MAST20029 Cheatsheet · Engineering Mathematics

University of Melbourne · Cheatsheet for the final exam. Semester 2 examination period, 2–20 November 2026. Full MAST20029 study guide →

A4 landscape · 2 sides · 6 columns each · revision memory sheetMAST20029 · University of Melbourne · S2 2026
MAST20029
Engineering Mathematics
University of Melbourne · S2 2026 · closed-book memory sheet · formula sheet provided
MAST20029 cheatsheet QR code – open the full MAST20029 study guide
Full guide + ask Sia
about MAST20029
SIDE 1/2   DECIDE THE METHOD · CHECK CONDITIONS · INTERPRET THE OUTPUTnothing of your own goes inMemory sheet for revision · Side 1

0 · Exam Mapcourse data · brief

Final 65%MST 15%3×5%Q 5%
Final65% · end of semester · in person · 15 min reading + 3 h writing · 11 questions · 120 marks
OtherMid-semester test 15% · assignments (3 × 5%) 15% · quizzes (best 5 of 6) 5%

“You will be given the 5-page formula sheet below in the final exam:” MAST20029 2026 SM1 Canvas exam-information page; 2026 SM1 exam front cover; 2025 S1 paper — confirm this semester’s rule on the LMS.

What the exam sheet gives you: the named 5-page formula sheet. This memory sheet instead drills the choices it cannot make: method, conditions, orientation, limits, interpretation and justified working.

Exam habit → Final marks reward correct method, full working, accuracy and notation. State the theorem/test; check the condition that licenses it; then execute.

1 · Trigger → Methodall sections

READ THE OBJECTclosed surfaceoutward fluxopen surface+ boundaryx′ = AxeigenvaluesIVP / step/ impulseGAUSSSTOKESEIGENLAPLACEseries: shape → testPDE: Δ=B²−4ACNAME · CONDITION · WORKING · CONCLUSION
Master selector: geometry or algebra supplies the trigger; the named condition supplies the mark.
WordingFirst move
Impossible inner antiderivativeSketch region; change order
Disk / cylinder / ballMatch coordinates + Jacobian
Closed-loop workTest curl F=0 first
Factorial / nth powerRatio / root test
Periodic symmetryOdd→sine; even→cosine
Rectangle + homogeneous edgesSeparate; BCs set modes

Eleven-question retrieval route

FamilyRecall before algebra
Vector calculusobject, geometry, orientation, Jacobian
ODE systemseigenvalue case; phase type; stability
Laplaceinitial terms; shift/impulse/convolution signal
Seriesterm limit first; test from shape; endpoints
Fourierperiod L; parity; value at jumps
PDEclass; homogeneous direction; match edge last

2 · Vector OperatorsCh 1 · Q1–Q3

Input → outputMeaning / check
∇φ → vectorsteepest increase; normal to level sets
∇·F → scalarnet outflow: + source, − sink
∇×F → vectorlocal rotation; determinant signs + − +

instant identitiescurl(grad φ)=0  ·  div(curl F)=0

Conservative test: on a simply connected domain, curl F=0 ⇔ F=∇φ. Recover φ by integrating one component; the “constant” is a function of the other variables, fixed by matching.

Exam bite: keep scalar divergence and three-component curl visibly distinct; the j-component carries the negative cofactor. Use the identities both as explanations and as arithmetic checks.

3 · Region → CoordinatesCh 2–3

REGION SHAPE?POLARCYLINDRICALSPHERICALJ=rJ=rJ=r²sinφlimits first → substitute → multiply Jacobian
Symmetry chooser. Course convention: φ is measured down from +z; θ is azimuth.
SystemSubstitutionBest for
Polarx=r cosθ, y=r sinθdisks, annuli, x²+y²
Cylindricalpolar + z unchangedcylinders / axis symmetry
Sphericalx=r cosθ sinφ; y=r sinθ sinφ; z=r cosφballs, spheres
Geometric phraseTypical limit
first quadrant0≤θ≤π/2
full disk/cylinder0≤θ≤2π
full ball0≤θ≤2π, 0≤φ≤π
radius a0≤r≤a

Change-order routine: draw R; new outer variable takes its full numeric range; new inner limits are the curves met by a strip. Never exchange symbols mechanically. An inner ∫e^(x³)dx is the classic swap signal.

Two setup drills

Quarter-annulus: x,y≥0 and 1≤x²+y²≤9 gives 1≤r≤3, 0≤θ≤π/2. With density 1/(x²+y²), density × Jacobian =1/r, so mass=(π/2)ln3. The answer diagnoses the setup: without J=r the radial integral would be wrong.

Before integratingsketch → system/order → bounds → Jacobian
After integratingrange/sign/units plausible? region covered once?

4 · Curves & SurfacesCh 4

Asked forSet-up
Scalar line∫ f(r(t))|r′(t)|dt
Work∫ F(r(t))·r′(t)dt
Surface scalar∬ g√(fₓ²+fᵧ²+1)dA
Upward flux∬(−F₁fₓ−F₂fᵧ+F₃)dA

Before work: compute curl F. If zero on a simply connected domain, find φ and use ∫CF·dr=φ(end)−φ(start). A closed loop then gives 0 without edge-by-edge integration.

Orientation: increasing t directs r(t); reversing the path reverses work. “Downward” reverses an upward-flux sign.

ObjectParametrise / projectCheck
circle radius ar(t)=(a cos t,a sin t,0)t increasing is anticlockwise from +z
segment A→Br(t)=tB+(1−t)A, 0≤t≤1r(0)=A, r(1)=B
graph z=f(x,y)project to R in xy-planeuse the correct scalar or flux factor

Do not merge the two surface formulas. The square-root factor belongs to scalar accumulation/area; the vector normal factor belongs to flux. Ask whether the integrand is a scalar g or a field F before choosing.

5 · GaussCh 5 · flux trigger

Vn̂ outcaps are part of S
Source redraw, Ch 5: Gauss needs the whole closed boundary and outward normal.

licence + conversionclosed S=∂V, outward n̂ ⇒ ∬SF·n̂ dS = ∭V∇·F dV

Use Gauss when divergence over the solid is easier than summing face fluxes. For an open side, add the cap, apply Gauss, then subtract cap flux. Choose coordinates from V and carry the Jacobian.

Check → constant divergence collapses to constant × volume.

Closed-surface micro-drill

For F=(x,y,z) on the sphere x²+y²+z²=a², ∇·F=3. Gauss gives flux=∭V3dV=3·(4/3)πa³=4πa³. This is not “formula recall”: the marks lie in recognising closed + outward and converting to constant × volume.

Given surfaceLegal route
whole sphere / closed cylinderGauss immediately

6 · StokesCh 6 · circulation

S openCn̂right-hand rule: fingers C, thumb n̂
Source redraw, Ch 6: boundary direction and normal are one compatible orientation.

licence + conversionopen oriented S with boundary C ⇒ ∬S(∇×F)·n̂ dS = ∮CF·dr

Choose the easier side. Simple curl → surface flux; simple boundary → circulation. Replace an awkward cap by any easier surface sharing C. Reversing C or n̂ flips the sign.

FeatureGaussStokes
Surfaceclosedopen
Operatordivcurl
Other sidevolumeboundary
Orientationoutwardright-hand

Choice test

Given / requestedCompare
circulation around simple circledirect parametrisation versus curl over flat disk
curl flux through awkward capsame boundary, replace by flat surface
conservative F around loopquote zero; no integral needed
surface without boundarynot Stokes; inspect Gauss if closed

Sign sentence: “The upward normal pairs with anticlockwise traversal as viewed from above.” If the problem states the opposite traversal, either reverse C or keep n̂ and attach a minus sign—never silently mix conventions.

7 · Guardrailsshort-answer marks

  • curl F=0 needs a simply connected domain.
  • Gauss needs closure; Stokes needs C=∂S.
  • Keep the same boundary when changing a Stokes surface.
  • The Jacobian belongs to dA/dV after coordinate change.
  • Name the theorem before substitution.

curl(grad φ)=0 · div(curl F)=0

7b · Stokes Micro-DrillCh 6 worked

Surface independencesame C + compatible orientation → same curl flux
Sign checkviewed from n̂, positive C is anticlockwise

8 · Systems x′=AxCh 7 · Q4

Pipeline: det(A−λI)=0 → eigenvalues → eigenvectors → correct solution form → initial conditions. Eigenvector directions are straight-line orbits; sign(Re λ) controls in/out behaviour.

EigenvaluesSolution move
distinct realα₁w₁e^(λ₁t)+α₂w₂e^(λ₂t)
repeated, two wtwo independent eigenmodes
deficient(wt+u)e^(λt), (A−λI)u=w
a±biRe/Im of we^(λt); no i remains

A centre requires pure imaginary eigenvalues: real part exactly zero. Asymptotic stability instead needs negative real parts.

9 · Classify & SketchCh 8 · Q5

EIGENVALUES OF J / Arealopposite signsrealsame signcomplex a±biSADDLEunstableNODE− stable+ unstablea=0a≠0CENTRESPIRALrepeated λ: check geom. mult.
Classification is immediate; the marks move to eigen-lines, arrows, slopes and long-time behaviour.
  1. Name type + stability.
  2. Draw eigenvector lines; λ>0 out, λ<0 in.
  3. Add trajectories and t→±∞ asymptotes.
  4. Use dy/dx=ẏ/ẋ for axis slopes.
  5. Non-linear: solve critical points, evaluate J at each, classify locally. Borderline centres may be inconclusive.
SpectrumTypeFlow
λ₁,λ₂<0 realstable nodeboth modes in
λ₁,λ₂>0 realunstable nodeboth modes out
λ₁λ₂<0saddleone in, one out
a±bi, a<0stable spiralspirals in
a±bi, a>0unstable spiralspirals out
±bicentreclosed orbits

Linearisation frame: solve f(x,y)=g(x,y)=0; form J=[[fₓ,fᵧ],[gₓ,gᵧ]]; substitute each critical point; classify its eigenvalues separately. For spirals/centres, determine rotation from one test velocity vector rather than guessing.

Eigenvector geometry: along w, Aw=λw remains on the same line, so the orbit can be straight. A generic initial vector combines modes and bends between eigen-directions. For a saddle, the negative-λ line is the stable direction and the positive-λ line the unstable direction.

10 · Laplace RecognitionCh 9 · Q6

Pipeline: transform IVP with initial terms → solve X(s) → partial fractions → invert → name each shift/convolution theorem.

SignalDecision
s replaced by s+as-shift
e^(−as)t-shift; delayed u(t−a)
δ(t−a)transform e^(−as)
∫₀ᵗf(τ)g(t−τ)dτconvolution → F(s)G(s)
piecewise forcingjumps × unit steps
2513+3−4f=2+3u(t−1)−4u(t−3)
Source redraw, Ch 9: each step coefficient is the jump, not the new height.

derivativesL{x′}=sX−x(0)
L{x″}=s²X−sx(0)−x′(0)

11 · Series & FourierCh 10–11

ShapeChoose
aₙ↛0divergence: diverges
bounded / nᵖcomparison with p-series
n!, cⁿratio; L=1 silent
(...)ⁿroot; L=1 silent
(−1)ⁿbₙLeibniz, then absolute

Power series: find R, then test both endpoints. Fourier: even→cosine; odd→sine. At a jump use ½[f(t⁻)+f(t⁺)].

benchmarksΣarⁿ converges iff |r|<1; sum=a/(1−r)
Σ1/nᵖ converges iff p>1

Taylor move: identify the standard series, match the centre, then substitute the requested value; for approximation, the polynomial matches f and its derivatives at the centre and the remainder controls the gap.

12 · PDE DecisionCh 12 · Q11

ELLIPTICPARABOLICHYPERBOLICB²−4AC<0=0>0Laplacediffusionwavesteady stateheat spreadstravels at cclassify → separate → modes → match
Source redraw, Ch 12: the discriminant names the physics; boundary conditions select modes.
Separation stageWhat to show
Productu=XY (or XT for time)
asksia.ai/cheatsheet/unimelb-mast20029 · Full MAST20029 study guide →
Forward this to your tutorial group · Shared by ________
AskSiav2 · 2026-10-25
Semester 2 examination period, 2–20 November 2026.
Your exam is closed book with a formula sheet provided: nothing of your own goes in. Use this to revise, not to carry.
MAST20029
Engineering Mathematics · Self-test Drills
University of Melbourne · S2 2026 · closed-book memory sheet · formula sheet provided
MAST20029 cheatsheet QR code – open the full MAST20029 study guide
Full guide + ask Sia
about MAST20029
SIDE 2/2   WHICH-IS-TRUE TRAPS · SHORT-ANSWER FRAMES · WORKED APPLICATIONSrevise · retrieve · justifyMemory sheet for revision · Side 2

13 · Integral JudgementCh 2–6

Which is true?Correct contrast
Swap by exchanging dx,dy?No: sketch R; re-read strips.
Jacobian optional?No: it scales dA/dV.
curl F=0 always conservative?Only on simply connected domain.
Gauss on curved wall alone?Close with caps, then correct.
Stokes surface fixed?No; boundary fixed, surface may change.
vertical: dy dxhorizontal: dx dyy=x²same R
Source redraw, Ch 2: changing order re-describes the same region; outer limits are numeric extremes.

14 · Worked A · Change Order5 marks

Evaluate ∫₀¹∫√y¹ e^(x³) dx dy.

  1. 0≤y≤1, √y≤x≤1 ⇒ y≤x².
  2. Same R: 0≤x≤1, 0≤y≤x².
  3. Swap: ∫₀¹∫₀ˣ²e^(x³)dy dx = ∫₀¹e^(x³)x²dx.
  4. u=x³, du=3x²dx ⇒ (e−1)/3.

The picture licenses the swap and exposes the substitution.

15 · Worked B · Gauss8 marks

F=(x³,y³,z³); closed cylinder x²+y²≤4, 0≤z≤2.

∇·F=3(x²+y²+z²)
dV=r dr dθ dz

Flux=3∫₀²π∫₀²∫₀²(r²+z²)r dz dr dθ. Inner z: 2r²+8/3; then ∫₀²(2r³+(8/3)r)dr=40/3. Restore 3 and 2π: 80π.

Trap → dropping the Jacobian r changes the problem.

15b · Conservative DrillCh 4 worked

F=(2xy+z²,x²,2xz). Mixed partials match, so curl F=0 on the stated simply connected domain. Integrate φₓ=2xy+z²:

Direct pathneeded when curl F≠0 or domain condition fails
Potential shortcutendpoint difference; closed loop gives zero

15c · Polar DrillCh 3 worked

Region cueannulus + first quadrant
Algebra cuex²+y² becomes r²
Essential factorJ=r turns 1/r² into 1/r

16 · ODE Which-Is-TrueCh 7–8

ClaimVerdict / repair
Opposite-sign λ is stable node.False: saddle; unstable.
Pure imaginary λ gives spiral.False: centre.
Repeated λ always has two w.False: check geometric multiplicity.
λ<0 arrows point out.False: e^(λt) decays; point in.
A sketch alone is enough.False: justify lines, slopes, limits.
y=x, λ=4 OUTy=−x, λ=−2 IN
Source redraw, Ch 8 worked saddle: λ=4 along y=x; λ=−2 along y=−x.

17 · Worked C · Saddle7 marks

x′=x+3y, y′=3x+y.

  1. λ=1±3 ⇒ 4,−2 ⇒ unstable saddle.
  2. w₁=(1,1): y=x, outgoing. w₂=(1,−1): y=−x, incoming.
  3. t→+∞: generic orbits follow y=x; t→−∞: y=−x.
  4. dy/dx=(3x+y)/(x+3y): y-axis slope 1/3; x-axis slope 3.

Frame: “Eigenvalues __ hence __; eigen-lines __ with arrows __; slopes __; therefore t→±∞ __.”

18 · Centre DrillCh 7 worked

x′=ax−by, y′=x+ay gives λ=a±√(−b). Centre: a=0,b>0. If a=0,b=4, λ=±2i and w=(2,−i):

x(t)=α₁(2cos2t,sin2t)+α₂(2sin2t,−cos2t)

No i remains; closed orbits are neutral, not asymptotically decaying.

18b · Sketch Checklistjustify every feature

FeatureEvidence to write
typeeigenvalue signs / real parts
straight orbitseigenvector lines
arrow directiongrowth/decay of e^(λt)
axis crossingsdy/dx=ẏ/ẋ at x=0 or y=0
rotationevaluate one velocity vector
non-linear pointJacobian evaluated at that point

Repeated-eigenvalue trap: two independent eigenvectors give the ordinary two-mode form; a deficient matrix has one eigenvector and needs a generalised eigenvector u satisfying (A−λI)u=w, producing the t e^(λt) behaviour.

19 · Laplace JudgementCh 9

Which is true?Answer
e^(−2s) changes s to s+2.False: delay in t.
δ(t−a)→e^(−as).True.
Step coefficient is new level.False: it is the jump.
Convolution becomes a sum.False: product F(s)G(s).
Initial terms can be added later.False: include immediately.
aau(t−a)δ(t−a)both ↦ e^(−as)
Source redraw, Ch 9: a delayed step persists; a Dirac delta is an idealised unit impulse.

20 · Worked D · Impulse IVP6 marks

x″+3x′=δ(t−2), x(0)=1, x′(0)=0.

  1. (s²X−s)+3(sX−1)=e^(−2s).
  2. X=1/s+e^(−2s)/[s(s+3)].
  3. 1/[s(s+3)]=(1/3)[1/s−1/(s+3)].
  4. By t-shifting, delayed inverse is (1/3)[1−e^(−3(t−2))]u(t−2).

answerx(t)=1+(1/3)[1−e^(−3(t−2))]u(t−2)

Before t=2, u=0 and x=1; the exponential contains (t−2).

21 · Answer Framesmethod marks

Shift: “X=e^(−as)F; by t-shifting, x=f(t−a)u(t−a).”

Convolution: “Integral=(y∗g)(t), hence transform YG; solve for Y.”

Piecewise: “Initial level + Σ(new−old)u(t−a).”

21b · Transform Auditbefore inversion

Audit questionFailure it catches
Did every derivative include initial data?lost constants in X(s)
Is the rational part proper?partial fractions before division
What base pair is on the table?wrong sine/cosine/exponential scale
Is e^(−as) factored outside?delay applied to only part of f
Did I name the theorem?missing explicit method mark

Step-function reconstruction

s-shiftmultiplication by e^(−at) shifts F(s) to F(s+a)
t-shiftfactor e^(−as) delays the whole base inverse and adds u(t−a)

22 · Convergence SelectorCh 10 · Q7–8

READ THE SHAPE OF aₙFIRST: if aₙ↛0, divergence test ⇒ divergesbounded/nᵖCOMPAREn!, cⁿRATIOwhole (...)ⁿROOT(−1)ⁿbₙLEIBNIZL<1 conv · L>1 divL=1 silentLeibniz: bₙ↓0then test Σ|aₙ|POWER SERIES: R, THEN BOTH ENDPOINTS
Source redraw, Ch 10: the divergence test comes first; a zero term limit proves nothing.
TrapRepair
L=1choose another test
alternating convergesmay be conditional; test absolute
radius foundstill test endpoints
sin n numeratorbound it; do not evaluate

23 · Application · Comparison + R6 marks

(a) Σ(2+sin n)/(n²+1):

0≤(2+sin n)/(n²+1)≤3/n²

Σ3/n² converges (p=2>1), so the original converges by comparison.

(b) Σn!(x−3)ⁿ/nⁿ:

L=(n/(n+1))ⁿ|x−3| → |x−3|/e

L<1 ⇒ |x−3|<e, so R=e. If the interval is requested, test both endpoints.

Say it → Name the test and quote the inequality/limit that licenses the conclusion.

23b · Fast Recall Tableconditions matter

TestRequired shapeConclusion
comparisonnon-negative terms; useful upper/lower benchmarkdirection must support claim
limit comparisonaₙ,bₙ>0 and aₙ/bₙ→c∈(0,∞)same convergence fate
integralpositive, continuous, decreasingΣ and improper integral agree
Leibnizbₙ decreases to 0alternating series converges
ratio/rootabsolute-value limit existsL<1 conv; L>1 div

Taylor recall

Approximation questionwrite polynomial + remainder/error statement
Numerical-sum questionname standard series + substitution value
Power seriesradius first; interval needs endpoints

24 · Fourier JudgementCh 11 · Q9–10

ODD → SINEEVEN → COSINEjump → 0no jump
Source redraw, Ch 11: reflect through the origin for odd; across the vertical axis for even; repeat with period 2L.
Which is true?Correct rule
Odd extension has cosine.False: bₙ sine only.
Even extension has sine.False: a₀,aₙ cosine only.
At a jump series equals f.False: one-sided midpoint.
Non-periodic uses series.Use Fourier integral.

Decide symmetry first. On (0,L), use factor 2/L; keep ω=π/L consistent; show integration-by-parts boundary terms.

25 · Application · Sine Series6 marks

f(t)=t, 0<t<2, odd extension: L=2, ω=π/2.

bₙ=∫₀²t sin(nπt/2)dt
=4(−1)^(n+1)/(nπ)

seriesf(t)=Σn≥1[4(−1)^(n+1)/(nπ)]sin(nπt/2)

At t=2 the extension jumps; the series tends to 0, the midpoint.

26 · Answer Framesconvergence

Symmetry: “The extension is odd/even, so __ coefficients vanish.”

Jump: “By midpoint convergence, limit=½[f(t⁻)+f(t⁺)]=__.”

Integral: “Non-periodic; use continuous frequency, cosine-only if even or sine-only if odd.”

26b · Fourier Fast RecallCh 11

PromptJudgement before integration
f given on (0,L)choose odd/even periodic extension
odd extensiona₀=aₙ=0; compute bₙ over (0,L)
even extensionbₙ=0; compute a₀,aₙ over (0,L)
jump pointreport midpoint of one-sided limits
periodic forcingexpand forcing; solve harmonics term-by-term
non-periodic signalFourier cosine/sine integral by parity

Gibbs distinction: partial sums develop persistent overshoot near a jump, while at the jump itself the series converges exactly to the midpoint. Do not confuse local overshoot with the pointwise limit.

Coefficient audit

27 · PDE RectangleCh 12 · Q11

u(x,1)=f(x) · MATCH LASTinhomogeneous edgeu(0,y)=0u(1,y)=0u(x,0)=0uₓₓ+uᵧᵧ=0
Source redraw, Ch 12: three homogeneous edges select modes; the remaining edge fixes coefficients.
StageMark-bearing line
Separateu=XY ⇒ X″/X=−Y″/Y=−λ
BC problemX(0)=X(1)=0 ⇒ λₙ=(nπ)², Xₙ=sin(nπx)
Other factorYₙ=sinh(nπy), vanishes at y=0
Superposeu=Σcₙsin(nπx)sinh(nπy)
Matchimpose u(x,1)=f(x) last

28 · Application · Laplace Square8 marks

If u(x,1)=sin(2πx)+3sin(5πx), matching gives c₂sinh(2π)=1, c₅sinh(5π)=3; all other cₙ=0.

solutionu=[sin(2πx)sinh(2πy)]/sinh(2π)
+3[sin(5πx)sinh(5πy)]/sinh(5π)

Do not integrate coefficients when data is already a finite sine sum. Check all four boundary conditions.

28b · PDE Trap Tablelong-question defence

TrapCorrection
classify by summing coefficientsuse B²−4AC
write only one λ caseshow why other signs are trivial
choose cosh for Y(0)=0choose sinh, which vanishes
fit f(x) too earlyhomogeneous BCs select modes first
integrate a finite sine summatch coefficients directly
omit final checksubstitute every boundary edge

29 · Exam-Morning Planrevision only

1234TRIGGERSCONDITIONSONE DRILLSTOPmethodselectortheoremstests + BCsweakestsectionexact formscalm setup
Final retrieval loop: choices and conditions, not a fresh topic.
    asksia.ai/cheatsheet/unimelb-mast20029 · Full MAST20029 study guide →
    Forward this to your tutorial group · Shared by ________
    AskSiav2 · 2026-10-25
    Semester 2 examination period, 2–20 November 2026.
    Your exam is closed book with a formula sheet provided: nothing of your own goes in. Use this to revise, not to carry.

    Want one for YOUR exact syllabus?

    Sia is your free desktop study agent. Drop your University of Melbourne MAST20029 slides — Sia builds a sheet tailored to YOUR exam. Better than this library because it knows YOUR materials.

    ↓ Download Sia · Free