Lent 2017 VECTOR CALCULUS – EXAMPLES 1 G.Taylor

A star means ‘optional’ and not necessarily ‘harder’.

1. The curve given parametrically by (a cos3 t, a sin3 t) with 0 6 t 6 2π is called an astroid.Sketch it, and find its length.

2. The curve defined by y2 = x3 is called Neile’s parabola. Sketch the segment of Neile’s parabolawith 0 6 x 6 4, and find the length of this segment.

*3. In R2 a path is defined in polar coordinates by r = f(θ), α 6 θ 6 β, with a function

f ∈ C1[α, β]. Show that the length of the path is

L =

∫ β

α

√

(

f(θ))2

+(

f ′(θ))2

dθ.

The curves r = aθ, r = aebθ and r = a(1+cos θ) are called an Archimedean spiral, a logarithmicspiral and a cardioid, respectively. For a, b > 0 and 0 6 θ 6 2π, sketch the curves and findtheir lengths.

4. Given a function f(x) in two dimensions, use the chain rule to express its partial derivativeswith respect to Cartesian coordinates (x, y) in terms of its partial derivatives with respect topolar coordinates (ρ, φ). From the relationship between the basis vectors in these coordinatesystems, deduce that

∇f =∂f

∂ρeρ +

1

ρ

∂f

∂φeφ.

5. In three dimensions, use suffix notation and the summation convention to show that

(i) ∇(a · x) = a and (ii) ∇rn = nrn−2x,

where a is a constant vector and r = |x|.Obtain the same results using spherical polar coordinates.

Note/hint: in spherical polars, for a function f of r and θ only, ∇f = ∂f∂r er +

1

r∂f∂θ eθ.

6. Find the equation of the plane which is tangent to the surface z = 3x2y sin(π2x) at the point

x = y = 1.

Taking East to be in the direction (1, 0, 0) and North to be (0, 1, 0), in which direction will amarble roll if placed on the surface at x = 1, y = 1

2?

7. Evaluate explicitly each of the line integrals

(a)

∫

(x dx+ y dy + z dz), (b)

∫

(y dx+ x dy + dz), (c)

∫

(y dx− x dy + ex+y dz)

along (i) the straight line path joining (0, 0, 0) to (1, 1, 1), and (ii) the parabolic path given byx(t) = (t, t, t2) with 0 6 t 6 1.

For which of these integrals do the two paths give the same results, and why?

8. Let F = (3x2y2z, 2x3yz, x3y2) and G = (3x2yz2, 2x3yz, x3z2) be vector fields.

Show that F is a conservative field and that G is not. Find a scalar potential for F.

By exploiting similarities between F and G, or otherwise, evaluate the line integral∫

G · dxfrom (0, 0, 0) to (1, 1, 1) along the path x(t) = (t, sin π

2t, sin π

2t) with 0 6 t 6 1.

1

9. (i) A curve C is given parametrically in Cartesian coordinates by

x(t) =(

cos(sinnt) cos t, cos(sinnt) sin t, sin(sinnt))

, 0 6 t 6 2π,

where n is some fixed integer. Using spherical polar coordinates, sketch C.

(ii) Let F(x) =(

− yx2+y2

, xx2+y2

, 0)

. Show explicitly that∫

C F ·dx = 2π, where C is traversed

in the direction of increasing t.

(iii) Find a scalar function f such that ∇f = F. Comment on this, given (ii).

*10. Without changing the order of integration, show that

∫

1

0

[∫

1

0

x− y

(x+ y)3dy

]

dx =1

2, and

∫

1

0

[∫

1

0

x− y

(x+ y)3dx

]

dy = −1

2.

Comment on these results.

11. Use the substitution x = r cos θ, y = 1

2r sin θ to evaluate

∫

A

x2

x2 + 4y2dA,

where A is the region between the two ellipses x2 + 4y2 = 1, x2 + 4y2 = 4.

12. The closed curve C in the (x, y) plane consists of the arc of the parabola y2 = 4ax (a > 0)between the points (a,±2a) and the straight line joining (a,∓2a). The area enclosed by C is A.By calculating the integrals explicitly, show that

∫

C(x2y dx+ xy2 dy) =

∫

A(y2 − x2) dA =

104

105a4,

where C is traversed anticlockwise.

13. The region A is bounded by the straight line segments x = 0, 0 6 y 6 1, y = 0, 0 6 x 6 1,y = 1, 0 6 x 6

3

4, and by an arc of the parabola y2 = 4(1−x). Consider a mapping into the

(x, y) plane from the (u, v) plane defined by the transformation x = u2 − v2, y = 2uv. SketchA and also the two regions in the (u, v) plane which are mapped into it.

Hence calculate

∫

A

dA

(x2 + y2)1/2.

*14. (a) The curve x3 + y3 − 3axy = 0 with a > 0 is called the Cartesian Leaf. Using polar coor-dinates, sketch the Cartesian Leaf and find the area bounded by it in the first quadrant.

(b) A tricylinder is the body formed by intersecting the three solid cylinders given by theequations x2 + y2 6 a2, y2 + z2 6 a2 and z2 + x2 6 a2, with a > 0. Using cylindricalpolar coordinates, show that the volume of a tricylinder is 8(2−

√2)a3.

15. Let T be the tetrahedron with vertices at (0, 0, 0), (1, 0, 0), (0, 1, 0) and (0, 0, 1). Find thevolume V of T , and also the centre of volume, given by

1

V

∫

Tx dV .

2

16. A solid cone is bounded by the surface θ = α (in spherical polar coordinates) and the surfacez = a. Its mass density is ρ0 cos θ. Show that its mass is 2

3πρ0a

3(secα− 1).

You can use spherical or cylindrical polars for the calculation, but there are coordinates whichmake the limits of the integrals nicer.

*17. [Tripos, 2005/III/12]

Express the integral

I =

∫

∞

0

dx

∫

1

0

dy

∫ x

0

dz x e−Ax/y−Bxy−Cyz

in terms of the new variables α = x/y, β = xy, γ = yz. Hence show that

I =1

2A(A+B)(A+B + C).

Assume that A, B and C are positive.

3

Lent 2017 VECTOR CALCULUS – EXAMPLES 2 G.Taylor

A star means ‘optional’ and not necessarily ‘harder’.

1. A circular helix is given by x = (a cos t, a sin t, ct). Calculate the tangent t, principal normal n,curvature κ, binormal b, and torsion τ . Sketch the helix for a, c > 0, showing t, n, b at somepoint on the helix.

2. (a) Show that a curve in the plane given by x(t) = (x(t), y(t)) has curvature

κ(t) =|xy − xy|

(x2 + y2)3/2.

(b) Find the maximum and minimum curvature of the ellipsex2

a2+y2

b2= 1, for a > b > 0.

*(c) A curve in the plane is given in polar coordinates as r = f(θ). Find an expression for itscurvature as a function of θ.

3. (a) Let ψ(x) be a scalar field and v(x) a vector field. Using suffix notation, show that

∇ · (ψv) = (∇ψ) · v + ψ∇ ·v and ∇× (ψv) = (∇ψ)× v + ψ∇×v.

(b) Evaluate the divergence and curl of the following vector fields on R3 :

rx, a(b · x), a× x,x− a

|x− a|3 ,

where r = |x|, and a and b are fixed vectors.

Use part (a) where you can. Thinking about the definitions of divergence and curl, couldyou have guessed any of the answers in advance?

4. Let u and v be vector fields. Show, using suffix notation, that

(i) ∇ · (u× v) = v · (∇×u)− u · (∇×v)

(ii) ∇× (u× v) = u(∇ ·v) + (v ·∇)u− v(∇ ·u)− (u ·∇)v

(iii) ∇(u · v) = u× (∇×v) + (u ·∇)v + v × (∇×u) + (v ·∇)u

Deduce from (iii) that (u ·∇)u = ∇(12u2)− u× (∇×u).

Feel free to omit any of (i), (ii), (iii) that were proved in lectures.

5. Show that the vector field

H =(

3x2 tan z − y2e−xy2 sin y, (cos y − 2xy sin y)e−xy2 , x3 sec2 z)

is conservative. Find the most general scalar potential for H, and hence calculate the lineintegral

∫

H · dx from the point (0, 0, 0) to the point (1, π/2, π/4).

6. Verify that the vector field

u = ex(x cos y + cos y − y sin y)i+ ex(−x sin y − sin y − y cos y)j

is irrotational and express it as the gradient of a scalar field ϕ.

Verify also that u is also solenoidal and find a vector potential for it in the form ψk, for somefunction ψ.

1

7. (a) A vector field B(x) is parallel to the normals of a family of surfaces f(x) = constant.Show that B · (∇×B) = 0.

(b) The vector fields v(x) and B(x) are everywhere parallel and are both solenoidal. Showthat B ·∇(v/B) = 0, where v = |v| and B = |B| 6= 0.

*(c) The tangent vector at each point on a curve is parallel to a non-vanishing vector fieldB(x). Show that the curvature of the curve is given by |B|−3|B× (B ·∇)B|.

8. A fluid flow has the velocity vector v = (0, 0, z + a) in Cartesian coordinates, where a is aconstant. Calculate the volume flux of fluid flowing across the open hemispherical surfacer = a, z > 0, and also that flowing across the disc r 6 a, z = 0. Verify that the divergencetheorem holds.

Recall that ‘volume flux of fluid’ =∫

S v · dS.

9. [Tripos, 2002/III/4]

State the divergence theorem. Consider the integral I =∫

S rnr · dS, where n > 0 and S is the

sphere of radius R centred at the origin. Evaluate I directly, and by means of the divergencetheorem.

10. Let F(x) = (x3 + 3y + z2, y3, x2 + y2 + 3z2), and let S be the open surface

x2 + y2 = 1− z, 0 6 z 6 1.

Use the divergence theorem (and cylindrical polar coordinates) to evaluate∫

S F · dS.You should find that dS = (2ρ cosφ, 2ρ sinφ, 1)ρ dρ dφ.

*Verify your result by calculating the integral directly.

11. Verify Stokes’ theorem for the vector field F(x) = (y,−x, z) and the open surface r = 1, z > 0.

12. By applying the divergence theorem to the vector field c×B, where c is an arbitrary constantvector and B(x) is a vector field, show that

∫

V∇×B dV = −

∫

AB× dA,

where the surface A encloses the volume V .

Verify this result when B(x) = (z, 0, 0) and V is the cuboid 0 6 x 6 a, 0 6 y 6 b, 0 6 z 6 c.

13. By applying Stokes’ theorem to the vector field c×B, where c is an arbitrary constant vectorand B(x) is a vector field, show that

∮

Cdx×B =

∫

A(dA×∇)×B,

where the curve C bounds the open surface A.

Verify this result when B(x) = x and A is the unit square in the (x, y) plane with oppositevertices at (0, 0, 0) and (1, 1, 0).

14. [Tripos, 2005/III/10]

Write down Stokes’ theorem for a vector field B(x) on R3.

Consider the bounded surface S defined by z = x2 + y2, 1

46 z 6 1. Sketch the surface

and calculate the surface element dS. For the vector field B = (−y3, x3, z3), calculateI =

∫

S(∇×B) · dS directly.

Show using Stokes’ theorem that I may be rewritten as a line integral and verify this yieldsthe same result.

2

*15. [Tripos, 2001/III/11]

State the divergence theorem for a vector field u(r) in a closed region V bounded by a smoothsurface S.

Let Ω(r) be a scalar field. By choosing u = cΩ for arbitrary constant vector c, show that

∫

V∇Ω dV =

∫

SΩ dS. (∗)

Let V be the bounded region enclosed by the surface S which consists of the cone (x, y, z) =(r cos θ, r sin θ, r/

√3) with 0 6 r 6

√3 and the plane z = 1, where r, θ, z are cylindrical polar

coordinates. Verify that (∗) holds for the scalar field Ω = (a− z), where a is a constant.

*16. [Tripos, 2002/III/11]

The first part of this question was to prove question 4(ii) above, so I have omitted it here.

S is an open orientable surface in R3 with unit normal n, and v(x) is any continuously

differentiable vector field such that n ·v = 0 on S. Let m be a continuously differentiable unitvector field which coincides with n on S. By applying Stokes’ theorem to m× v, show that

∫

S(δij − ninj)

∂vi∂xj

dS =

∮

Cu · v ds,

where s denotes arc-length along the boundary C of S, and u is such that u ds = ds× n.

Verify this result by taking v = r and S to be the disc |r| 6 R in the z = 0 plane.

*17. (a) By considering

A(x) = −∫

1

0

tx×B(tx) dt,

show that if ∇ ·B = 0 then ∇×A = B.

(b) By considering

ϕ(x) =

∫

1

0

x ·B(tx) dt,

show that if x× (∇×B) = 0 then ∇ϕ = B.

Show that B(x) = r−2c × x, where c is a non-zero constant vector, satisfies ∇×B 6= 0and x × (∇×B) = 0. Why does this not contradict the usual necessary and sufficientcondition for B to be the gradient of a scalar?

3

Lent 2017 VECTOR CALCULUS – EXAMPLES 3 G.Taylor

A star means ‘optional’ and not necessarily ‘harder’.

1. (a) Using the operator ∇ in Cartesian coordinates and in spherical polar coordinates, calcu-late the gradient of ψ = Ez = Er cos θ in both systems, where E is a constant.

By considering the relationship between the basis vectors, check that your answers agree.

(b) Calculate, in three ways, the curl of the vector field

B = −yex + xey = ρeφ = r sin θ eφ,

by applying the standard formulae in Cartesian, cylindrical, and spherical coordinates.

2. Show that the unit basis vectors of cylindrical polar coordinates satisfy

∂eρ∂φ

= eφ and∂eφ∂φ

= −eρ

and that all other derivatives of the three basis vectors are zero.

Given that the operator ∇ in cylindrical polars is

∇ = eρ∂

∂ρ+ eφ

1

ρ

∂

∂φ+ ez

∂

∂z,

derive expressions for ∇ ·A and ∇×A, where A = Aρeρ +Aφeφ +Azez.

Also derive an expression for ∇2f , for a scalar function f .

Don’t just quote div/curl formulae from notes – the intention is for you to derive them.

*3. Show that if the scalar field f(x) has cylindrical symmetry about the line through the originin the direction of the fixed non-zero vector a, then a · (x×∇f) = 0.

Deduce that if f has cylindrical symmetry about three lines through the origin in linearlyindependent directions, then f is spherically symmetric.

4. The vector field B is given in cylindrical polar coordinates (ρ, φ, z) by B = 1

ρeφ, for ρ 6= 0.

Using the formula derived in question 2, show that ∇×B = 0 for ρ 6= 0. Calculate∮

C B · dxwith C the circle given by ρ = R, 0 6 φ 6 2π, z = 0. Why does Stokes’ theorem not apply?

5. Let F satisfy ∇ ·F = 0 in the volume V and F · n = 0 on the boundary ∂V . By considering∂

∂xj(xiFj), show that

∫

V F dV = 0.

Explain why this result makes sense, given the two conditions on F.

*6. Let V be a volume with boundary ∂V, and let P,Q be two solenoidal vector fields. Show that∫

V(Q · ∇2P−P · ∇2Q) dV =

∫

∂V

(

Q× (∇×P)−P× (∇×Q))

· dS.

7. Maxwell’s equations for electric and magnetic fields E(x, t) and B(x, t) are

∇ ·E =ρ

ε0, ∇×E = −∂B

∂t, ∇ ·B = 0, ∇×B = µ0j+ ε0µ0

∂E

∂t,

Show that these imply that the charge density ρ(x, t) and current density j(x, t) satisfy theconservation equation ∇ · j = −∂ρ/∂t. Show also that if j is zero then

U = 1

2(ε0E

2 + µ−10

B2) and P = µ−10

E×B satisfy ∇ ·P = −∂U/∂t .

1

8. (a) The scalar field ϕ(r) depends only on the radial distance r = |x| in R3. Use Cartesian

coordinates and the chain rule to show that

∇ϕ = ϕ′(r)x

r, ∇2ϕ = ϕ′′(r) +

2

rϕ′(r).

What are the corresponding results when working in R2 rather than R

3?

(b) Show that the radially symmetric solutions of Laplace’s equation in two dimensions havethe form ϕ = α+ β log r, where α and β are constants.

(c) Find the solution of ∇2ϕ = 1 in the region r 6 1 in R3 which is not singular at the origin

and satisfies ϕ(1) = 1.

9. (a) Find all solutions of Laplace’s equation, ∇2f = 0, in two dimensions that can be writtenin the separable form f(r, θ) = R(r)Φ(θ), where r and θ are plane polar coordinates.

(b) Consider the following boundary value problem:

∇2f = 0, f(a, θ) = sin θ.

Find the solution for r 6 a which is not singular at the origin.

Find the solution for r > a that satisfies f(r, θ) → 0 as r → ∞.

Find the solution for a 6 r 6 b that satisfies ∂f∂n(b, θ) = 0. Recall that ∂f

∂n = n ·∇f .

10. A spherical shell has density given by

ρ(r) =

0 for 0 6 r < aρ0r/a for a 6 r 6 b0 for b < r <∞

Find the gravitational field everywhere by three different methods, namely

(i) direct solution of Poisson’s equation,

(ii) Gauss’s flux theorem,

*(iii) the integral form of the general solution of Poisson’s equation, ϕ(y) =1

4π

∫

V

−ρ(x)|x− y| dV .

You should assume that the potential is a function only of r, is not singular at the origin andthat the potential and its first derivative are continuous at r = a and r = b.

(Note that you are asked to find the gravitational field, not the potential.)

*11. [Tripos, 2004/III/11]

Let S1 be the 3-dimensional sphere of radius 1 centred at (0, 0, 0), S2 be the sphere of radius1

2centred at (1

2, 0, 0) and S3 be the sphere of radius 1

4centred at (−1

4, 0, 0).

The eccentrically shaped planet Zog is composed of rock of uniform density ρ occupying theregion within S1 and outside S2 and S3. The regions inside S2 and S3 are empty. Give anexpression for Zog’s gravitational potential at a general coordinate x that is outside S1.

Is there a point in the interior of S3 where a test particle would remain stably at rest? Justifyyour answer.

12. The surface S encloses a volume in which the scalar field ϕ satisfies the Klein-Gordon equation∇2ϕ = m2ϕ, wherem is a real non-zero constant. Prove that ϕ is uniquely determined if eitherϕ or ∂ϕ/∂n is given on S.

2

13. Show that the solution to Laplace’s equation in a volume V with boundary condition

g∂ϕ

∂n+ ϕ = f on ∂V

is unique if g(x) > 0 on ∂V .

Find a non-zero (and so non-unique) solution of Laplace’s equation defined on r 6 1 whichsatisfies the boundary condition above with f = 0 and g = −1 on r = 1.

Don’t assume that the solution is spherically symmetric. (Why not?)

14. The functions u(x) and v(x) on V satisfy ∇2u = 0 on V and v = 0 on ∂V . Show that

∫

V∇u ·∇v dV = 0.

Let w be a function on V which satisfies w = u on ∂V . By considering v = w − u, show that

∫

V|∇w|2 dV >

∫

V|∇u|2 dV,

i.e. the solution of the Laplace problem minimises∫

V |∇w|2 dV .

15. The scalar field ϕ is harmonic (i.e., solves Laplace’s equation) in a volume V bounded by aclosed surface S. Given that V does not contain the origin (r = 0), show that

∫

S

(

ϕ∇

(

1

r

)

−(

1

r

)

∇ϕ

)

· dS = 0.

Now let V be the volume given by ε 6 r 6 a and let S1 be the surface r = a. Given that ϕ(x)is harmonic for r 6 a, use this result, in the limit ε→ 0, to show that

ϕ(0) =1

4πa2

∫

S1

ϕ(x) dS.

Deduce that if ϕ is harmonic in a general volume V , then it attains its maximum and minimumvalues on S.

*16. [Tripos, 2005/III/9]

The first part of this question was essentially question 14 above, so I have omitted it here.

Consider the partial differential equation∂w

∂t= ∇2w, for w = w(t,x), with initial condition

w(0,x) = w0(x) in V , and boundary condition w(t,x) = f(x) on S for all t > 0.

Show that∂

∂t

∫

V|∇w|2 dV 6 0, (∗)

with equality holding only when w(t,x) = u(x). (This is the u from the question 14 part.)

Show that (∗) remains true with the boundary condition

∂w

∂t+ α(x)

∂w

∂n= 0

on S, provided α(x) > 0.

3

Lent 2017 VECTOR CALCULUS – EXAMPLES 4 G.Taylor

A star means ‘optional’ and not necessarily ‘harder’.

1. If u(x) is a vector field, show that ∂ui/∂xj transforms as a second-rank tensor.

If σ(x) is a second-rank tensor field, show that ∂σij/∂xj transforms as a vector.

2. Given vectors u = (1, 0, 1), v = (0, 1,−1) and w = (1, 1, 0), find all components of the second-rank and third-rank tensors defined by

Tij = uivj + viwj , Sijk = uivjwk − viujwk + viwjuk − wivjuk + wiujvk − uiwjvk .

3. Any 3 × 3 matrix A can be decomposed in the form Ax = αx + ω × x + Bx, where α is ascalar, ω is a vector, and B is a traceless symmetric matrix. Verify this claim by finding α,ωk and Bij explicitly in terms of Aij .

Check your calculations are correct by finding α, ω and B for the matrix A =

1 2 34 5 61 2 3

.

4. In a particular system of Cartesian coordinates xi, the vector field ui has the components

u1 = x1x22 , u2 = x2x

23 , u3 = x3x

21 .

Express the tensor ∂ui/∂xj as a linear combination of εijkwk (where wk is a vector to bedetermined) and a symmetric tensor eij .

Verify that (0, 0, 1) is one of the principal axes of eij at the point x1 = 2, x2 = 3, x3 = 0, andfind the others.

*5. [Tripos, 2005/I/8 – Algebra & Geometry]

Given a non-zero vector vi, any 3 × 3 symmetric matrix Tij can be expressed in the formTij = Aδij + Bvivj + (Civj + Cjvi) +Dij for some numbers A and B, some vector Ci and asymmetric matrix Dij , where Civi = 0, Dii = 0, Dijvj = 0, and the summation convention isimplicit.

Show that the above statement is true by finding A, B, Ci and Dij explicitly in terms of Tijand vj , or otherwise. Explain why A, B, Ci and Dij together provide a space of the correctdimension to parameterise an arbitrary symmetric 3× 3 matrix Tij .

6. The electrical conductivity tensor σij has components

σij =

2 −1 −1−1 2 −1−1 −1 2

.

Show that there is a direction along which no current flows, and find the direction(s) alongwhich the current flow is largest, for an electric field of fixed magnitude.

7. A body has symmetry such that it is unchanged by rotations of π about three perpendicularaxes which form a basis B. Show that any second-rank tensor calculated for the body will bediagonal in this basis, although the diagonal elements need not be equal.

Find the inertia tensor of a cuboid of uniform density with sides of length 2a, 2b and 2c aboutthe centre of the cuboid.

1

8. For any second-rank tensor Tij , prove using the transformation law that the quantities

α = Tii, β = TijTji, and γ = TijTjkTki

are the same in all bases.

If Tij is a symmetric tensor, express these invariants in terms of the eigenvalues. Deduce thatthe cubic equation for the eigenvalues is

λ3 − αλ2 + 1

2(α2 − β)λ− 1

6(α3 − 3αβ + 2γ) = 0.

9. Let S be the surface of the unit sphere.

(a) Calculate the following integrals using properties of isotropic tensors:

(i)

∫

Sxi dS, (ii)

∫

Sxixj dS.

*Verify your answers using the tensor divergence theorem.

(b) For the second-rank tensor Tij = δij + εijkxk, calculate the following integrals:

(i)

∫

STij dS, (ii)

∫

STijTjk dS.

10. Evaluate the following integrals over the whole of R3 for positive γ and r2 = xpxp :

(i)

∫

r−3e−γr2xixj dV, (ii)

∫

r−4e−γr2xixjxk dV, *(iii)

∫

r−5e−γr2xixjxkxℓ dV.

11. (a) Given that the most general isotropic rank 4 tensor is λδijδkℓ + µδikδjℓ + νδiℓδjk forλ, µ, ν ∈ R, show that

εijkεiℓm = δjℓδkm − δjmδkℓ.

*(b) Prove the following result:

εijkεpqr =

∣

∣

∣

∣

∣

∣

δip δiq δirδjp δjq δjrδkp δkq δkr

∣

∣

∣

∣

∣

∣

.

12. (a) A tensor of rank 3 satisfies Tijk = Tjik and Tijk = −Tikj . Show that Tijk = 0.

(b) A tensor of rank 4 satisfies Tjikℓ = −Tijkℓ = Tijℓk and Tijij = 0. Show that

Tijkℓ = εijpεkℓqSpq, where Spq = −Trqrp.

*13. The array dijk with 33 elements is such that dijksjk is a vector for every symmetric second-ranktensor sjk. Show that dijk need not be a tensor, but that dijk + dikj must be.

14. In linear elasticity, the symmetric second-rank stress tensor σij depends on the symmetricsecond-rank strain tensor eij according to σij = cijkℓekℓ. Assuming that cijkℓ = cijℓk, explainwhy cijkℓ must be a fourth-rank tensor. Show that in an isotropic material,

σij = λδijekk + 2µeij (∗)

with two scalars λ and µ. Assume below that µ > 0 and λ > −2

3µ.

Invert (∗) to find an expression for eij in terms of σij , and explain why the principal axes ofof σij and eij coincide.

The elastic energy density resulting from a deformation of the material is given by E = 1

2eijσij .

Show that E > 0 for any non-zero strain eij .

2

*15. A conductor positioned in a magnetic field H carries a steady current density J = ∇×H,and the magnetic flux intensity B = µH satisfies ∇ ·B = 0. The mechanical force per unitvolume acting on the conductor can be written as J×B. If the permeability µ is a constant,show that this force per unit volume can be written as ∂sik/∂xk in terms of a tensor

sik = µ(

HiHk − 1

2HmHmδik

)

.

*16. [Tripos, 2011/III/9]

Write down the most general isotropic tensors of rank 2 and 3. Use the tensor transformationlaw to show that they are, indeed, isotropic.

Let V be the sphere 0 6 r 6 a. Explain briefly why Ti1...in =∫

V xi1 . . .xindV is an isotropictensor for any n. Hence show that

∫

Vxixj dV = αδij ,

∫

Vxixjxk dV = 0,

∫

Vxixjxkxℓ dV = β(δijδkℓ + δikδjℓ + δiℓδjk)

for some scalars α and β, which should be determined using suitable contractions of the indicesor otherwise. Deduce the value of of

∫

Vx× (Ω× x) dV,

where Ω is a constant vector.

*17. [Tripos, 2009/III/11]

A second-rank tensor T (y) is defined by

Tij(y) =

∫

S(yi − xi)(yj − xj)|y − x|2n−2 dA(x) ,

where y is a fixed vector with |y| = a, n > −1, and the integration is over all points x lyingon the surface S of the sphere of radius a, centred on the origin. Explain briefly why T mightbe expected to have the form

Tij = αδij + βyiyj ,

where α and β are scalar constants.

Show that y · (y−x) = a2(1− cos θ), where θ is the angle between y and x, and find a similarexpression for |y − x|2. Using suitably chosen spherical polar coordinates, show that

yiTijyj =πa2(2a)2n+2

n+ 2.

Hence, by evaluating another scalar integral, determine α and β, and find the value of n forwhich T is isotropic.

3