Multivariable Calculus
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 -style math throughout.
Question 1 — Fields, Potentials, and the Unification of Integral Theorems (34 marks)
Consider the vector field on
(a) Compute and . Show all component derivatives. (6)
(b) The field 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 where , and find the scalar potential for the remaining part. (9)
(c) Let be the circle , , oriented counterclockwise viewed from above. Compute directly by parametrization. (7)
(d) A student claims: "By Stokes' theorem, for any surface bounded by ." 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 . Identify what plays the role of , , and in each. (8)
Question 2 — Constrained Optimization meets Physics: Equilibrium of a Charged Particle (33 marks)
A particle constrained to move on the surface experiences a potential energy
Equilibria correspond to critical points of on .
(a) Use Lagrange multipliers to find all points on where (with ). Give all critical points and the corresponding . (9)
(b) Evaluate at each critical point and identify the global minimum and maximum of on (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 . Set up and solve the two-constraint Lagrange system with . 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:
Show the region and the reversed limits. Give an exact value. (8)
(b) Using the change of variables , , evaluate
where is the square with vertices ... correction: let be the region bounded by , , , . Compute the Jacobian and evaluate . (9)
(c) Let be the closed surface of the solid region bounded below by the cone and above by the sphere , oriented outward. For the field , use the Divergence theorem to compute the outward flux . Use spherical coordinates for the volume integral. (10)
(d) For the same solid , find its volume using cylindrical coordinates and confirm consistency: verify that your flux answer equals (since ). (6)
End of paper.
Answer keyMark scheme & solutions
Question 1
(a) [6 marks] Let with , , .
Curl components:
- .
- .
- . Here and . So .
Thus (4). Divergence: . , , . Sum (2).
(b) [9 marks] The domain is not simply connected (loops around the -axis cannot contract to a point) (3). Poincaré's lemma / the "curl-free conservative" implication requires a simply connected domain; it fails here (2). The vortex part is the gradient of , which is multivalued (jumps by around a loop). The remaining part is conservative with potential Check: , , . ✓ (4)
(c) [7 marks] Parametrize , , . The gradient part contributes around the closed loop (path-independent, closed curve). Only contributes. On : , so . . (marks: parametrization 2, drop gradient part 2, integrand 2, value 1)
(d) [4 marks] Any surface bounded by (which encircles the -axis) must cross the -axis, where (and ) is undefined (2). Stokes' theorem requires to be on an open set containing ; this fails because contains a singular point (2). Hence Stokes cannot be applied and the direct value stands.
(e) [8 marks] Generalized Stokes: .
- Green (-form on a plane region): = 2D region, = its boundary curve, , (2).
- Stokes (classical): = oriented surface in , = boundary curve, (-form), (2).
- Divergence: = solid region , = closed surface, the -form dual to (flux form), (2). Unifying idea: the exterior derivative generalizes grad/curl/div, and orientation of the boundary is induced from ; each theorem equates a boundary integral to an integral of the derivative over the interior (2).
Question 2
(a) [9 marks] , . Equations: (3). Constraint: : . Sum . So (4). Points: with : ; with : negatives thereof (2).
(b) [6 marks] . At the point: (2). At the point: (1). Global max ; global min (2). Existence: is continuous and the ellipsoid is compact (closed & bounded), so by the Extreme Value Theorem the max and min are attained (1).
(c) [10 marks] Now , . On : reduces to circle . : . Component eqns: . With : third gives (free, determined) (3). From first two: . Constraint : (4). Points: . Numerically , giving (same sign). . Max: , ; min (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
For a single constraint in vars, examine the sign of the last two leading principal minors of : an alternating sign pattern signals a constrained max, uniform sign a constrained min (4). Concretely makes negative definite on the tangent space → local max (); → min.
Question 3
(a) [8 marks] Region: , , i.e. . Reverse: , (3).