is a local extremum. Because fxx = 2 > 0, the surface has upward
concave curvature. Thus, the point (2, -1) is classified as a Local Minimum.

Multivariable Calculus: Student Study Guide Page 10 of 10