Noesis
4.4.11Multivariable Calculus

Gradient perpendicular to level curves - surfaces

1,640 words7 min readdifficulty · medium1 backlinks
🇮🇳 Yeh note Hinglish mein padho →

WHAT is being claimed

HOW to derive it (from first principles)

The cleanest derivation uses a curve lying inside the level set.

Step 1 — Parametrize a curve on the level set. Let r(t)\mathbf{r}(t) be any smooth curve that stays on the level set, so f(r(t))=cfor all t.f(\mathbf{r}(t)) = c \quad\text{for all } t. Why this step? The tangent vector r′(t)\mathbf{r}'(t) of such a curve is, by definition, a direction that lies along the level set. If ∇f⊥r′(t)\nabla f \perp \mathbf{r}'(t) for every such curve, then ∇f\nabla f is perpendicular to the whole level set.

Step 2 — Differentiate both sides with respect to tt. The left side is a composition, so use the multivariable chain rule: ddtf(r(t))=∇f(r(t))⋅r′(t).\frac{d}{dt}f(\mathbf{r}(t)) = \nabla f(\mathbf{r}(t)) \cdot \mathbf{r}'(t). The right side: ddt(c)=0\dfrac{d}{dt}(c) = 0. Why this step? The chain rule is exactly the tool that connects how ff changes to how the position changes. Setting it up lets the constancy of ff do the work.

Step 3 — Read off the conclusion. ∇f(r(t))⋅r′(t)=0.\nabla f(\mathbf{r}(t)) \cdot \mathbf{r}'(t) = 0. A dot product equal to zero means the two vectors are perpendicular. Since r′(t)\mathbf{r}'(t) is the tangent to the level set, ∇f\nabla f is normal to the level set. ■\blacksquare

Figure — Gradient perpendicular to level curves - surfaces

Consequence: equation of the tangent line / plane

Because ∇f\nabla f is the normal vector, you instantly get tangent objects:

Worked examples

Common mistakes

Flashcards

Why is ∇f\nabla f perpendicular to a level curve?
Differentiating f(r(t))=cf(\mathbf r(t))=c via chain rule gives ∇f⋅r′(t)=0\nabla f\cdot \mathbf r'(t)=0; the tangent dots to zero with ∇f\nabla f.
What does ∇f⋅u=0\nabla f\cdot \mathbf u=0 tell you geometrically?
u\mathbf u is a direction along the level set (no change in ff), i.e. tangent to it.
Direction of ∇f\nabla f relative to a level set?
Normal (perpendicular), pointing toward increasing ff (steepest ascent).
Tangent plane to f(x,y,z)=cf(x,y,z)=c at x0\mathbf x_0?
∇f(x0)⋅(x−x0)=0\nabla f(\mathbf x_0)\cdot(\mathbf x-\mathbf x_0)=0.
For f=x2+y2f=x^2+y^2 at (3,4)(3,4), what is ∇f\nabla f and is it perpendicular to the circle?
(6,8)(6,8); yes, it's radial, perpendicular to tangent (−4,3)(-4,3).
Normal to the GRAPH z=f(x,y)z=f(x,y) (not the level curve)?
(fx,fy,−1)(f_x,f_y,-1), from g=f(x,y)−zg=f(x,y)-z.
Implicit slope of f(x,y)=cf(x,y)=c in terms of partials?
dy/dx=−fx/fydy/dx=-f_x/f_y.
Recall Feynman: explain to a 12-year-old

Imagine a hill drawn as a map with circles showing equal heights (like a contour map). If you walk along one circle, you never go up or down — you stay at the same height. The steepest "straight up the hill" direction always cuts straight across those circles, like a "T". It can't run along a circle, because along a circle the height doesn't change at all! That "straight up" arrow is the gradient, and it's always at a right angle to the contour lines.

Connections

Concept Map

f stays constant

points toward

lies inside

tangent vector

differentiate f=c

gives

zero dot product means

opposite in spirit

grad f is normal vector

orthogonal to

Level set f=c

No change in f

Gradient grad f

Steepest increase

Curve r of t on level set

r prime of t along level set

Chain rule

grad f dot r prime = 0

Gradient perpendicular to level set

Tangent line or plane

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Socho ek contour map jaisa — har circle pe height same hai. Yeh circles hi level curves hote hain, jahan function ff ki value constant rehti hai. Ab gradient ∇f\nabla f woh arrow hai jo function ke sabse tezi se badhne ki direction batata hai (steepest ascent). Main idea: yeh dono ek doosre ke bilkul perpendicular (90°) hote hain.

Kyun? Agar tum level curve ke along chalo, toh ff change hi nahi hota. Iska matlab us direction me change rate zero hai, yaani ∇f⋅u=0\nabla f \cdot \mathbf{u} = 0. Aur dot product zero ka matlab hai dono vectors perpendicular. Derivation simple hai: curve r(t)\mathbf r(t) ko level curve pe rakho, f(r(t))=cf(\mathbf r(t))=c likho, dono taraf tt se differentiate karo (chain rule), aur ∇f⋅r′(t)=0\nabla f \cdot \mathbf r'(t)=0 mil jaata hai. Bas ho gaya proof.

Yeh kyun important hai? Kyunki ek baar tumhe pata chal gaya ki ∇f\nabla f ek normal vector hai, toh tangent plane ki equation turant nikal aati hai: ∇f⋅(x−x0)=0\nabla f \cdot (\mathbf x - \mathbf x_0) = 0. Sphere, hyperbola, koi bhi surface ho — same trick. Aur yahi concept aage Lagrange multipliers aur gradient descent me kaam aata hai.

Ek common galti: log sochte hain gradient curve ke along hota hai. Galat! Gradient hamesha curve ke across (perpendicular), uphill direction me hota hai. Yaad rakho: "no change ⟂ most change".

Go deeper — visual, from zero

D1FoundationsEvery symbol from zero · 1,912 wordsD2Visual walkthroughThe derivation in pictures · 2,067 wordsD3Worked examplesEvery scenario covered · 2,895 wordsD4ExercisesGraded, full solutions · 1,989 wordsD5Question bankConcept traps, quick reveals · 1,493 words

Test yourself — Multivariable Calculus

L1 RecognitionL2 RecallL3 ProductionL4 ApplicationL5 Mastery

Connections

Directional derivative — definition, formulaMaths · 4.4.8Chain rule for multivariable functions — all casesMaths · 4.4.7Lagrange multipliers — one and two constraintsMaths · 4.4.15Gradient descent and variants — convergence analysisMaths · 4.10.20Gradient vector ∇f — definition, propertiesMaths · 4.4.9