Calculus III: Gradient and Directional Derivatives

In multivariable calculus, understanding how a function changes isn't limited to the x- or y-axis. The concepts of the directional derivative and the gradient vector unlock the ability to analyze rates of change in any direction and to find the path of steepest ascent on a surface. These tools are indispensable in engineering for modeling phenomena like heat dissipation, optimizing material properties, and navigating complex potential fields.

Understanding the Directional Derivative

The partial derivative $f_x$ tells you the rate of change of a function $f(x,y)$ as you move in the positive x-direction. Similarly, $f_y$ gives the rate of change in the y-direction. But what if you need to know how $f$ changes as you travel in an arbitrary direction, say northwest? This is precisely what the directional derivative measures.

Formally, the directional derivative of $f$ at a point $(x_0,y_0)$ in the direction of a unit vector $\mathbf{u}=⟨a,b⟩$ is defined as: $$D_{\mathbf{u}}f(x_0,y_0)=\underset{h\to 0}{\lim }\frac{f(x_0+ha,y_0+hb)-f(x_0,y_0)}{h}.$$ Intuitively, it is the slope of the function $f$ in the specific direction of $\mathbf{u}$. For practical computation, if $f$ is differentiable, the directional derivative can be calculated using a dot product: $$D_{\mathbf{u}}f(x,y)=f_x(x,y)a+f_y(x,y)b.$$ Consider the function $f(x,y)=x^2+xy$ at the point $(1,2)$. The partials are $f_x=2x+y$ and $f_y=x$. At $(1,2)$, $f_x=4$ and $f_y=1$. To find the rate of change in the direction of the vector $⟨3,4⟩$, you first convert it to a unit vector: $\mathbf{u}=⟨\frac{3}{5},\frac{4}{5}⟩$. The directional derivative is then $D_{\mathbf{u}}f(1,2)=(4)(\frac{3}{5})+(1)(\frac{4}{5})=\frac{16}{5}$.

The Gradient Vector and Steepest Ascent

Notice that the formula $f_{x}a+f_{y}b$ is the dot product of the vector $⟨f_x,f_y⟩$ and the unit vector $\mathbf{u}$. This special vector of partial derivatives is called the gradient, denoted $\nabla f$ (read "nabla f"). $$\nabla f(x,y)=⟨f_x(x,y),f_y(x,y)⟩.$$ For a function of three variables, $\nabla f=⟨f_x,f_y,f_z⟩$.

Using the gradient, the directional derivative formula becomes beautifully concise: $$D_{\mathbf{u}}f=\nabla f\cdot \mathbf{u}.$$ Recall that the dot product $\mathbf{v}\cdot \mathbf{u}=\Vert \mathbf{v}\Vert \Vert \mathbf{u}\Vert \cos \theta =\Vert \mathbf{v}\Vert \cos \theta$, where $\theta$ is the angle between the vectors. Since $\Vert \mathbf{u}\Vert =1$, we have $D_{\mathbf{u}}f=\Vert \nabla f\Vert \cos \theta$. This reveals a critical property: the maximum possible value of $\cos \theta$ is 1, which occurs when $\theta =0$—that is, when $\mathbf{u}$ points in the same direction as $\nabla f$. Therefore, the gradient vector $\nabla f$ points in the direction of the steepest ascent of the function at that point. The magnitude $\Vert \nabla f\Vert$ gives the maximum rate of increase (the slope in that steepest direction).

The Gradient and Level Curves

A level curve of a function $f(x,y)$ is the set of points satisfying $f(x,y)=k$ for some constant $k$. Imagine a topographic map where each contour line is a level curve of the elevation function. A key geometric property of the gradient is that it is always perpendicular to the level curve (or level surface in 3D) passing through that point.

Why? Along a level curve, the function value does not change. Therefore, the directional derivative in any direction tangent to the level curve must be zero. For a tangent direction vector $\mathbf{T}$, we have $0=D_{\mathbf{T}}f=\nabla f\cdot \mathbf{T}$. A dot product of zero means the vectors are perpendicular. This property is foundational for constructing tangent lines and planes.

Finding Tangent Planes Using Gradients

For a surface defined by $F(x,y,z)=0$ (e.g., $z-f(x,y)=0$ or $x^2+y^2+z^2-1=0$), the gradient vector $\nabla F$ at a point $P_0=(x_0,y_0,z_0)$ on the surface is normal (perpendicular) to the surface at that point. This provides the most direct method for finding the equation of the tangent plane.

If $\nabla F(P_0)=⟨A,B,C⟩$, then the equation of the tangent plane at $P_0$ is: $$A(x-x_0)+B(y-y_0)+C(z-z_0)=0.$$ For a surface given explicitly as $z=f(x,y)$, you can rewrite it as $F(x,y,z)=f(x,y)-z=0$. Then $\nabla F=⟨f_x,f_y,-1⟩$, and the tangent plane equation simplifies to the more familiar form: $z-z_0=f_x(x_0,y_0)(x-x_0)+f_y(x_0,y_0)(y-y_0)$.

Applications: Heat Flow and Potential Fields

These concepts are not just abstract; they are the language of physical models in engineering.

• Heat Flow: Temperature in a medium is a scalar field $T(x,y,z)$. Heat flows from hot to cold, and it does so in the direction where the temperature decreases most rapidly. This is precisely the direction of $-\nabla T$, the negative gradient. The magnitude $\Vert \nabla T\Vert$ quantifies the temperature gradient, which drives the rate of heat transfer (governed by Fourier's law). Engineers use this to analyze heat dissipation in electronic components or insulation effectiveness.
• Potential Fields: In fluid dynamics or electromagnetism, we often work with a potential function $\varphi (x,y,z)$. The force or velocity field experienced by a particle is given by the negative gradient of the potential: $\mathbf{F}=-\nabla \varphi$. For instance, in a gravitational field, the potential $\varphi =-GM/r$ leads to the force vector $\mathbf{F}=-\nabla \varphi$, which points toward the mass $M$. The gradient points "uphill" in potential, but the physical force pushes objects "downhill." Level curves of $\varphi$ are equipotential lines, and, as established, the force field $\mathbf{F}$ is perpendicular to these lines.

Common Pitfalls

1. Using a Non-Unit Vector for Direction: A frequent error is computing $D_{\mathbf{u}}f$ using a direction vector $\mathbf{v}$ that is not a unit vector. The formula $D_{\mathbf{u}}f=\nabla f\cdot \mathbf{u}$ requires $\mathbf{u}$ to have length 1. Always remember to normalize: $\mathbf{u}=\frac{\mathbf{v}}{\Vert \mathbf{v}\Vert }$.
2. Confusing Gradient with Directional Derivative: The gradient $\nabla f$ is a vector that points in the direction of steepest ascent. The directional derivative $D_{\mathbf{u}}f$ is a scalar value representing the rate of change in a specific direction. They are related but distinct concepts.
3. Misinterpreting the "Steepest" Direction: The gradient points in the direction of steepest ascent. The direction of steepest descent is the exact opposite: $-\nabla f$. In optimization algorithms like gradient descent, you move against the gradient to minimize a function.
4. Incorrect Tangent Plane for Implicit Surfaces: When finding a tangent plane to a surface defined implicitly as $F(x,y,z)=c$, the normal vector is $\nabla F$, not $\nabla z$. You must compute the partial derivatives of $F$ itself with respect to $x$, $y$, and $z$.

Summary

• The directional derivative $D_{\mathbf{u}}f$ quantifies the rate of change of a multivariable function in any given direction specified by a unit vector $\mathbf{u}$. It is computed as $D_{\mathbf{u}}f=\nabla f\cdot \mathbf{u}$.
• The gradient vector $\nabla f=⟨f_x,f_y⟩$ is the fundamental tool. It points in the direction of the steepest ascent of the function, and its magnitude equals the maximum rate of increase.
• Geometrically, the gradient at a point is perpendicular to the level curve (or level surface) of the function passing through that point.
• This perpendicular property allows the gradient to provide a normal vector for tangent planes. For a surface $F(x,y,z)=0$, the tangent plane at $P_0$ is given by $\nabla F(P_0)\cdot ⟨x-x_0,y-y_0,z-z_0⟩=0$.
• In engineering, these concepts model real-world phenomena: heat flows in the direction of the negative temperature gradient ($-\nabla T$), and force fields are often the negative gradient of a potential function ($\mathbf{F}=-\nabla \varphi$).