The Gradient Vector

---

Concept Explanation

The Vector Differential Operator Del ($\nabla$)

The symbol $\nabla$ is the vector differential operator "del", defined in Cartesian coordinates as:

When del acts upon a scalar function $f(x, y, z)$, it transforms the scalar field into a vector field $\nabla f$. At each spatial coordinate $(x, y, z)$, $\nabla f$ assigns a specific vector with both magnitude and spatial orientation.

       Level Curve f(x, y) = c_2
             .---.
            /     \       ^  grad f (Direction of steepest ascent)
           /   P   \     /|
          |    *----|---' |  (Orthogonal to tangent vector T)
           \   |   /      |
            \  v  /       +-- Tangent vector T along level curve
             '---'
       Level Curve f(x, y) = c_1   (where c_2 > c_1)

Three Fundamental Geometric Properties

1. Direction of Steepest Ascent: If you stand at a point $P(x_0, y_0)$ on a topographic terrain $z = f(x, y)$, walking in the exact compass direction of $\nabla f(x_0, y_0)$ produces the steepest possible uphill climb. Conversely, moving in the opposite direction, $-\nabla f(x_0, y_0)$, gives the direction of steepest descent (the path naturally carved by rainwater runoff).

2. Magnitude as Maximum Rate of Change: The Euclidean norm of the gradient vector:

   $$\|\nabla f\| = \sqrt{\left(\frac{\partial f}{\partial x}\right)^2 + \left(\frac{\partial f}{\partial y}\right)^2} \quad \text{or} \quad \|\nabla f\| = \sqrt{f_x^2 + f_y^2 + f_z^2}$$
   quantifies the maximum possible slope or instantaneous rate of change per unit distance traveled in the domain.

3. Orthogonality to Level Sets: Let $\vec{r}(t) = \langle x(t), y(t) \rangle$ parameterize a smooth level curve $f(x, y) = c$. Differentiating both sides with respect to $t$ via the Multivariable Chain Rule:

   $$\frac{d}{dt}[f(x(t), y(t))] = \frac{d}{dt}[c] \implies \frac{\partial f}{\partial x}\frac{dx}{dt} + \frac{\partial f}{\partial y}\frac{dy}{dt} = 0$$
   Recognizing the left side as a vector dot product:
   $$\nabla f \cdot \vec{r}'(t) = 0$$
   Because $\vec{r}'(t)$ is the velocity vector tangent to the level curve, $\nabla f$ is strictly perpendicular (orthogonal) to the tangent line of every level curve (and tangent plane of every level surface).

Step-by-Step Evaluation Strategy

1. Differentiate Symbolically First: Compute all first-order partial derivatives $f_x, f_y$ (and $f_z$) treating all other variables as constants. Never substitute numerical values prior to differentiation.
2. Assemble the Vector: Combine the scalar partial derivatives into bracket format $\langle f_x, f_y, f_z \rangle$ or unit vector form $f_x\hat{i} + f_y\hat{j} + f_z\hat{k}$.
3. Evaluate at the Specified Point: Substitute the coordinate values of point $P(x_0, y_0, z_0)$ into the components to obtain a constant vector in $\mathbb{R}^2$ or $\mathbb{R}^3$.

---

Engineering Applications

• Civil & Environmental Engineering: Hydrologists and site grading engineers model retention basins and municipal watersheds with elevation surfaces $h(x, y)$. Water runoff naturally cascades along the direction of steepest descent, $-\nabla h$, while retaining walls and road alignments follow contour curves orthogonal to $\nabla h$.
• Electrical Engineering: Electrostatic potential fields $V(x, y, z)$ generate electric fields via the negative gradient: $\vec{E} = -\nabla V$. Charged particles accelerate along lines of electric force governed by $\nabla V$.
• Mechanical & Materials Engineering: Fourier's Law of Heat Conduction states that conductive heat flux $\vec{q}$ is proportional to the negative temperature gradient: $\vec{q} = -k\nabla T$. Thermal engineers align cooling fins and heat sinks directly parallel to $-\nabla T$ to maximize heat dissipation away from microprocessors.

---

Example Problems

Example 1: Two-Variable Gradient Computation & Evaluation

Consider the surface $f(x, y) = 4x^3y^2 - 3x^2 + 7y$.

1. Find the general gradient vector $\nabla f(x, y)$.
2. Evaluate $\nabla f$ at the point $P(1, -2)$.
3. State the direction of steepest ascent and the maximum rate of change at $P$.

Solution

1. Compute the first-order partial derivatives:

   $$f_x(x, y) = \frac{\partial}{\partial x}(4x^3y^2 - 3x^2 + 7y) = 12x^2y^2 - 6x$$
   $$f_y(x, y) = \frac{\partial}{\partial y}(4x^3y^2 - 3x^2 + 7y) = 8x^3y + 7$$
   The general gradient vector is:
   $$\nabla f(x, y) = \langle 12x^2y^2 - 6x, \, 8x^3y + 7 \rangle = (12x^2y^2 - 6x)\hat{i} + (8x^3y + 7)\hat{j}$$

2. Evaluate at $P(1, -2)$:

   $$f_x(1, -2) = 12(1)^2(-2)^2 - 6(1) = 12(4) - 6 = 48 - 6 = 42$$
   $$f_y(1, -2) = 8(1)^3(-2) + 7 = -16 + 7 = -9$$
   $$\nabla f(1, -2) = \langle 42, -9 \rangle = 42\hat{i} - 9\hat{j}$$

3. Direction of steepest ascent and maximum rate of change:

   • Direction of steepest ascent: $\vec{u} = \frac{\nabla f}{\|\nabla f\|} = \frac{\langle 42, -9 \rangle}{\sqrt{42^2 + (-9)^2}} = \frac{\langle 42, -9 \rangle}{\sqrt{1764 + 81}} = \frac{\langle 42, -9 \rangle}{\sqrt{1845}} = \left\langle \frac{14}{\sqrt{205}}, -\frac{3}{\sqrt{205}} \right\rangle$
   • Maximum rate of change: $\|\nabla f(1, -2)\| = \sqrt{1845} = 3\sqrt{205} \approx 42.95$

---

Example 2: Three-Variable Transcendental Gradient

Let $f(x, y, z) = z^2\ln(x) + x e^{2y}$.

1. Compute the general gradient $\nabla f(x, y, z)$.
2. Evaluate $\nabla f$ at the coordinate point $Q(1, 0, 3)$.

Solution

1. Compute the three partial derivatives:

   $$f_x = \frac{\partial}{\partial x}[z^2\ln(x) + x e^{2y}] = \frac{z^2}{x} + e^{2y}$$
   $$f_y = \frac{\partial}{\partial y}[z^2\ln(x) + x e^{2y}] = 2x e^{2y}$$
   $$f_z = \frac{\partial}{\partial z}[z^2\ln(x) + x e^{2y}] = 2z\ln(x)$$
   The general gradient vector is:
   $$\nabla f(x, y, z) = \left\langle \frac{z^2}{x} + e^{2y}, \, 2x e^{2y}, \, 2z\ln(x) \right\rangle$$

2. Evaluate at $Q(1, 0, 3)$:

   $$f_x(1, 0, 3) = \frac{3^2}{1} + e^{2(0)} = 9 + 1 = 10$$
   $$f_y(1, 0, 3) = 2(1)e^{2(0)} = 2(1)(1) = 2$$
   $$f_z(1, 0, 3) = 2(3)\ln(1) = 6(0) = 0$$
   $$\nabla f(1, 0, 3) = \langle 10, 2, 0 \rangle = 10\hat{i} + 2\hat{j} + 0\hat{k}$$

---

Connections & References

• Prerequisites:
  • Partial Derivatives — Foundation for component calculations $f_x, f_y, f_z$.
  • Vectors — Vector notation, dot products, and unit vector algebra.
  • Multivariable Functions — Domains, scalar fields, and coordinate spaces.
• Direct Successors & Core Skills:
  • Directional Derivatives — Evaluating rates along arbitrary direction vectors: $D_{\vec{u}} f = \nabla f \cdot \vec{u}$.
  • Tangent Planes to Level Surfaces — Utilizing $\vec{n} = \nabla F$ to construct tangent plane equations.
  • Level Curves and Contour Maps — Visualizing gradient vectors orthogonal to elevation contours.
• Course Sessions:
  • Session 12 - Gradient and Directional Derivatives