Directional Derivatives in Non-Unit Directions
You'll learn how to compute directional derivatives when the given direction vector isn't normalized to length 1.
The Core Challenge
Up until now, you've computed directional derivatives using the formula:
D_u f = ∇f · u
But this formula requires that u be a unit vector (length = 1). What happens when someone hands you a direction like v = ⟨3, 4⟩ or v = ⟨1, 2, -2⟩? These aren't unit vectors—they're just directions with arbitrary lengths.
The Solution: Normalize First
The fix is simple: convert your direction vector into a unit vector before applying the formula.
Step-by-Step Process
- Given: A function f and a direction vector v (not necessarily unit length)
- Find the magnitude: ||v|| = √(v₁² + v₂² + ... + vₙ²)
- Normalize: u = v/||v|| = ⟨v₁/||v||, v₂/||v||, ...⟩
- Compute: D_u f = ∇f · u
Example
Find the directional derivative of f(x,y) = x² + 3y at point (1, 2) in direction v = ⟨3, 4⟩.
- First, normalize: ||v|| = √(3² + 4²) = 5
- So u = ⟨3/5, 4/5⟩
- Gradient: ∇f = ⟨2x, 3⟩, so ∇f(1,2) = ⟨2, 3⟩
- Directional derivative: D_u f = ⟨2, 3⟩ · ⟨3/5, 4/5⟩ = 6/5 + 12/5 = 18/5
Why This Matters
The directional derivative measures rate of change per unit distance. If you don't normalize, you're measuring rate per scaled distance, which distorts the meaning.
Key Takeaway: Always normalize non-unit direction vectors by dividing by their magnitude before computing directional derivatives; the formula D_u f = ∇f · u requires a unit vector.