Level 5 — MasteryMultivariable Calculus

Multivariable Calculus

2 minutes100 marksprintable — key stays hidden on paper

Time limit: 2 hours 30 minutes Total marks: 100 Instructions: Attempt all three questions. Full rigour is expected: state theorems you invoke and verify their hypotheses. Calculators/CAS may be used only for arithmetic; all derivations must be shown. Use LaTeX\LaTeX-style math throughout.


Question 1 — Fields, Potentials, and the Unification of Integral Theorems (34 marks)

Consider the vector field on R3{z-axis}\mathbb{R}^3\setminus\{z\text{-axis}\}

F(x,y,z)=(yx2+y2+2xz)i+(xx2+y2)j+(x2+cosz)k.\mathbf{F}(x,y,z) = \left(\frac{-y}{x^2+y^2} + 2xz\right)\mathbf{i} + \left(\frac{x}{x^2+y^2}\right)\mathbf{j} + \left(x^2 + \cos z\right)\mathbf{k}.

(a) Compute ×F\nabla\times\mathbf{F} and F\nabla\cdot\mathbf{F}. Show all component derivatives. (6)

(b) The field F\mathbf{F} has zero curl on its domain, yet is not conservative there. Explain precisely why zero curl does not imply conservative here, referencing the topology of the domain. Then decompose F=G+ϕ\mathbf{F} = \mathbf{G} + \nabla\phi where G=(yx2+y2,xx2+y2,0)\mathbf{G} = \left(\tfrac{-y}{x^2+y^2}, \tfrac{x}{x^2+y^2}, 0\right), and find the scalar potential ϕ\phi for the remaining part. (9)

(c) Let CC be the circle x2+y2=4x^2+y^2=4, z=1z=1, oriented counterclockwise viewed from above. Compute CFdr\oint_C \mathbf{F}\cdot d\mathbf{r} directly by parametrization. (7)

(d) A student claims: "By Stokes' theorem, CFdr=S(×F)dS=0\oint_C \mathbf{F}\cdot d\mathbf{r} = \iint_S (\nabla\times\mathbf{F})\cdot d\mathbf{S} = 0 for any surface SS bounded by CC." Identify the flaw and state precisely which hypothesis of Stokes' theorem fails. (4)

(e) Explain in 4–6 sentences how Green's theorem, Stokes' theorem, and the Divergence theorem are all special cases of the generalized Stokes theorem Mω=Mdω\int_{\partial M}\omega = \int_M d\omega. Identify what plays the role of MM, M\partial M, and ω\omega in each. (8)


Question 2 — Constrained Optimization meets Physics: Equilibrium of a Charged Particle (33 marks)

A particle constrained to move on the surface S:  x2+y2+4z2=12S:\; x^2 + y^2 + 4z^2 = 12 experiences a potential energy

U(x,y,z)=x+2y+z.U(x,y,z) = x + 2y + z.

Equilibria correspond to critical points of UU on SS.

(a) Use Lagrange multipliers to find all points on SS where U=λg\nabla U = \lambda \nabla g (with g=x2+y2+4z212g = x^2+y^2+4z^2-12). Give all critical points and the corresponding λ\lambda. (9)

(b) Evaluate UU at each critical point and identify the global minimum and maximum of UU on SS (a compact ellipsoid). State why a global max/min must exist. (6)

(c) Now add a second constraint: the particle is further confined to the plane z=0z = 0. Set up and solve the two-constraint Lagrange system U=λg+μh\nabla U = \lambda\nabla g + \mu\nabla h with h=zh = z. Find the constrained extrema. (10)

(d) Write a short pseudocode / Python-style numerical scheme (gradient projection or Newton on the Lagrange system) that would locate the part-(a) extrema numerically, and explain how the second-order condition (bordered Hessian) confirms max vs. min. You need not run it; correctness of the algorithm and the classification criterion is assessed. (8)


Question 3 — Integration, Change of Variables, and Flux (33 marks)

(a) Evaluate by first changing the order of integration:

I=01y1sin(πx3)xdxdy.I = \int_0^1 \int_{\sqrt{y}}^{1} \frac{\sin(\pi x^3)}{x}\, dx\, dy.

Show the region and the reversed limits. Give an exact value. (8)

(b) Using the change of variables u=xyu = x - y, v=x+yv = x + y, evaluate

J=R(xy)2ex+ydA,J = \iint_R (x-y)^2\, e^{x+y}\, dA,

where RR is the square with vertices (0,0),(1,0),(1,1),(0,1)(0,0),(1,0),(1,1),(0,1)... correction: let RR be the region bounded by x+y=0x+y=0, x+y=2x+y=2, xy=0x-y=0, xy=2x-y=2. Compute the Jacobian and evaluate JJ. (9)

(c) Let SS be the closed surface of the solid region EE bounded below by the cone z=x2+y2z=\sqrt{x^2+y^2} and above by the sphere x2+y2+z2=8x^2+y^2+z^2=8, oriented outward. For the field F=(x,y,z)\mathbf{F} = (x, y, z), use the Divergence theorem to compute the outward flux SFdS\iint_S \mathbf{F}\cdot d\mathbf{S}. Use spherical coordinates for the volume integral. (10)

(d) For the same solid EE, find its volume using cylindrical coordinates and confirm consistency: verify that your flux answer equals 3Vol(E)3\,\text{Vol}(E) (since F=3\nabla\cdot\mathbf{F}=3). (6)


End of paper.

Answer keyMark scheme & solutions

Question 1

(a) [6 marks] Let F=(P,Q,R)\mathbf{F}=(P,Q,R) with P=yx2+y2+2xzP=\frac{-y}{x^2+y^2}+2xz, Q=xx2+y2Q=\frac{x}{x^2+y^2}, R=x2+coszR=x^2+\cos z.

Curl components:

  • (×F)x=RyQz=00=0(\nabla\times\mathbf F)_x = R_y - Q_z = 0 - 0 = 0.
  • (×F)y=PzRx=2x2x=0(\nabla\times\mathbf F)_y = P_z - R_x = 2x - 2x = 0.
  • (×F)z=QxPy(\nabla\times\mathbf F)_z = Q_x - P_y. Here Qx=(x2+y2)x(2x)(x2+y2)2=y2x2(x2+y2)2Q_x=\frac{(x^2+y^2)-x(2x)}{(x^2+y^2)^2}=\frac{y^2-x^2}{(x^2+y^2)^2} and Py=(x2+y2)+y(2y)(x2+y2)2=y2x2(x2+y2)2P_y=\frac{-(x^2+y^2)+y(2y)}{(x^2+y^2)^2}=\frac{y^2-x^2}{(x^2+y^2)^2}. So QxPy=0Q_x-P_y=0.

Thus ×F=0\nabla\times\mathbf F=\mathbf 0 (4). Divergence: F=Px+Qy+Rz\nabla\cdot\mathbf F = P_x+Q_y+R_z. Px=2xy(x2+y2)2+2zP_x=\frac{2xy}{(x^2+y^2)^2}+2z, Qy=2xy(x2+y2)2Q_y=\frac{-2xy}{(x^2+y^2)^2}, Rz=sinzR_z=-\sin z. Sum =2zsinz=2z-\sin z (2).

(b) [9 marks] The domain R3{z-axis}\mathbb R^3\setminus\{z\text{-axis}\} is not simply connected (loops around the zz-axis cannot contract to a point) (3). Poincaré's lemma / the "curl-free \Rightarrow conservative" implication requires a simply connected domain; it fails here (2). The vortex part G\mathbf G is the gradient of θ=atan2(y,x)\theta=\operatorname{atan2}(y,x), which is multivalued (jumps by 2π2\pi around a loop). The remaining part FG=(2xz,0,x2+cosz)\mathbf F-\mathbf G=(2xz,\,0,\,x^2+\cos z) is conservative with potential ϕ=x2z+sinz+const\phi = x^2 z + \sin z + \text{const} Check: ϕx=2xz\phi_x=2xz, ϕy=0\phi_y=0, ϕz=x2+cosz\phi_z=x^2+\cos z. ✓ (4)

(c) [7 marks] Parametrize r(t)=(2cost,2sint,1)\mathbf r(t)=(2\cos t, 2\sin t, 1), t[0,2π]t\in[0,2\pi], r(t)=(2sint,2cost,0)\mathbf r'(t)=(-2\sin t, 2\cos t, 0). The gradient part ϕ\nabla\phi contributes 00 around the closed loop (path-independent, closed curve). Only G\mathbf G contributes. On CC: x2+y2=4x^2+y^2=4, so G=(y4,x4,0)=(sint2,cost2,0)\mathbf G=(-\tfrac{y}{4},\tfrac{x}{4},0)=(-\tfrac{\sin t}{2},\tfrac{\cos t}{2},0). Gr=(sint2)(2sint)+(cost2)(2cost)=sin2t+cos2t=1\mathbf G\cdot\mathbf r' = (-\tfrac{\sin t}{2})(-2\sin t)+(\tfrac{\cos t}{2})(2\cos t)=\sin^2 t+\cos^2 t=1. CFdr=02π1dt=2π.\oint_C \mathbf F\cdot d\mathbf r=\int_0^{2\pi}1\,dt=2\pi. (marks: parametrization 2, drop gradient part 2, integrand 2, value 1)

(d) [4 marks] Any surface SS bounded by CC (which encircles the zz-axis) must cross the zz-axis, where F\mathbf F (and ×F\nabla\times\mathbf F) is undefined (2). Stokes' theorem requires F\mathbf F to be C1C^1 on an open set containing SS; this fails because SS contains a singular point (2). Hence Stokes cannot be applied and the direct value 2π02\pi\neq0 stands.

(e) [8 marks] Generalized Stokes: Mω=Mdω\int_{\partial M}\omega=\int_M d\omega.

  • Green (11-form on a plane region): MM = 2D region, M\partial M = its boundary curve, ω=Pdx+Qdy\omega=P\,dx+Q\,dy, dω=(QxPy)dxdyd\omega=(Q_x-P_y)\,dx\wedge dy (2).
  • Stokes (classical): MM = oriented surface in R3\mathbb R^3, M\partial M = boundary curve, ω=Fdr\omega=\mathbf F\cdot d\mathbf r (11-form), dω=(×F)dSd\omega=(\nabla\times\mathbf F)\cdot d\mathbf S (2).
  • Divergence: MM = solid region EE, M\partial M = closed surface, ω=\omega= the 22-form dual to F\mathbf F (flux form), dω=(F)dVd\omega=(\nabla\cdot\mathbf F)\,dV (2). Unifying idea: the exterior derivative dd generalizes grad/curl/div, and orientation of the boundary is induced from MM; each theorem equates a boundary integral to an integral of the derivative over the interior (2).

Question 2

(a) [9 marks] U=(1,2,1)\nabla U=(1,2,1), g=(2x,2y,8z)\nabla g=(2x,2y,8z). Equations: 1=2λx,  2=2λy,  1=8λz1=2\lambda x,\;2=2\lambda y,\;1=8\lambda z x=12λ,y=1λ,z=18λ\Rightarrow x=\frac{1}{2\lambda}, y=\frac{1}{\lambda}, z=\frac{1}{8\lambda} (3). Constraint: x2+y2+4z2=12x^2+y^2+4z^2=12: 14λ2+1λ2+4164λ2=12\frac{1}{4\lambda^2}+\frac{1}{\lambda^2}+4\cdot\frac{1}{64\lambda^2}=12 1λ2(14+1+116)=12\Rightarrow \frac{1}{\lambda^2}\left(\frac14+1+\frac{1}{16}\right)=12. Sum =4+16+116=2116=\frac{4+16+1}{16}=\frac{21}{16}. So 2116λ2=12λ2=21192=764λ=±78\frac{21}{16\lambda^2}=12\Rightarrow \lambda^2=\frac{21}{192}=\frac{7}{64}\Rightarrow\lambda=\pm\frac{\sqrt7}{8} (4). Points: with λ=78\lambda=\frac{\sqrt7}{8}: x=47,y=87,z=17x=\frac{4}{\sqrt7}, y=\frac{8}{\sqrt7}, z=\frac{1}{\sqrt7}; with λ=78\lambda=-\frac{\sqrt7}{8}: negatives thereof (2).

(b) [6 marks] U=x+2y+zU=x+2y+z. At the λ>0\lambda>0 point: U=47+167+17=217=37U=\frac{4}{\sqrt7}+\frac{16}{\sqrt7}+\frac{1}{\sqrt7}=\frac{21}{\sqrt7}=3\sqrt7 (2). At the λ<0\lambda<0 point: U=37U=-3\sqrt7 (1). Global max =377.937=3\sqrt7\approx7.937; global min =37=-3\sqrt7 (2). Existence: UU is continuous and the ellipsoid SS is compact (closed & bounded), so by the Extreme Value Theorem the max and min are attained (1).

(c) [10 marks] Now g=x2+y2+4z212g=x^2+y^2+4z^2-12, h=zh=z. On z=0z=0: gg reduces to circle x2+y2=12x^2+y^2=12. U=λg+μh\nabla U=\lambda\nabla g+\mu\nabla h: (1,2,1)=λ(2x,2y,8z)+μ(0,0,1)(1,2,1)=\lambda(2x,2y,8z)+\mu(0,0,1). Component eqns: 1=2λx,  2=2λy,  1=8λz+μ1=2\lambda x,\;2=2\lambda y,\; 1=8\lambda z+\mu. With z=0z=0: third gives μ=1\mu=1 (free, determined) (3). From first two: x=12λ,y=1λx=\frac{1}{2\lambda}, y=\frac1\lambda. Constraint x2+y2=12x^2+y^2=12: 14λ2+1λ2=1254λ2=12λ2=548λ=±548\frac{1}{4\lambda^2}+\frac1{\lambda^2}=12\Rightarrow\frac{5}{4\lambda^2}=12\Rightarrow\lambda^2=\frac{5}{48}\Rightarrow\lambda=\pm\sqrt{\frac{5}{48}} (4). Points: (12λ,1λ,0)\left(\frac{1}{2\lambda},\frac1\lambda,0\right). Numerically λ=±0.3227\lambda=\pm0.3227, giving (±1.549,±3.098,0)(\pm1.549,\pm3.098,0) (same sign). U=x+2y=12λ+2λ=52λU=x+2y=\frac{1}{2\lambda}+\frac2\lambda=\frac{5}{2\lambda}. Max: λ>0\lambda>0, U=52λ=5215/48=5248/5=60=2157.746U=\frac{5}{2\lambda}=\frac52\cdot\frac{1}{\sqrt{5/48}}=\frac52\sqrt{48/5}=\sqrt{60}=2\sqrt{15}\approx7.746; min =215=-2\sqrt{15} (3).

(d) [8 marks] Algorithm (Newton on the KKT/Lagrange system) — award for correctness:

# unknowns (x,y,z,lam)
F(x,y,z,lam) = [ grad U - lam*grad g ;  g(x,y,z) ]   # 4 eqns
solve F=0 by Newton: 
  guess w0
  repeat: w <- w - J(w)^{-1} F(w)  until ||F||<tol
J = Jacobian of the 4 residuals (includes Hessian of Lagrangian block)

(4) for a coherent scheme (projected-gradient alternative equally valid: project ∇U onto tangent plane of S, step, renormalize onto S). Classification via bordered Hessian: form

Hˉ=(0gTgHL),HL=2(Uλg)=2λI  (diag(2λ,2λ,8λ)).\bar H=\begin{pmatrix}0 & \nabla g^T\\ \nabla g & H_L\end{pmatrix},\quad H_L=\nabla^2(U-\lambda g)=-2\lambda I\;(\text{diag}(-2\lambda,-2\lambda,-8\lambda)).

For a single constraint in n=3n=3 vars, examine the sign of the last two leading principal minors of Hˉ\bar H: an alternating sign pattern signals a constrained max, uniform sign a constrained min (4). Concretely λ>0\lambda>0 makes HLH_L negative definite on the tangent space → local max (U=37U=3\sqrt7); λ<0\lambda<0 → min.


Question 3

(a) [8 marks] Region: 0y10\le y\le1, yx1\sqrt y\le x\le1, i.e. yx2y\le x^2. Reverse: 0x10\le x\le1, 0yx20\le y\le x^2 (3).