A directional derivative measures how fast a multivariable function changes as you move away from a point in a chosen direction. For a surface z = f(x, y), it is the slope you feel if you stand on the surface and walk in a particular direction in the xy-plane. This matters because real quantities such as temperature, height, pressure, and electric potential often change differently depending on direction.
Directional derivatives connect local geometry, rates of change, and optimization in one useful idea.
The most efficient way to compute a directional derivative is to use the gradient vector and a unit direction vector. If u is a unit vector, then D_u f(a, b) = grad f(a, b) · u, so the directional derivative is a dot product. The gradient points in the direction of greatest increase, its opposite gives greatest decrease, and any direction perpendicular to it gives zero instantaneous change.
This makes directional derivatives useful for contour maps, tangent planes, steepest ascent paths, and local linear approximation.
Understanding Calculus: Directional Derivatives
A useful way to understand this idea is to turn a two-variable problem into a one-variable problem. Imagine choosing a straight path across a map, starting at a fixed location. Your east-west and north-south coordinates both change as you travel along that path.
The function value becomes a new function of distance traveled. Ordinary single-variable differentiation then gives the rate of change along the path at the starting instant. This is why directional derivatives are closely connected to the chain rule.
The two partial derivatives tell how the function responds to separate coordinate movements. A chosen direction combines those two responses according to how much of the movement lies along each coordinate axis.
The direction must have length one because a directional derivative is intended to describe change per unit distance. A vector such as three units east and four units north describes a direction, but its length is five. Using it without adjustment measures change over a movement scaled by five, so the result is five times too large for the usual slope interpretation.
Normalizing a vector removes this hidden scaling. This is one of the most common calculation mistakes. Students should first find the vector pointing in the requested direction, then divide each component by its length before using it.
The sign of the result carries physical meaning. A positive value means the measured quantity rises as you begin moving that way. A negative value means it falls.
A value of zero means there is no first-order change along that path at that point. On a contour map, moving along a contour gives zero directional derivative because the function value stays constant to first order.
Moving across closely spaced contours produces a larger magnitude because the quantity changes rapidly over a short distance. This helps explain weather maps, where a large temperature change across a small region signals a strong temperature gradient.
Directional derivatives describe local behavior, not necessarily what happens after a long walk. A path can begin level yet soon rise or fall because the surface may curve. Likewise, a direction of steepest increase can change from point to point.
The gradient method works reliably when the function is differentiable at the point. If a surface has a sharp corner, cusp, or other irregular feature, partial derivatives may not give a complete description of every direction. When solving problems, sketching the direction and checking the unit length can prevent errors.
It is equally important to state the units. For a height function measured in meters over horizontal distance in meters, a directional derivative is meters of vertical change per meter traveled.
Key Facts
- Directional derivative definition: D_u f(a, b) = lim h->0 [f(a + h u1, b + h u2) - f(a, b)] / h.
- Gradient in two variables: grad f(x, y) = <f_x(x, y), f_y(x, y)>.
- Computing formula: D_u f(a, b) = grad f(a, b) · u, where u is a unit vector.
- Unit vector requirement: if v = <p, q>, then u = v / |v| = <p, q> / sqrt(p^2 + q^2).
- Maximum directional derivative: max D_u f(a, b) = |grad f(a, b)|, in the direction of grad f(a, b).
- Minimum directional derivative: min D_u f(a, b) = -|grad f(a, b)|, in the direction of -grad f(a, b).
Vocabulary
- Directional derivative
- The instantaneous rate of change of a function at a point as you move in a specified direction.
- Gradient
- A vector made from the partial derivatives of a function that points in the direction of steepest increase.
- Unit vector
- A vector with length 1 that gives direction without changing the scale of the rate of change.
- Tangent plane
- The plane that best approximates a smooth surface near a given point.
- Level curve
- A curve in the input plane along which a function has a constant value.
Common Mistakes to Avoid
- Using a non-unit direction vector, which makes the directional derivative too large or too small because the vector length scales the dot product.
- Confusing the gradient with the directional derivative, which is wrong because the gradient is a vector while the directional derivative is a scalar rate.
- Forgetting to evaluate the gradient at the given point, which gives a formula for many points instead of the requested local rate of change.
- Assuming the steepest direction is always along the x-axis or y-axis, which is wrong because the steepest direction depends on both partial derivatives together.
Practice Questions
- 1 For f(x, y) = x^2 + 3y^2, find D_u f(1, 2) in the direction u = <3/5, 4/5>.
- 2 For f(x, y) = e^x y, find the directional derivative at (0, 2) in the direction of v = <1, -1>.
- 3 At a point, grad f = <6, 8>. Explain which direction gives the greatest increase, which gives the greatest decrease, and which directions give zero instantaneous change.