Key Concepts of Partial Derivatives

Why This Matters

Partial derivatives are your gateway to understanding how multivariable functions behave. In Calculus II, you need to analyze functions that depend on more than one variable at a time. This isn't just an extension of single-variable calculus; it's a fundamentally different way of thinking about change. You'll need to master how functions respond when you tweak one variable while holding others fixed, how the gradient captures the direction of steepest increase, and how optimization works when you have multiple inputs.

These concepts connect directly to exam applications: finding tangent planes, optimizing functions with constraints, and understanding how composite functions behave through the multivariable chain rule. Don't just memorize formulas. Know why we treat other variables as constants, when Clairaut's theorem applies, and how the gradient relates to directional derivatives.

---

Foundations: What Partial Derivatives Measure

A partial derivative isolates the effect of one variable on a function's output while treating all other variables as constants. Think of it this way: if you have a surface in 3D, a partial derivative tells you the slope of that surface along one coordinate axis at a time.

Definition of Partial Derivatives

• Measures the rate of change of a function with respect to one variable while all other variables remain fixed. Geometrically, you're slicing a 3D surface with a vertical plane parallel to one axis and finding the slope of the resulting curve.
• Conceptually distinct from ordinary derivatives. You're examining change along one axis only, not total change across all variables simultaneously.

Notation for Partial Derivatives

The symbol $\partial$ distinguishes partial derivatives from ordinary derivatives ($d$). You'll see three common notations used interchangeably:

• Leibniz notation: $\frac{\partial f}{\partial x}$
• Subscript notation: $f_x$ or $\partial_x f$
• Higher-order notation uses repeated symbols: $\frac{\partial^2 f}{\partial x^2}$ (or $f_{xx}$) for second partials, and $\frac{\partial^2 f}{\partial x \partial y}$ (or $f_{xy}$) for mixed partials

Subscript notation is faster for computation. $f_{xx}$ means differentiate twice with respect to $x$, while $f_{xy}$ means differentiate first with respect to $x$, then with respect to $y$.

Calculating Partial Derivatives

Here's the process:

1. Identify your active variable before you start differentiating. This prevents the most common careless errors.
2. Treat all other variables as constants. For example, to find $\frac{\partial}{\partial x}$ of $f(x,y) = x^2y + y^3$, the $y$ in the first term acts like a coefficient and the $y^3$ term is just a constant. The result is $2xy + 0 = 2xy$.
3. Apply standard differentiation rules (product, quotient, chain) exactly as in single-variable calculus. The only difference is what counts as a constant.

---

Higher-Order Behavior: Second Derivatives and Beyond

Higher-order partial derivatives reveal curvature and concavity of multivariable functions. They're essential for optimization and understanding surface geometry.

Higher-Order Partial Derivatives

• Second partial derivatives like $f_{xx}$ and $f_{yy}$ measure how the rate of change itself changes. This is analogous to concavity in single-variable calculus: $f_{xx}$ tells you about curvature along the $x$-direction.
• These are central to the second derivative test for classifying critical points as maxima, minima, or saddle points (covered in the optimization section below).

Mixed Partial Derivatives

• Differentiate with respect to different variables in succession. $f_{xy}$ means: first compute $f_x$, then differentiate that result with respect to $y$.
• A nonzero mixed partial indicates variable interaction. It tells you that changing one variable affects how the function responds to changes in the other.
• Watch the order of notation. In Leibniz form, $\frac{\partial^2 f}{\partial y \partial x}$ means differentiate with respect to $x$ first, then $y$ (read right to left). In subscript form, $f_{xy}$ means differentiate with respect to $x$ first, then $y$ (read left to right).

Clairaut's Theorem (Equality of Mixed Partials)

Clairaut's theorem states that $f_{xy} = f_{yx}$ provided both mixed partials are continuous near the point in question. For virtually every function you'll encounter in this course, this condition holds.

Why does this matter? It simplifies calculations dramatically. If one order of differentiation involves simpler algebra than the other, you can choose the easier path and get the same answer.

---

The Chain Rule and Implicit Differentiation

When functions are composed or defined implicitly, you need specialized techniques. The multivariable chain rule tracks how changes propagate through nested dependencies.

The Chain Rule for Partial Derivatives

If $z = f(x,y)$ where $x = g(t)$ and $y = h(t)$, then:

$\frac{dz}{dt} = \frac{\partial f}{\partial x}\frac{dx}{dt} + \frac{\partial f}{\partial y}\frac{dy}{dt}$

Each term accounts for one pathway through which $t$ affects $z$. Tree diagrams are genuinely helpful here: draw arrows from the final output back through intermediate variables to the independent variables. Each path through the tree contributes one term to the chain rule expression.

This generalizes naturally. If $x$ and $y$ each depend on two variables $s$ and $t$, you get partial derivatives of $z$ with respect to $s$ and $t$, each with two terms summing over the intermediate variables.

Implicit Differentiation with Partial Derivatives

When you can't solve explicitly for one variable, implicit differentiation handles it. Given a surface defined by $F(x,y,z) = 0$, you can find $\frac{\partial z}{\partial x}$ without ever isolating $z$.

The shortcut formula is:

$\frac{\partial z}{\partial x} = -\frac{F_x}{F_z} \quad \text{(when } F_z \neq 0\text{)}$

This comes from differentiating $F(x,y,z) = 0$ with respect to $x$, treating $y$ as constant and $z$ as a function of $x$ and $y$. Every term containing $z$ picks up a factor of $\frac{\partial z}{\partial x}$ via the chain rule. Then you solve for $\frac{\partial z}{\partial x}$.

---

Gradients and Directional Derivatives

The gradient unifies all first-order partial derivatives into a single vector that points toward steepest ascent. Directional derivatives generalize partial derivatives to arbitrary directions.

Gradient Vector

• Defined as $\nabla f = \langle f_x, f_y \rangle$ in 2D (or $\langle f_x, f_y, f_z \rangle$ in 3D). It collects all first partial derivatives into one vector.
• Points in the direction of steepest increase of $f$. This is the single most important geometric fact about gradients.
• Its magnitude $\|\nabla f\|$ gives the maximum rate of change of the function at that point.
• Perpendicular to level curves (in 2D) and level surfaces (in 3D). This is why the gradient appears in tangent plane and normal line calculations.

Directional Derivatives

The directional derivative measures the rate of change of $f$ in any direction, not just along coordinate axes.

$D_{\mathbf{u}}f = \nabla f \cdot \mathbf{u}$

where $\mathbf{u}$ is a unit vector in the desired direction. A common mistake is forgetting to normalize: if you're given a direction vector $\mathbf{v}$, divide by its magnitude to get $\mathbf{u} = \frac{\mathbf{v}}{\|\mathbf{v}\|}$ before computing the dot product.

• Maximum value of $D_{\mathbf{u}}f$ is $\|\nabla f\|$, achieved when $\mathbf{u}$ points in the gradient direction.
• Minimum value is $-\|\nabla f\|$, achieved in the opposite direction.
• Zero directional derivative occurs when $\mathbf{u}$ is perpendicular to $\nabla f$ (moving along a level curve).

---

Geometric Applications: Tangent Planes and Linearization

Partial derivatives let you construct linear approximations to surfaces. This is the multivariable analog of tangent lines.

Tangent Planes and Normal Lines

For a surface $z = f(x,y)$ at the point $(a, b, f(a,b))$, the tangent plane is:

$z - f(a,b) = f_x(a,b)(x - a) + f_y(a,b)(y - b)$

The normal vector to this plane is $\langle f_x(a,b),\; f_y(a,b),\; -1 \rangle$. For an implicit surface $F(x,y,z) = 0$, the normal vector is simply $\nabla F$.

The normal line passes through the point in the direction of the normal vector. You can write it parametrically using the normal vector as the direction.

Taylor Series for Multivariable Functions

Multivariable Taylor series extend polynomial approximation to functions of several variables.

• First-order approximation: $f(x,y) \approx f(a,b) + f_x(a,b)(x-a) + f_y(a,b)(y-b)$. Notice this is exactly the tangent plane equation.
• Second-order terms bring in $f_{xx}$, $f_{yy}$, and $f_{xy}$, giving a better approximation near $(a,b)$. These same second-order terms appear in the discriminant used for the second derivative test.

---

Optimization and Applications

Partial derivatives are the primary tool for finding extrema of multivariable functions.

Partial Derivatives in Optimization Problems

Finding and classifying critical points follows a clear procedure:

1. Find critical points by setting $\nabla f = \mathbf{0}$. This means solving $f_x = 0$ and $f_y = 0$ simultaneously.
2. Compute the discriminant at each critical point: $D = f_{xx}f_{yy} - (f_{xy})^2$
3. Classify using $D$:
   • $D > 0$ and $f_{xx} > 0$: local minimum
   • $D > 0$ and $f_{xx} < 0$: local maximum
   • $D < 0$: saddle point
   • $D = 0$: test is inconclusive
4. Check boundaries if the domain is a constrained region. Critical points alone aren't enough; extrema can also occur on the boundary.

Applications in Physics and Engineering

• Heat equation and diffusion use $\frac{\partial u}{\partial t}$ and $\nabla^2 u$ (the Laplacian) to model how temperature distributes over time.
• Fluid dynamics relies on partial derivatives to express conservation laws and flow rates.
• Economics uses partial derivatives for marginal analysis. For instance, $\frac{\partial C}{\partial x}$ gives the marginal cost with respect to one production input while holding others fixed.

---

Quick Reference Table

---

Self-Check Questions

1. If $f(x,y) = x^3y^2 + \sin(xy)$, what is $f_{xy}$? Verify that $f_{xy} = f_{yx}$ and explain why Clairaut's theorem guarantees this.

2. Compare the gradient vector and the directional derivative: how does knowing $\nabla f$ at a point let you find the rate of change in any direction?

3. For the surface $x^2 + y^2 + z^2 = 9$, use implicit differentiation to find $\frac{\partial z}{\partial x}$. What does this derivative represent geometrically?

4. A critical point has $f_{xx} = 4$, $f_{yy} = 3$, and $f_{xy} = 5$. Calculate the discriminant $D$ and classify the critical point. What would change if $f_{xy} = 2$ instead?

5. Contrast finding the tangent plane to $z = f(x,y)$ at a point versus finding the tangent plane to an implicitly defined surface $F(x,y,z) = 0$. What role does the gradient play in each case?