2.2 Surface Integrals; Green, Stokes and Gauss Theorems
Integrals over surfaces, and the three theorems that trade an integral over a boundary for an integral over what it encloses: Green's theorem in the plane, Stokes' theorem for surfaces in space, and the Gauss divergence theorem for solids.
The fundamental theorem of calculus says that integrating a derivative over an interval gives the values of the function at the two end points — the "boundary" of the interval. Section 2.1 extended this to curves: the line integral of a gradient depends only on the end points. This section completes the picture with three theorems of the same shape, one for each dimension. Green's theorem turns the circulation of a field around a closed plane curve into the integral of its scalar curl over the region inside. Stokes' theorem does the same for a curved surface in space. The Gauss divergence theorem turns the flux of a field out of a closed surface into the integral of its divergence over the solid inside. Along the way we need surfaces, their area, their orientation and the flux of a field through them.
These theorems are the working tools of fluid mechanics, electromagnetism and heat flow — Gauss's law for electric fields and the continuity equation for fluids are the divergence theorem in disguise — and they are the most reliable source of quick marks in GATE MA. A typical question hands you a line integral or a flux integral that is painful to compute directly and easy once the right theorem converts it; other questions test whether you notice that a theorem does not apply, because the field is singular inside the region or the orientation is reversed. We prove all three theorems (for the regions that matter in practice), and collect the traps in the ⚠️ box.
Green's Theorem in the Plane
A simple closed curve in $\mathbb{R}^2$ is the image of a simple closed path; by the Jordan curve theorem it divides the plane into a bounded "inside" and an unbounded "outside". A Green region is a bounded set $D$ whose boundary $\partial D$ consists of finitely many disjoint, piecewise $C^1$ simple closed curves. The boundary is positively oriented if each boundary curve is traversed so that $D$ lies on the left: the outer curve counterclockwise, each inner curve (around a hole) clockwise.
$D$ is of type I if $D=\{(x,y):a\le x\le b,\ g_1(x)\le y\le g_2(x)\}$, and of type II if $D=\{(x,y):c\le y\le d,\ h_1(y)\le x\le h_2(y)\}$, with continuous, piecewise $C^1$ functions $g_1\le g_2$, $h_1\le h_2$. Discs, rectangles, triangles and ellipses are of both types.
Proof. Step 1: a type I region and the $P$ term. Let $D=\{a\le x\le b,\ g_1(x)\le y\le g_2(x)\}$. We show $\oint_{\partial D}P\,dx=-\iint_DP_y\,dA$. By Fubini's theorem and the fundamental theorem of calculus (§1.3, Double and Triple Integrals),
$$\iint_DP_y\,dA=\int_a^b\int_{g_1(x)}^{g_2(x)}P_y\,dy\,dx=\int_a^b\Big[P\big(x,g_2(x)\big)-P\big(x,g_1(x)\big)\Big]dx.$$The positively oriented boundary consists of four pieces: the lower graph $y=g_1(x)$ traversed from $x=a$ to $x=b$, contributing $\int_a^bP(x,g_1(x))\,dx$; the right vertical segment $x=b$ and the left one $x=a$, on which $dx=0$ (either may reduce to a point); and the upper graph traversed from $x=b$ back to $x=a$, contributing $-\int_a^bP(x,g_2(x))\,dx$. The sum is exactly $-\iint_DP_y\,dA$.
Step 2: a type II region and the $Q$ term. In the same way, for $D=\{c\le y\le d,\ h_1(y)\le x\le h_2(y)\}$ the right graph $x=h_2(y)$ is traversed upwards and the left one downwards, so $\oint_{\partial D}Q\,dy$ equals $\int_c^d\big[Q(h_2(y),y)-Q(h_1(y),y)\big]dy$, which is $\iint_DQ_x\,dA$.
Step 3: regions of both types. Adding Steps 1 and 2 proves the theorem for any region that is both of type I and of type II.
Step 4: general regions. Cut $D$ by finitely many segments or arcs into pieces $D_1,\dots,D_m$, each of both types. (For the annulus $1\le x^2+y^2\le4$, the two coordinate axes cut it into four quarter-annuli, each of both types.) Apply Step 3 to each piece and add. The double integrals add up to $\iint_D$. Each cut is part of the boundary of two pieces and is traversed once in each direction, so its contributions cancel; what remains is the integral over $\partial D$, with exactly the positive orientation — in particular, the inner boundary curves come out clockwise. ∎
Every Green region met in practice can be cut up as in Step 4; the theorem holds for all Green regions, but the general proof needs more topology than we want here.
Proof. Apply Green's theorem to $(P,Q)=(0,x)$, $(-y,0)$ and $\left(-\tfrac y2,\tfrac x2\right)$; in each case $Q_x-P_y=1$, so the right side is $\iint_D1\,dA$. ∎
This is how a planimeter works — a mechanical arm that a surveyor runs round the edge of a region on a map, reading off its area — and how a computer finds the area of a polygon from its vertices (the "shoelace formula", Exercise I.4).
Proof. Along a positively oriented boundary the outward normal points to the right of the direction of motion, so $\mathbf{F}\cdot\mathbf{n}\,ds=P\,dy-Q\,dx$ (Definition 2.7). This is the tangential integral $\oint\tilde P\,dx+\tilde Q\,dy$ of the field $(\tilde P,\tilde Q)=(-Q,P)$, and Green's theorem gives $\iint(\tilde Q_x-\tilde P_y)\,dA=\iint(P_x+Q_y)\,dA$. ∎
Let $D$ be the region inside a simple closed curve $C_0$ and outside the disjoint simple closed curves $C_1,\dots,C_k$ (which lie inside $C_0$), all oriented counterclockwise, and let $P,Q$ be $C^1$ on an open set containing $D\cup\partial D$. Then
$$\oint_{C_0}P\,dx+Q\,dy-\sum_{j=1}^k\oint_{C_j}P\,dx+Q\,dy=\iint_D(Q_x-P_y)\,dA.$$In particular, if $Q_x=P_y$ on $D$, then $\oint_{C_0}=\sum_j\oint_{C_j}$: a closed curve can be deformed across any region where the scalar curl vanishes without changing the circulation.
Proof. The positively oriented boundary of $D$ is $C_0$ counterclockwise together with each $C_j$ clockwise, that is $C_0-C_1-\dots-C_k$. Apply Green's theorem, and use the fact that reversing a curve changes the sign of a line integral (§2.1, Gradient, Divergence, Curl and Line Integrals). ∎
Parametric Surfaces and Surface Area
A parametric surface is a $C^1$ map $\mathbf{r}:D\to\mathbb{R}^3$, $\mathbf{r}(u,v)=\big(x(u,v),y(u,v),z(u,v)\big)$, defined on a Green region $D$ of the $uv$-plane; its image $S=\mathbf{r}(D)$ is the surface. The partial derivatives $\mathbf{r}_u,\mathbf{r}_v$ are tangent vectors to the coordinate curves, and
$$\mathbf{N}(u,v)=\mathbf{r}_u\times\mathbf{r}_v$$is the (non-unit) normal vector. The parametrisation is regular at $(u,v)$ if $\mathbf{N}\ne\mathbf{0}$ there; the tangent plane at $\mathbf{r}(u_0,v_0)$ is then $\mathbf{N}(u_0,v_0)\cdot\big(\mathbf{x}-\mathbf{r}(u_0,v_0)\big)=0$. We allow $\mathbf{N}=\mathbf{0}$ and failure of injectivity on the boundary of $D$ (the poles of a sphere, the seam of a cylinder): a set of zero area does not affect any integral below. A piecewise smooth surface is a finite union of such surfaces meeting along curves (the six faces of a cube).
To find the area of $S$, cut $D$ into small rectangles $[u,u+\Delta u]\times[v,v+\Delta v]$. Each is mapped onto a small curved patch that is approximately the parallelogram spanned by $\mathbf{r}_u\Delta u$ and $\mathbf{r}_v\Delta v$, whose area is $|\mathbf{r}_u\times\mathbf{r}_v|\,\Delta u\,\Delta v$. Adding and refining leads to:
Proof. By the chain rule $\tilde{\mathbf{r}}_s=u_s\mathbf{r}_u+v_s\mathbf{r}_v$ and $\tilde{\mathbf{r}}_t=u_t\mathbf{r}_u+v_t\mathbf{r}_v$. Expanding the cross product and using $\mathbf{r}_u\times\mathbf{r}_u=\mathbf{r}_v\times\mathbf{r}_v=\mathbf{0}$ and $\mathbf{r}_v\times\mathbf{r}_u=-\mathbf{r}_u\times\mathbf{r}_v$,
$$\tilde{\mathbf{r}}_s\times\tilde{\mathbf{r}}_t=(u_sv_t-v_su_t)\,\mathbf{r}_u\times\mathbf{r}_v=J\,\mathbf{r}_u\times\mathbf{r}_v.$$Hence $|\tilde{\mathbf{r}}_s\times\tilde{\mathbf{r}}_t|=|\mathbf{r}_u\times\mathbf{r}_v|\circ\Phi\cdot|J|$, and by the change-of-variables formula for double integrals (§1.3),
$$\iint_{D'}|\tilde{\mathbf{r}}_s\times\tilde{\mathbf{r}}_t|\,ds\,dt=\iint_D|\mathbf{r}_u\times\mathbf{r}_v|\,du\,dv.$$The statement about the direction of $\tilde{\mathbf{N}}$ is the sign of $J$ in $\tilde{\mathbf{N}}=J\,\mathbf{N}$. ∎
Orientation and Flux
A smooth surface $S$ is orientable if it carries a continuous unit normal field $\mathbf{n}$; a choice of such a field is an orientation. A connected orientable surface has exactly two orientations, $\mathbf{n}$ and $-\mathbf{n}$. A parametrisation $\mathbf{r}$ is compatible with $\mathbf{n}$ if $\mathbf{n}=\mathbf{N}/|\mathbf{N}|$. The boundary of a bounded solid (a sphere, the surface of a cube) is given the outward orientation unless stated otherwise; a graph $z=g(x,y)$ is usually oriented upwards, with $\mathbf{N}=(-g_x,-g_y,1)$.
If $S$ is oriented by $\mathbf{n}$ and its boundary $\partial S$ consists of closed curves, the induced orientation of $\partial S$ is the direction in which you walk along $\partial S$ with your head pointing along $\mathbf{n}$ and the surface on your left (right-hand rule: thumb along $\mathbf{n}$, fingers along $\partial S$). For the upper hemisphere with the outward normal, $\partial S$ is the equator traversed counterclockwise when seen from above.
Not every surface is orientable. The Möbius strip $\mathbf{r}(u,v)=(x,y,z)$ with $x=(2+u\cos\tfrac v2)\cos v$, $\ y=(2+u\cos\tfrac v2)\sin v$, $\ z=u\sin\tfrac v2$ ($|u|\le\tfrac12$, $0\le v\le2\pi$) has $\mathbf{N}(0,0)=(0,0,2)$ but $\mathbf{N}(0,2\pi)=(0,0,-2)$ at the same point: carrying a normal once round the strip reverses it, so no continuous unit normal exists (Exercise III.8). Flux through a Möbius strip is not defined.
- The flux is the same for all compatible parametrisations and changes sign when the orientation is reversed.
- For the graph $z=g(x,y)$, $(x,y)\in D$, oriented upwards, and $\mathbf{F}=(P,Q,R)$: $$\iint_S\mathbf{F}\cdot\mathbf{n}\,dS=\iint_D\Big(-P\,g_x-Q\,g_y+R\Big)\,dx\,dy,$$ with $P,Q,R$ evaluated at $(x,y,g(x,y))$.
- For the sphere of radius $a$ centred at $\mathbf{0}$, oriented outwards: $\mathbf{n}=\mathbf{r}/a$ and $\mathbf{F}\cdot\mathbf{n}\,dS=a\,\mathbf{F}\cdot\mathbf{r}\,\sin\varphi\,d\varphi\,d\theta$.
Proof. (1) By Theorem 2.15, $\tilde{\mathbf{N}}=J\mathbf{N}$; for compatible parametrisations $J\gt0$, and the change-of-variables formula gives equal integrals; reversing the orientation replaces $\mathbf{n}$ by $-\mathbf{n}$. (2) $\mathbf{r}=(x,y,g(x,y))$ has $\mathbf{r}_x\times\mathbf{r}_y=(1,0,g_x)\times(0,1,g_y)=(-g_x,-g_y,1)$, which points upwards. (3) $\mathbf{N}=a\sin\varphi\,\mathbf{r}$ points outwards. ∎
Stokes' Theorem
Green's theorem relates the circulation around a plane curve to the curl inside it. Stokes' theorem says that the same is true for a curve bounding a curved surface in space, provided we integrate the normal component of the curl.
Proof (for graphs; sketch in general). Let $S$ be the graph $z=g(x,y)$ over a Green region $D$, with $g$ of class $C^2$ and the upward orientation, so that $\partial S$ is the image of the positively oriented $\partial D$ under $(x,y)\mapsto(x,y,g(x,y))$. On $\partial S$ we have $dz=g_x\,dx+g_y\,dy$, so
$$\oint_{\partial S}P\,dx+Q\,dy+R\,dz=\oint_{\partial D}\big(P+Rg_x\big)dx+\big(Q+Rg_y\big)dy,$$where $P,Q,R$ are evaluated at $(x,y,g(x,y))$. Apply Green's theorem to the right side. By the chain rule,
$$\partial_x\big(Q+Rg_y\big)=Q_x+Q_zg_x+\big(R_x+R_zg_x\big)g_y+Rg_{yx},$$ $$\partial_y\big(P+Rg_x\big)=P_y+P_zg_y+\big(R_y+R_zg_y\big)g_x+Rg_{xy}.$$Subtracting, the terms $R_zg_xg_y$ cancel, and $Rg_{yx}-Rg_{xy}=0$ because $g$ is $C^2$:
$$\partial_x(Q+Rg_y)-\partial_y(P+Rg_x)=(Q_x-P_y)-g_x(R_y-Q_z)-g_y(P_z-R_x).$$This is exactly $(\nabla\times\mathbf{F})\cdot(-g_x,-g_y,1)$, and by Theorem 2.16(2) its integral over $D$ is the flux of $\nabla\times\mathbf{F}$ through $S$. That proves the theorem for graphs. A general surface is cut into finitely many pieces, each a graph over one of the three coordinate planes; applying the result to each piece and adding, the integrals along the cuts cancel in pairs (adjacent pieces induce opposite orientations on a common edge), leaving $\oint_{\partial S}$. ∎
Let $\mathbf{F}$ be $C^1$ on an open set $U\subseteq\mathbb{R}^3$.
- Surface independence. If $S_1,S_2\subseteq U$ are oriented surfaces with the same boundary curve, carrying the same induced orientation, then $\iint_{S_1}(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=\iint_{S_2}(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS$.
- Closed surfaces. If $S\subseteq U$ is a closed surface (no boundary), $\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=0$.
- Curl as circulation density. For $\mathbf{p}\in U$ and a unit vector $\mathbf{n}$, let $C_\varepsilon$ be the circle of radius $\varepsilon$ about $\mathbf{p}$ in the plane normal to $\mathbf{n}$, oriented by the right-hand rule. Then $$(\nabla\times\mathbf{F})(\mathbf{p})\cdot\mathbf{n}=\lim_{\varepsilon\to0}\frac{1}{\pi\varepsilon^2}\oint_{C_\varepsilon}\mathbf{F}\cdot d\mathbf{r}.$$
Proof. (1) Both fluxes equal $\oint_{\partial S}\mathbf{F}\cdot d\mathbf{r}$. (2) Cut $S$ by a closed curve $\Gamma$ into two surfaces $S_1,S_2$; the orientations they induce on $\Gamma$ are opposite, so the two fluxes are $\oint_\Gamma$ and $-\oint_\Gamma$, which add to $0$. (3) By Stokes' theorem for the flat disc $D_\varepsilon$ bounded by $C_\varepsilon$, $\oint_{C_\varepsilon}\mathbf{F}\cdot d\mathbf{r}=\iint_{D_\varepsilon}g\,dS$ with $g=(\nabla\times\mathbf{F})\cdot\mathbf{n}$ continuous. Given $\eta\gt0$, for small $\varepsilon$ we have $|g(\mathbf{q})-g(\mathbf{p})|\lt\eta$ on $D_\varepsilon$, so $\left|\frac1{\pi\varepsilon^2}\oint_{C_\varepsilon}-g(\mathbf{p})\right|\le\frac1{\pi\varepsilon^2}\iint_{D_\varepsilon}|g-g(\mathbf{p})|\,dS\le\eta$. ∎
Part (3) justifies the interpretation of curl given in §2.1, and with it Stokes' theorem completes the argument there that curl-free fields on simply connected domains are conservative: a closed curve that bounds a surface inside the domain has circulation $\iint(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=0$.
The Gauss Divergence Theorem
Proof. Step 1. Call $E$ $z$-simple if it consists of the points $(x,y,z)$ with $(x,y)\in D$ and $u_1(x,y)\le z\le u_2(x,y)$, for a Green region $D$ and piecewise $C^1$ functions $u_1\le u_2$. We show that for such $E$,
$$\iint_{\partial E}R\,(\mathbf{k}\cdot\mathbf{n})\,dS=\iiint_ER_z\,dV.$$By Fubini and the fundamental theorem of calculus, $\iiint_ER_z\,dV=\iint_D\big[R(x,y,u_2)-R(x,y,u_1)\big]dA$. The boundary has three parts. On the top $z=u_2(x,y)$ the outward normal points upwards, $\mathbf{N}=(-u_{2,x},-u_{2,y},1)$, so by Theorem 2.16(2) the top contributes $\iint_DR(x,y,u_2)\,dA$. On the bottom $z=u_1(x,y)$ the outward normal points downwards, $\mathbf{N}=(u_{1,x},u_{1,y},-1)$, contributing $-\iint_DR(x,y,u_1)\,dA$. On the vertical sides (above $\partial D$) the normal is horizontal, $\mathbf{k}\cdot\mathbf{n}=0$, and the contribution is $0$. The two sides of the identity agree.
Step 2. If $E$ is also $x$-simple and $y$-simple, the same argument gives $\iint_{\partial E}P\,(\mathbf{i}\cdot\mathbf{n})\,dS=\iiint_EP_x\,dV$ and $\iint_{\partial E}Q\,(\mathbf{j}\cdot\mathbf{n})\,dS=\iiint_EQ_y\,dV$. Adding the three identities gives the theorem, because $\mathbf{F}\cdot\mathbf{n}=P\,\mathbf{i}\cdot\mathbf{n}+Q\,\mathbf{j}\cdot\mathbf{n}+R\,\mathbf{k}\cdot\mathbf{n}$. Balls, boxes, cylinders, cones and tetrahedra are simple in all three directions.
Step 3. A general region (a spherical shell, a solid torus, an L-shaped block) is cut by finitely many surfaces into pieces of the kind in Step 2. Apply Step 2 to each piece and add: the volume integrals add up to $\iiint_E$, and each internal cut appears twice with opposite outward normals, so its fluxes cancel, leaving the flux through $\partial E$. For a shell $a\le|\mathbf{x}|\le b$, the boundary is the outer sphere with normal pointing away from the centre and the inner sphere with normal pointing towards the centre — outward from the solid. ∎
- Divergence as flux density. If $\mathbf{F}$ is $C^1$ near $\mathbf{p}$ and $B_\varepsilon$ is the ball of radius $\varepsilon$ about $\mathbf{p}$, then $\displaystyle\nabla\cdot\mathbf{F}(\mathbf{p})=\lim_{\varepsilon\to0}\frac{1}{\tfrac43\pi\varepsilon^3}\iint_{\partial B_\varepsilon}\mathbf{F}\cdot\mathbf{n}\,dS$.
- Gauss's law. Let $E$ be as in Theorem 2.19 with $\mathbf{0}\notin\partial E$. Then $$\iint_{\partial E}\frac{\mathbf{r}\cdot\mathbf{n}}{r^3}\,dS=\begin{cases}4\pi,&\mathbf{0}\text{ in the interior of }E,\\ 0,&\mathbf{0}\notin E.\end{cases}$$
- Volume. $\operatorname{vol}(E)=\dfrac13\displaystyle\iint_{\partial E}\mathbf{r}\cdot\mathbf{n}\,dS$.
- Green's identities. For $C^2$ functions $f,g$ on a neighbourhood of $E$, writing $\dfrac{\partial g}{\partial n}=\nabla g\cdot\mathbf{n}$: $$\iiint_E\big(f\Delta g+\nabla f\cdot\nabla g\big)\,dV=\iint_{\partial E}f\,\frac{\partial g}{\partial n}\,dS,$$ $$\iiint_E\big(f\Delta g-g\Delta f\big)\,dV=\iint_{\partial E}\Big(f\,\frac{\partial g}{\partial n}-g\,\frac{\partial f}{\partial n}\Big)dS.$$ In particular $\iint_{\partial E}\frac{\partial g}{\partial n}\,dS=\iiint_E\Delta g\,dV$, which is $0$ if $g$ is harmonic.
Proof. (1) is the same continuity argument as in Theorem 2.18(3), with the divergence theorem in place of Stokes' theorem. (2) Let $\mathbf{G}=\mathbf{r}/r^3$, which is $C^1$ with $\nabla\cdot\mathbf{G}=0$ on $\mathbb{R}^3\setminus\{0\}$ (the theorem on radial fields in §2.1). If $\mathbf{0}\notin E$, apply Theorem 2.19 directly: the flux is $0$. If $\mathbf{0}$ is interior, choose $\varepsilon$ so that the ball $B_\varepsilon$ lies inside $E$ and apply Theorem 2.19 to $E'=E\setminus B_\varepsilon$, where $\mathbf{G}$ is $C^1$: $0=\iint_{\partial E}\mathbf{G}\cdot\mathbf{n}\,dS-\iint_{\partial B_\varepsilon}\mathbf{G}\cdot\frac{\mathbf{r}}{\varepsilon}\,dS$ (on the inner sphere the outward normal of $E'$ is $-\mathbf{r}/\varepsilon$). On $\partial B_\varepsilon$, $\mathbf{G}\cdot\mathbf{r}/\varepsilon=\varepsilon^2/\varepsilon^4=\varepsilon^{-2}$, and the area is $4\pi\varepsilon^2$, so the flux is $4\pi$. (3) $\nabla\cdot\mathbf{r}=3$. (4) Apply Theorem 2.19 to $f\nabla g$, whose divergence is $f\Delta g+\nabla f\cdot\nabla g$ (the product rule for $\nabla\cdot(f\mathbf{F})$, §2.1); for the second identity subtract the same formula with $f,g$ exchanged. ∎
Gauss's law is the reason the inverse-square law of gravity and electrostatics is so special: the flux of the field of a point charge through any closed surface around it is the same, $4\pi$ (times the charge), however the surface is shaped — the singular point carries all the divergence.
- Line integral round a closed plane curve: Green. Compute $Q_x-P_y$; awkward terms such as $e^{x^2}$ in $P$ or $\sin(y^3)$ in $Q$ usually vanish. Check the orientation (clockwise = minus) and that $P,Q$ are $C^1$ inside — if there is a singular point inside, cut it out with a small circle (Theorem 2.14).
- Line integral round a closed space curve: Stokes. Choose the simplest spanning surface (often a flat piece of a plane), find its normal from the right-hand rule, integrate $(\nabla\times\mathbf{F})\cdot\mathbf{N}$.
- Flux of a curl through an open surface: Stokes backwards (a line integral round the rim), or replace the surface by a simpler one with the same rim.
- Flux out of a closed surface: divergence theorem; with $\nabla\cdot\mathbf{F}$ constant the answer is constant × volume. Singular point inside? Gauss's law or a small sphere.
- Flux through an open surface with complicated $\mathbf{F}$: close it with a simple lid $L$, then $\iint_S=\iiint_E\nabla\cdot\mathbf{F}\,dV-\iint_L$ (both outward).
- The field must be $C^1$ on the whole region, not just on the boundary. For $(-y,x)/(x^2+y^2)$ on the unit disc, $\oint=2\pi$ but $\iint(Q_x-P_y)\,dA=0$; for $\mathbf{r}/r^3$ on the unit ball, the flux is $4\pi$ but $\iiint\nabla\cdot\mathbf{F}=0$. Cut the singularity out with a small circle or sphere.
- Holes reverse the inner orientation. For an annulus, the positively oriented boundary is the outer circle counterclockwise and the inner circle clockwise.
- Clockwise is negative. Green's theorem and the area formula assume the region is on the left; for a clockwise curve both change sign.
- Outward means downward on a bottom face and inward-pointing on the inner sphere of a shell. Using $\mathbf{N}=(-g_x,-g_y,1)$ on the bottom of a solid flips that contribution.
- Not every surface is orientable. On the Möbius strip a normal carried once round comes back reversed; flux through it and Stokes' theorem for it make no sense.
- $dS$ is not $dA$. $\iint_Sf\,dS$ needs the factor $\sqrt{1+g_x^2+g_y^2}$ (graph) or $a^2\sin\varphi$ (sphere); dropping it gives the integral over the shadow, not over the surface.
- Cover the surface once. $\theta\in[0,4\pi]$ doubles every surface integral; degenerate points such as the poles ($\mathbf{N}=\mathbf{0}$) are harmless.
- Flux of a curl vs flux of a field. Through a closed surface, $\iint(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=0$ always; $\iint\mathbf{F}\cdot\mathbf{n}\,dS$ need not vanish — not even when $\nabla\cdot\mathbf{F}=0$ wherever $\mathbf{F}$ is defined ($\mathbf{r}/r^3$).
- Divergence theorem only for closed surfaces. For an open surface, close it with a lid and subtract the lid's flux (Example 7(b)).
- Unit normal or not. $\iint_S\mathbf{F}\cdot\mathbf{n}\,dS=\iint_D\mathbf{F}\cdot\mathbf{N}\,du\,dv$ with the non-unit $\mathbf{N}=\mathbf{r}_u\times\mathbf{r}_v$: the factor $|\mathbf{N}|$ in $dS$ cancels the one in $\mathbf{n}=\mathbf{N}/|\mathbf{N}|$. Normalising and also keeping $|\mathbf{N}|$ out is a classic slip.
- Only some line integrals give area. $\oint x\,dy$, $-\oint y\,dx$ and $\tfrac12\oint(x\,dy-y\,dx)$ do; $\oint x\,dx$ and $\oint y\,dy$ are always $0$ (exact differentials).
Exercises
- 1 Verify Green's theorem for $P=xy$, $Q=x+y$ on the unit square $[0,1]^2$ by computing both sides.
- 2 Evaluate $\oint_{\partial D}(x^2-y^3)\,dx+(x^3+y^2)\,dy$ for the annulus $D:\ 1\le x^2+y^2\le4$ with positively oriented boundary.
- 3 Use Theorem 2.12 to find the area inside the cardioid $\mathbf{r}(t)=\big((1+\cos t)\cos t,\ (1+\cos t)\sin t\big)$, $0\le t\le2\pi$.
- 4 (Shoelace formula) Show that the area of a polygon with vertices $(x_1,y_1),\dots,(x_m,y_m)$ listed counterclockwise is $\tfrac12\sum_{k=1}^m(x_ky_{k+1}-x_{k+1}y_k)$, with $(x_{m+1},y_{m+1})=(x_1,y_1)$. Compute it for $(0,0),(4,0),(3,2),(1,3)$.
- 5 Compute the outward flux of $\mathbf{F}=(x^3,y^3)$ across the circle $x^2+y^2=4$, by Theorem 2.13 and directly.
- 6 Let $C$ be a positively oriented simple closed curve not passing through the origin. Find all possible values of $\oint_C\dfrac{x\,dx+y\,dy}{x^2+y^2}$ and of $\oint_C\dfrac{x\,dy-y\,dx}{x^2+y^2}$.
- 7 Show that $\mathbf{F}=\dfrac{(-y,\,x)}{4x^2+y^2}$ has zero scalar curl on $\mathbb{R}^2\setminus\{0\}$, and compute its circulation around the unit circle by deforming the circle into the ellipse $4x^2+y^2=1$.
- 8 Let $D$ be a Green region of area $A$ with centroid $(\bar x,\bar y)$. Show that $\bar x=\dfrac1{2A}\oint_{\partial D}x^2\,dy$ and $\bar y=-\dfrac1{2A}\oint_{\partial D}y^2\,dx$.
- 1 For the cone $\mathbf{r}(u,v)=(u\cos v,u\sin v,u)$ find $\mathbf{N}$ at $(u,v)=(1,\tfrac\pi2)$ and the tangent plane at the corresponding point.
- 2 Find the area of the part of the plane $x+2y+2z=6$ in the first octant.
- 3 Show that the area of the spherical cap $x^2+y^2+z^2=a^2$, $z\ge h$ ($-a\le h\le a$) is $2\pi a(a-h)$ — proportional to its height (Archimedes).
- 4 Find the area of the torus $\mathbf{r}(\varphi,\theta)=\big((R+a\cos\varphi)\cos\theta,\ (R+a\cos\varphi)\sin\theta,\ a\sin\varphi\big)$, $0\lt a\lt R$.
- 5 Compute $\iint_S(x^2+y^2)\,dS$ over the sphere of radius $a$ (two ways: directly, and by symmetry using $\iint_S(x^2+y^2+z^2)\,dS$).
- 6 Compute $\iint_Sxyz\,dS$, where $S$ is the part of the plane $x+y+z=1$ in the first octant.
- 7 Show that rotating the graph $y=h(x)\ge0$, $a\le x\le b$, about the $x$-axis gives a surface of area $2\pi\int_a^bh(x)\sqrt{1+h'(x)^2}\,dx$, and evaluate it for $h=\sqrt x$, $0\le x\le2$.
- 8 Prove Lagrange's identity $|\mathbf{r}_u\times\mathbf{r}_v|^2=EG-F^2$, where $E=\mathbf{r}_u\cdot\mathbf{r}_u$, $F=\mathbf{r}_u\cdot\mathbf{r}_v$, $G=\mathbf{r}_v\cdot\mathbf{r}_v$, and compute $E,F,G$ for the sphere of radius $a$.
- 1 Compute the outward flux of $\mathbf{F}=(x,y,z)$ through the sphere of radius $a$ directly from Theorem 2.16(3).
- 2 Compute the upward flux of $\mathbf{F}=(x,y,2z)$ through the triangle $x+y+z=1$, $x,y,z\ge0$.
- 3 Compute the outward flux of $\mathbf{F}=(0,0,z)$ through the upper hemisphere $x^2+y^2+z^2=a^2$, $z\ge0$.
- 4 Verify Stokes' theorem for $\mathbf{F}=(z,x,y)$ on the upper unit hemisphere with the outward normal.
- 5 Let $\mathbf{F}=\left(-y+z\sin x,\ x+z^2,\ e^{xy}\right)$ and let $S$ be the part of $z=9-x^2-y^2$ with $z\ge0$, oriented upwards. Use Stokes' theorem to compute $\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS$.
- 6 Let $C$ be the circle in which the sphere $x^2+y^2+z^2=4$ meets the plane $z=1$, counterclockwise seen from above. Compute $\oint_Cy\,dx+z\,dy+x\,dz$ with Stokes' theorem, and check directly.
- 7 Give two proofs that $\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=0$ when $S$ is the boundary of a solid $E$ and $\mathbf{F}$ is $C^2$ near $E$.
- 8 For the Möbius strip parametrisation in the text, compute $\mathbf{N}$ at $(u,v)=(0,0)$ and $(0,2\pi)$ and explain why the strip is not orientable.
- 1 Find the outward flux of $\mathbf{F}=(x^3,y^3,z^3)$ through the sphere of radius $a$.
- 2 Find the outward flux of $\mathbf{F}=(x^2,y^2,z^2)$ through the surface of the cube $[0,1]^3$.
- 3 Find the outward flux of $\mathbf{F}=(xy,yz,zx)$ through the boundary of the cylinder $x^2+y^2\le1$, $0\le z\le3$.
- 4 Find the outward flux of $\mathbf{r}=(x,y,z)$ through the boundary of the tetrahedron $x,y,z\ge0$, $x+y+z\le1$, by the divergence theorem and directly (which faces contribute?).
- 5 Use Theorem 2.20(3) to find the volume of a cone of height $h$ and base radius $R$ (vertex at the origin, axis along $\mathbf{k}$).
- 6 Compute $\iint_S\dfrac{\mathbf{r}\cdot\mathbf{n}}{r^3}\,dS$ (outward) for (a) the sphere $(x-3)^2+y^2+z^2=1$; (b) the surface of the cube $[-1,1]^3$; (c) the ellipsoid $\tfrac{x^2}4+\tfrac{y^2}9+z^2=1$.
- 7 Show that if $g$ is harmonic on a neighbourhood of a solid $E$ and $g=0$ on $\partial E$, then $g\equiv0$ on $E$. (Use Green's first identity with $f=g$.)
- 8 Let $\mathbf{F}=(x+zy^2,\ y,\ z+x^2)$. Find the upward flux of $\mathbf{F}$ through the upper unit hemisphere by closing it with the unit disc.
- 9 Prove that if $\mathbf{F}$ is $C^1$ on $\mathbb{R}^3$ and its outward flux through every sphere is $0$, then $\nabla\cdot\mathbf{F}\equiv0$.
E1. GATE MA · MCQ Let $C$ be the boundary of the triangle with vertices $(0,0)$, $(2,0)$, $(0,2)$, traversed clockwise. Then $\displaystyle\oint_C\left(2y+e^{x^3}\right)dx+\left(5x-ye^y\right)dy$ equals (A) $6$; (B) $-6$; (C) $3$; (D) $-12$.
E2. GATE MA · MCQ Let $S$ be the ellipsoid $\dfrac{x^2}4+\dfrac{y^2}9+z^2=1$ with the outward normal, and $\mathbf{F}=\left(x+\sin(yz),\ y+e^{xz},\ z\right)$. The flux $\iint_S\mathbf{F}\cdot\mathbf{n}\,dS$ equals (A) $24\pi$; (B) $8\pi$; (C) $36\pi$; (D) $72\pi$.
E3. GATE MA · MSQ Let $S$ be the unit sphere $x^2+y^2+z^2=1$ with outward normal $\mathbf{n}$. Which of the following are true? (One or more options may be correct.) (A) $\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS=0$ for every $C^2$ vector field $\mathbf{F}$ on $\mathbb{R}^3$. (B) $\iint_S\mathbf{G}\cdot\mathbf{n}\,dS=0$ for every $C^1$ field $\mathbf{G}$ on $\mathbb{R}^3\setminus\{0\}$ with $\nabla\cdot\mathbf{G}=0$ on $\mathbb{R}^3\setminus\{0\}$. (C) $\iint_S(x^2,y^2,z^2)\cdot\mathbf{n}\,dS=0$. (D) $\iint_S\nabla f\cdot\mathbf{n}\,dS=0$ for every $C^2$ function $f$ on $\mathbb{R}^3$ with $\Delta f=0$.
E4. GATE MA · MSQ Let $S$ be the part of the cone $z=\sqrt{x^2+y^2}$ with $0\le z\le1$. Which of the following are true? (One or more options may be correct.) (A) The area of $S$ is $\sqrt2\,\pi$. (B) $\iint_Sz\,dS=\dfrac{2\sqrt2\,\pi}3$. (C) The upward unit normal to $S$ at $(x,y,z)\ne\mathbf{0}$ is $\dfrac{1}{\sqrt2}\left(\dfrac xz,\dfrac yz,1\right)$. (D) The upward flux of $\mathbf{F}=(0,0,1)$ through $S$ is $\sqrt2\,\pi$.
E5. GATE MA · NAT Let $\mathbf{F}=\left(z-y,\ x+z,\ 2y\right)$ and let $C$ be the curve of intersection of the cylinder $x^2+y^2=1$ with the plane $x+y+z=2$, oriented counterclockwise when viewed from above. The value of $\oint_C\mathbf{F}\cdot d\mathbf{r}$ is ______ (correct to two decimal places).
E6. GATE MA · NAT The area of the helicoid $\mathbf{r}(u,v)=(u\cos v,\ u\sin v,\ v)$, $0\le u\le1$, $0\le v\le2\pi$, is ______ (correct to two decimal places).
Suggested plan — 4 lectures × 60 min.
- Green's theorem. Definition 2.10 with pictures of type I/II regions and an annulus. The proof of Theorem 2.11 (Steps 1 and 4 on the board), Theorems 2.12–2.14, Examples 1–3. Project Interactive 1: first $(-y,x)$ with 1, 9, 121 cells (the sum never changes), then the vortex on the disc and on the annulus.
- Surfaces. Definition 2.11 and the four standard parametrisations; the parallelogram heuristic for $dS$; Theorem 2.15 (the one-line cross-product computation); Example 4. Orientation, the Möbius strip (bring a paper strip and a marker), flux, Theorem 2.16, Example 5.
- Stokes' theorem. Statement with the right-hand rule, the proof for graphs (it is Green plus the chain rule — let students do the subtraction), Theorem 2.18 and Example 6 both ways.
- Divergence theorem. Proof for $z$-simple regions, cut-and-cancel for general regions, Theorem 2.20 (Gauss's law in detail), the strategy box, Example 7, Interactive 2 (point source: centred, shifted by 0.5, shifted by 2.2). GATE corner E2 and E3 live.
Discussion questions. (1) A butterfly net (a hemisphere-like bag on a circular rim) and a flat disc share a rim. Why does the flux of $\nabla\times\mathbf{F}$ through them agree, but the flux of $\mathbf{F}$ need not? (2) Green's theorem is a special case of both Stokes' theorem and the divergence theorem — explain both ways. (3) Why is the flux of $\mathbf{r}/r^3$ through a closed surface around the origin independent of the shape of the surface, while the flux of $\mathbf{r}/r^2$ is not?
Misconceptions to watch for. Applying Green or Gauss across a singularity; forgetting that inner boundary curves run clockwise; taking the upward normal on the bottom of a solid; $dS=dx\,dy$; using the divergence theorem on an open surface; thinking the Möbius strip just has "two sides that meet"; confusing $\iint\mathbf{F}\cdot\mathbf{n}\,dS$ with $\iint|\mathbf{F}|\,dS$.
Suggested homework. I(2, 4, 6, 7), II(2, 3, 4, 7), III(2, 4, 5, 6), IV(1, 3, 4, 6, 8); GATE corner E1, E4, E5, E6.
Research question. A farmer measures the area of an irregular field by walking its boundary with a phone GPS that records a position every few metres, each with an error of about $3$ m. How accurate is the area computed from these points, and how should the walk be planned?
Why it matters. Land records, crop-insurance claims and subsidy payments in many countries (India's land-record digitisation, the European Union's area-based farm payments) depend on field areas measured exactly this way. The formula every GPS app uses is Green's theorem in its discrete form — the shoelace formula — and the same mathematics drove the mechanical planimeters that surveyors used from the 1850s.
What to do.
- Derive the shoelace formula from $\tfrac12\oint(x\,dy-y\,dx)$ (Exercise I.4).
- Take a test field with known area — an ellipse with semi-axes $150$ m and $80$ m, or a polygon you draw — and sample $N$ points along its boundary. Compute the area without noise and record the discretisation error.
- Add independent Gaussian noise ($\sigma=3$ m) to each coordinate and repeat $2000$ times. Estimate the bias and the standard deviation of the computed area for $N=20,50,100,400$.
- Compare your standard deviation with the linearised prediction $\tfrac\sigma2\sqrt{\sum_k|\mathbf{r}_{k+1}-\mathbf{r}_{k-1}|^2}$ (derive it by differentiating the shoelace sum with respect to each coordinate).
Tools: Python with numpy, or a spreadsheet; optionally a real GPS track exported from a phone.
Hand in: a one-page report with a table (N, discretisation error, noise bias, standard deviation) and a recommendation: how many points, and is it better to walk slowly or to walk twice?
Going further. Read how Amsler's polar planimeter (1854) turns $\oint x\,dy$ into the rotation of a small wheel, and why the instrument works even though its arm is not attached to a fixed Cartesian frame.
🧪 Self-check quiz
Q1. If $C$ is the positively oriented boundary of a Green region $D$, which integral equals the area of $D$?
Q2. The outward flux of $\mathbf{F}=(x,y,z)$ through the sphere of radius $2$ centred at the origin is
Q3. Let $S$ be the boundary of a bounded solid and $\mathbf{F}$ a $C^2$ field on $\mathbb{R}^3$. Then $\iint_S(\nabla\times\mathbf{F})\cdot\mathbf{n}\,dS$
Q4. For $\mathbf{F}=(-y,x)/(x^2+y^2)$, the circulation around the ellipse $\dfrac{x^2}9+\dfrac{y^2}4=1$, counterclockwise, is
Q5. For the graph $z=g(x,y)$ over $D$, the surface area element is