Noesis
4.4.24Multivariable Calculus

Divergence — definition, physical meaning (flux density)

1,686 words8 min readdifficulty · medium1 backlinks
🇮🇳 Yeh note Hinglish mein padho →

WHAT is divergence?

The deep claim: these two definitions are the same thing. The geometric one tells you why you care; the partial-derivative one tells you how to compute. Let's derive the second from the first.

HOW: Derivation from first principles

We compute the flux out of a tiny box and divide by its volume.

Setup. Put a tiny box at point (x,y,z)(x,y,z) with sides Δx,Δy,Δz\Delta x,\Delta y,\Delta z, so V=Δx Δy ΔzV=\Delta x\,\Delta y\,\Delta z. The box has 6 faces; we add the outward flux through each.

Flux through the two faces perpendicular to the xx-axis. The outward normal on the right face (x+Δxx+\Delta x) is +x^+\hat x, on the left face (xx) it is −x^-\hat x. Only F1F_1 contributes (it's the xx-component dotted with ±x^\pm\hat x). Approximate each face's flux as (field value)×(area):

Fluxx≈F1(x+Δx,y,z) Δy Δz⏟out the right  −  F1(x,y,z) Δy Δz⏟out the left\text{Flux}_x \approx \underbrace{F_1(x+\Delta x,y,z)\,\Delta y\,\Delta z}_{\text{out the right}} \;-\; \underbrace{F_1(x,y,z)\,\Delta y\,\Delta z}_{\text{out the left}}

Why the minus? On the left face the outward normal points in −x^-\hat x, so F⃗⋅n^=−F1\vec F\cdot\hat n=-F_1.

Factor out the area Δy Δz\Delta y\,\Delta z: Fluxx≈[F1(x+Δx,y,z)−F1(x,y,z)] Δy Δz\text{Flux}_x \approx \big[F_1(x+\Delta x,y,z)-F_1(x,y,z)\big]\,\Delta y\,\Delta z

Why this step? The bracket is exactly a difference of F1F_1 over a step Δx\Delta x — that's a derivative waiting to happen. Multiply and divide by Δx\Delta x:

Fluxx≈F1(x+Δx,y,z)−F1(x,y,z)Δx Δx Δy Δz  →Δx→0  ∂F1∂x V\text{Flux}_x \approx \frac{F_1(x+\Delta x,y,z)-F_1(x,y,z)}{\Delta x}\,\Delta x\,\Delta y\,\Delta z \;\xrightarrow{\Delta x\to 0}\; \frac{\partial F_1}{\partial x}\,V

By identical reasoning for the yy- and zz-faces: Fluxy≈∂F2∂yV,Fluxz≈∂F3∂zV\text{Flux}_y\approx \frac{\partial F_2}{\partial y}V,\qquad \text{Flux}_z\approx \frac{\partial F_3}{\partial z}V

Add and divide by VV. Total flux ≈(∂F1∂x+∂F2∂y+∂F3∂z)V\approx\left(\dfrac{\partial F_1}{\partial x}+\dfrac{\partial F_2}{\partial y}+\dfrac{\partial F_3}{\partial z}\right)V, so

fluxV  →  ∂F1∂x+∂F2∂y+∂F3∂z=div⁡F⃗.■\frac{\text{flux}}{V}\;\to\;\frac{\partial F_1}{\partial x}+\frac{\partial F_2}{\partial y}+\frac{\partial F_3}{\partial z}=\operatorname{div}\vec F.\qquad\blacksquare

That's it — flux per unit volume = sum of partial derivatives. The geometric definition forced the formula into existence.

Figure — Divergence — definition, physical meaning (flux density)

WHY the formula "feels" right

Worked examples

Common mistakes (steel-manned)

Recall Feynman: explain to a 12-year-old

Pretend the arrows are tiny water streams. Stand at one spot and watch a tiny imaginary balloon around you. If MORE water leaves the balloon than enters, water must be magically appearing inside — that's positive divergence, like a tap. If more comes in than goes out, it's vanishing — like a drain, negative divergence. If exactly as much comes in as goes out, divergence is zero. Divergence is just counting the leftover water per second per tiny balloon.

Flashcards

What kind of quantity is the divergence of a vector field?
A scalar (a number at each point).
Give the geometric definition of divergence.
The limit of outward flux per unit volume as the volume shrinks to a point: lim⁡V→01V∯∂VF⃗⋅n^ dS\lim_{V\to0}\frac1V\oiint_{\partial V}\vec F\cdot\hat n\,dS.
Give the coordinate formula for div⁡F⃗\operatorname{div}\vec F.
∂F1/∂x+∂F2/∂y+∂F3/∂z=∇⋅F⃗\partial F_1/\partial x+\partial F_2/\partial y+\partial F_3/\partial z=\nabla\cdot\vec F.
Why does only ∂F1/∂x\partial F_1/\partial x (not ∂F1/∂y\partial F_1/\partial y) appear in divergence?
Because the flux through xx-faces depends on how the xx-component changes along xx; cross-terms describe rotation (curl), not net outflow.
Physical meaning of positive vs negative divergence?
Positive = source (more flows out, a "tap"); negative = sink ("drain"); zero = incompressible/conserved flow.
Compute div⁡(x,y,z)\operatorname{div}(x,y,z).
1+1+1=31+1+1=3.
Compute div⁡(−y,x,0)\operatorname{div}(-y,x,0) and interpret.
00; pure rotation, no spreading.
Does a large-magnitude uniform field have large divergence?
No — uniform field (5,0,0)(5,0,0) has divergence 00; size ≠ spreading.
Divergence uses which product of ∇\nabla and F⃗\vec F?
The dot product (curl uses cross).

Connections

  • Curl — rotation density (the cross-product sibling; same ∇\nabla, different operation)
  • Divergence Theorem (Gauss) (turns the local flux-density into a global flux: ∭div⁡F⃗ dV=∯F⃗⋅n^ dS\iiint \operatorname{div}\vec F\,dV=\oiint\vec F\cdot\hat n\,dS)
  • Gradient and the del operator (where ∇\nabla comes from)
  • Flux through a surface (the surface integral being averaged)
  • Continuity equation (physics: ∂ρ/∂t+∇⋅(ρv⃗)=0\partial\rho/\partial t+\nabla\cdot(\rho\vec v)=0)
  • Laplacian (∇2f=div⁡(grad⁡f)\nabla^2 f=\operatorname{div}(\operatorname{grad} f))

Concept Map

fluid velocity analogy

source

sink

why you care

derived via tiny box

difference quotient becomes

sum and divide by V

same thing as

equals

Vector field F

Divergence intuition: spreading out

Positive div: tap

Negative div: drain

Geometric definition: flux per unit volume

Flux through 6 faces

Partial derivatives

Coordinate formula del dot F

Scalar at each point

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Socho ki vector field F⃗\vec F ek behte hue paani ki velocity hai. Divergence batata hai ki kisi point pe paani "create" ho raha hai ya "destroy". Ek chhota sa imaginary balloon (tiny volume) lo us point ke around. Agar balloon se zyada paani bahar nikal raha hai jitna andar aa raha hai, to andar koi tap hai — yeh positive divergence. Agar zyada andar aa raha hai, to drain hai — negative divergence. Aur agar jitna aata utna jaata, to divergence zero.

Formula simple hai: div⁡F⃗=∂F1/∂x+∂F2/∂y+∂F3/∂z\operatorname{div}\vec F=\partial F_1/\partial x+\partial F_2/\partial y+\partial F_3/\partial z. Yeh sirf diagonal partial derivatives ka sum hai, aur output ek scalar number hai, vector nahi. Iski derivation bhi seedhi hai: ek chhote box ke 6 faces se flux nikaalo, add karo, volume se divide karo — limit me yahi formula aata hai. Isliye divergence ko hum "flux per unit volume" bolte hain.

Important baat: divergence field ke size se nahi, balki uske change/spreading se related hai. Ek uniform field (5,0,0)(5,0,0) ke bade arrows hone ke bawajood div =0=0 hota hai, kyunki jitna aata utna jaata. Aur rotation field (−y,x)(-y,x) me bhi div =0=0 — woh swirl hai (curl wala kaam), spreading nahi. Yaad rakho: dot product = divergence (tap/drain), cross product = curl (ghoomna). Aage chal ke yeh Divergence Theorem se connect hota hai jo local spreading ko poore surface ke total flux se jodta hai.

Go deeper — visual, from zero

D1FoundationsEvery symbol from zero · 2,247 wordsD2Visual walkthroughThe derivation in pictures · 2,120 wordsD3Worked examplesEvery scenario covered · 2,267 wordsD4ExercisesGraded, full solutions · 2,033 wordsD5Question bankConcept traps, quick reveals · 2,228 words

Test yourself — Multivariable Calculus

L1 RecognitionL2 RecallL3 ProductionL4 ApplicationL5 Mastery

Connections

Divergence theorem (Gauss's theorem) — statement, flux-divergence connectionMaths · 4.4.33Continuity equation — derivation (conservation of mass), ρAv = constPhysics · 2.2.12Understand the role of volume confirmationStock-Market · 3.1.9