You are on page 1of 18

Chapter 3 Cartesian Tensors

3.1 Suffix Notation and the Summation Convention

We will consider vectors in 3D, though the notation we shall introduce applies (mostly) just as well to n dimensions. For a general vector x = (x 1 , x2 , x 3 ) we shall refer to xi , the ith component of x. The index i may take any of the values 1, 2 or 3, and we refer to “the vector xi ” to mean “the vector whose components are (x1 , x2 , x3 )”. However, we cannot write x = xi , since the LHS is a vector and the RHS a scalar. Instead, we can write [x]i = xi , and similarly [x + y]i = xi + yi . Note that the expression yi = xi implies that y = x; the statement in suffix notation is implicitly true for all three possible values of i (one at a time!). Einstein introduced a convention whereby if a particular suffix (e.g., i) appears twice in a single term of an expression then it is implicitly summed. For example, in traditional notation

x . y = x1 y1 + x2 y2 + x3 y3 =

xi yi ;

using summation convention we simply write x . y = xi yi . All we are doing is not bothering to write down the !

The Rules of Summation Convention
Summation convention does not allow any one suffix to appear more than twice within a single term; so xi yi zi is meaningless. We have to take care to avoid this: for example, 46

© R. E. Hunt, 2002

a in vector notation). and one that appears precisely twice. b = ai b i = aj b j – the suffix we use for the summation is immaterial.. b)x. 3). We have a . It is an important precept of summation convention that the free suffixes must match precisely in every term (though dummy suffixes can be anything you like so long as they do not clash with the free suffixes). For a matrix A. Suffix notation can also be used with matrices. b)(x . We write either A = (aij ) or [A]ij = aij – these equations are equivalent – to indicate this. z = (x + y) .consider the vector relation y = (a . as in the equation ai zi = (xi + yi )ai (which reads a . Note that the RHS of this suffix notation equation does not mean z1 + z2 + z3 – no repeated suffix. E. y) = 0 ⇔ ai bi xj yj = 0. i in aj bj xi . there are two possible types of suffix: one that appears precisely once. b = ai bi . e.g. this equation reads (a . but we cannot write yi = ai bi xi as this would be ambiguous. b)z = x + |a|2 |b|2 y. in which case the matrix A would have entries Aij . How can we correct this? Note that a .. and all other suffixes are dummy ones. (Note that the order of variables in the final term of this equation in suffix notation is unimportant: we could equally well have written bj yk ai bj ai .) Examples of Summation Convention (i) 2x + y = z ⇔ 2xi + yi = zi . which is known as a dummy suffix. no sum! (ii) (a . 3 and j = 1. Hunt. e. 2. j in aj bj xi . 2002 . then.) There need not be any free suffixes at all.) So we can write y i = aj b j x i . we write aij to denote the entry in the ith row and j th column of A (for each i = 1. So in the equation aj bj zk = xk + ai ai yk bj bj every term has a free suffix k . 47 © R. In vector notation. which is known as a free suffix . In any given term. 2. (Sometimes the upper-case letter is used instead. (Compare with the use of dummy ∞ ∞ variables in integrations: 0 e−x dx = 0 e−y dy .g.

j in cij ) by a single dummy suffix (cii ) is known as contraction. For example. i.. E. k are equal (i. (v) The trace of a matrix C may be written as Tr C = cii .g. If I is the identity matrix 1 0 0 0 1 0 0 0 1 then [I ]ij = δij . Note that δij = δji and that ijk = jki = kij = − jik etc. (iv) The matrix multiplication C = AB (where A and B are 3 × 3 matrices) is written cij = [AB ]ij = aik bkj . 3. 123 = 231 = 312 = 1. 213 = 132 = 321 = −1. 2002 . i. 2. Not all expressions written in suffix notation can be recast in vector or matrix notation.2 The Kronecker Delta and the Alternating Tensor The Kronecker delta is defined by  1 i = j δij = 0 i = j and the alternating tensor is defined by     1 if (i. y = Ax is written yi = [Ax]i = aij xj (check that this is correct by writing it out long-hand for each possible value of the free suffix i). We see that xi = δij xj 48 © R. j and k ). k ) is an anti-cyclic permutation of (1. j.. j. Hunt. j . 2. Hence Tr(AB ) = aik bki . aijk = xi yj zk is a valid equation in suffix notation (each term has three free suffixes. i.e.(iii) In summation convention.e. c11 + c22 + c33 . k ) is a cyclic permutation of (1. Replacing two free suffixes (e. 3) ijk = −1 if (i. but there is no vector equivalent. all others are zero). 3)    0 if any of i.

b)ci = [(a .. noting that for example δjk δkm = δjm . 2002 .. Hunt. m.g. b)c]i .) Example: prove that a × (b × c) = (a . contract k and n and simplify. This makes many vector identities easy to prove. The determinant of a 3 × 3 matrix A = (aij ) is given by ijk a1i a2j a3k (check this by just expanding the product and sum in ijk 49 © R. b)c. z1 = important property of 123 x2 y3 ijk : + 132 x3 y2 = x2 y3 − x3 y2 . δik ajk is equal to aji .) There is one very ijk klm = δil δjm − δim δjl . or if any of l.g. c)b − (a . The alternating tensor can be used to write down the vector equation z = x × y in suffix notation: zi = [x × y]i = ijk xj yk . as required. c)bi − (a . The Kronecker delta just “selects” entries: e. k are equal. E. can also be used to calculate determinants. (The property may be proved by first proving the  δil δim  ijk lmn = det  δjl δjm δkl δkm generalisation  δin  δjn  .g. Once we have proved this generalisation. δkn Both sides clearly vanish if any of i. k = n = 3: both sides are clearly 1. n are. RHS = δ1j xj = δ11 x1 + δ12 x2 + δ13 x3 = x1 = LHS). c)b − (a . when i = 1. What is δii ? It is not 1.because (i) this is equivalent to x = I x. as required. or (ii) we can check for each value of i (e. (Check this: e. Now take i = l = 1. j = m = 2.. j. [a × (b × c)]i = = ijk aj [b × c]k ijk aj klm bl cm = (δil δjm − δim δjl )aj bl cm = aj b i c j − aj b j c i = (a . Finally consider the effect of swapping say i and j .

ei = vj δij = vi but also v . This looks like. det A = ijk a1i a2j a3k = jik a1j a2i a3k [swapping i and j ] =− ijk a2i a1j a3k . and that a vector v has components vi relative to axes along those vectors. ej . just measured with respect 50 © R. 2002 .) For example. e3 }? Let v = v 1 e 1 + v 2 e2 + v 3 e 3 = v j e j . This can be written in several other ways. but is not quite the same as. rotating the vector v round to a different vector v using a transformation matrix L.3 What is a Vector? A vector is more than just 3 real numbers. v and v are the same vector. In the present case. 3. L is called the rotation matrix.full). That is to say. E. This proves that swapping two rows of a matrix changes the sign of the determinant. ei = vj ej . so v . ei = vj ej . (The vector stays the same even if its components do not. in matrix notation. e2 . e2 . e3 } is a right-handed orthogonal set of unit vectors. Hunt. It is also a physical entity: if we know its 3 components with respect to one set of Cartesian axes then we know its components with respect to any other set of Cartesian axes. What are the components of v with respect to axes which have been rotated to align with a different set of unit vectors {e1 . ej = δij . Now ei . Then vi = lij vj (or. suppose that {e1 . ei = vj lij where we define the matrix L = (lij ) by lij = ei . for example. v = Lv where v is the column vector with components vi ). v = v 1 e 1 + v 2 e2 + v 3 e 3 = v j e j .

e 1 e 2 . we see that the ith row of L just consists of the components of ei measured with respect to the first frame:   e 1 . e3   = e1 T e2 T e3 T    [measured with respect to frame 1]. e2 . this is true for all vectors v. 51 © R. In suffix notation. Exactly the same discussion would lead to vi = ˆ lij vj where ˆ lij = ei . both of these identities will be useful. e1 e 1 . Hunt. The transformation matrix corresponding to the rotation {e1 . We conclude that LT L = I. ej is just the j th component of ei measured with respect to the first frame. Since ei .e. and LLT = I reads lik ljk = δij . the equation LT L = I reads lki lkj = δij . (Hence LLT = I also. e2 e 3 . Now consider the reverse of this argument. Another way of seeing that LLT = I (or. and furthermore. ej (we swap primed and unprimed quantities throughout the argument). e3 } → {e1 .. e3   L = e 2 . v=L We can deduce that v = LT Lv . e2 . 2002 .to different axes. equivalently. e 2 e 2 . hence ˆ = LT L and so ˆ v = LT v . e2 e 1 .) L is therefore an orthogonal matrix. e3 } is not L (in fact it is L−1 ). LT L = I ) is to consider the components of L. E. LT = L− 1 . e1 e 3 . e 3  e 3 . i. We note that ˆ lij = lji from their definitions.

Hunt.) More generally. e1 . so Ji = lij Jj 52 © R. J = σ E. Both are simply e1 measured with respect to the first frame. E. which is zero. 3. What happens if we measure J and E with respect to a different set of axes? We would expect the matrix σ to change too: let its new components be σij . Similarly. the ith column consists of the components of ei with respect to the second frame. like a vector. we find the dot product of the first row of L with the first column of LT . So.   5 0 0   J = 0 4 0 E 0 0 0 might represent a medium in which the conductivity is high in the x-direction but in which no current at all can flow in the z -direction. But J and E are vectors. But a matrix formulation may be more suitable in anisotropic media. Then Ji = σij Ej . Physical Motivation Recall the conductivity law. a crystalline lattice structure where vertical layers are electrically insulated. This is suitable for simple isotropic media. and which also has a physical significance independent of the frame. e3 . in suffix notation we have Ji = σij Ej where σ is the conductivity tensor. the top right component of LLT is e1 .4 Tensors Tensors are a generalisation of vectors. we see that LLT = I as required. considering all possible combinations of rows and columns.Alternatively. where E is the applied electric field and J is the resulting electric current. We think informally of a tensor as something which. which is 1. like a vector. for example. 2002 . To calculate the top left component of LLT . so we obtain e1 . can be measured component-wise in any Cartesian frame. (For instance. where the conductivity is the same in all directions.

a tensor of rank n is a mathematical object with n suffixes. Consider a mass m which is part of a rigid body. This is true for all vectors E . where L is the rotation matrix between frames.and Ei = lji Ej from the results regarding the transformation of vectors in §3.. Note: for second rank tensors such as σ . hence σij = lip ljq σpq . 2002 . 53 © R. Hunt. velocity) is a tensor of rank 1.. Tpqr. This tells us how σ transforms. at a location x within the body.. if it is true for all x then A must be zero. (ii) Temperature T is a tensor of rank 0 – known as a scalar – because it is the same in all frames (T = T ).3. though this is not the case if it is only true for some x’s... Definition: In general. because it has two suffixes (σij ). . because vi = lip vp . Compare this argument with the corresponding argument for the case Ax = 0 where A is a matrix. Hence σij Ej = Ji = lip Jp = lip σpq Eq = lip σpq ljq Ej =⇒ (σij − lip ljq σpq )Ej = 0. = lip ljq lkr . and hence the bracket must be identically zero. . x)x .g. . and its angular momentum is therefore mx × v = mx × (! × x) = m |x|2 ! − (! . If the body is rotating with angular velocity ! then the mass’s velocity is v = ! × x. Tijk. (iii) The inertia tensor. the transformation law Tij = lip ljq Tpq can be rewritten in matrix notation as T = LT LT – check this yourself! Examples of Tensors (i) Any vector v (e. σ is a second rank tensor.. which obeys the transformation law Tijk. E..

only on properties of the body itself.3). in suffix notation.e. then. ijk are isotropic tensors : that is. (iv) Susceptibility χ. which is exactly the right transformation law. we can show that it is a tensor: lip ljq δpq = lip ljp = δij = δij (from §3. 2002 . x)x dV. and apply arguments previously used for the conductivity tensor. their components are the same in 54 © R. Note that the tensor I does not depend on !. (v) The Kronecker delta itself. or. then for a simple medium we have M = (m) χ(m) B where χ(m) is the magnetic susceptibility. Surprisingly. We have defined δij without reference to frame. note that both h and ! are vectors. Both δij and all frames. To see that it is indeed a tensor. we see that the total angular momentum of a rigid body V is given by h= V ρ(x) |x|2 ! − (! . We can also show that ijk is a tensor of rank 3.Changing from a single mass m to a continuous mass distribution with density ρ(x). E. so that an infinitesimal mass element is ρ(x) dV . hi = V ρ(x)(xk xk ωi − ωj xj xi ) dV = V ρ(x)(xk xk δij − xj xi )ωj dV = Iij ωj where Iij = V ρ(x)(xk xk δij − xi xj ) dV is the inertia tensor of the rigid body.. so it may be calculated once and for all for any given body. Hunt. its components are by definition the same in all frames (δij ≡ δij ). This generalises to Mi = χij Bj (m) where χij is the magnetic susceptibility tensor. Similarly for polarization density (e) (e) P in a dielectric: Pi = χij Ej where E is the electric field and χij is the electric susceptibility tensor. If M is the magnetization (magnetic moment per unit volume) and B is the applied magnetic field. i.

and α. Contraction (also known as the Inner Product) If Tij is a tensor then Tii is a scalar. 3. stresses (forces) produce displacements of small volume elements within the body. Proof: Tii = lip liq Tpq = δpq Tpq = Tpp = Tii . We can now generalise this to the tensor formulation pij = kijkl ekl where kijkl is a fourth rank tensor. The stress tensor pij is defined as the j th component of the forces within the body acting on an imaginary plane perpendicular to the ith axis. β are scalars. E. so Tii has the same value in all frames as required. 55 © R. Hooke’s law for simple media says that stress ∝ strain.(vi) Stress and strain tensors. Let this displacement at a location x be u. 2002 . Proof: Tij = α Aij + β Bij = αlip ljq Apq + βlip ljq Bpq = lip ljq (αApq + βBpq ) = lip ljq Tpq as required. This result clearly extends to tensors of rank n. In an elastic body. then the strain tensor is defined to be eij = 1 2 ∂ui ∂uj + ∂xj ∂xi . which expresses the linear (but possibly anisotropic) relationship between p and e. Hunt. then Tij = αAij + βBij is a tensor.5 Properties of Tensors Linear Combination of Tensors If Aij and Bij are second rank tensors.

Then Tii = ai bi . . j if Tijk.. = Aijk.βγ.. lmβ lnγ ... Example: if a and b are vectors then a . For a second rank tensor we need not specify the indices as there are only two to choose from! For example.. Similarly (left as an exercise for the reader) we can show that if Note that |a|2 = a . . hence a .αβγ..... is a tensor of rank m + n. = Tijk.. .... Tpqr. being an outer product of two vectors. Proof: Tij = ai bj . 56 © = −Tjik..imn..lmn. is a tensor of rank n − 2... 2002 ...... = Tjik.. b/|a| |b| = cos θ is a scalar. as required. 3.... Hunt... j ) if Tijk. is a tensor of rank two. so that the angle between vectors is unaffected by a change of frame. = ljq lkr . lmβ lnγ . . being a contraction of a tensor.. b is a scalar. Blmn. is said to be symmetric in a pair of indices (say i.... E.. . Sqr. . is a scalar.. a and |b|2 are also scalars...We can extend this result: if Tijk. . Tpqr. . . is a tensor of rank m and Blmn. . ijk is anti-symmetric in any pair of indices.. δij is symmetric. = Tijk..... lmβ lnγ . = (lip liα )ljq lkr ..imn.6 Symmetric and Anti-Symmetric Tensors A tensor Tijk. liα lmβ lnγ . or anti-symmetric in i. = lip ljq lkr ... .. Tpqr.αβγ.. Outer Product If a and b are vectors then the outer product Tij defined by Tij = ai bj is a tensor of rank two. ... is a tensor of rank n then Sjk. .lmn. Proof: Tij = ai bj = lip ap ljq bq = lip ljq ap bq = lip ljq Tpq as required..αβγ. . then Tijk. Proof: Sjk. . . is a tensor of rank n.. = δpα ljq lkr . . .

E. (swapping dummy i and j ) as required. and simpler. Suppose that Sij is a symmetric tensor and Aij an anti-symmetric tensor. by considering appropriate pairs of components.e. In most situations the the strain tensor eij = 2 stress tensor is also symmetric. conductivity and susceptibility tensors usually are. Similarly for an anti-symmetric tensor. Try to work out also how to see this “by inspection”. Proof: Sij Aij = −Sij Aji = −Sji Aji = −Sij Aij =⇒ 2Sij Aij = 0. Then Sij Aij = 0. Hunt. but in some circumstances (for instance in crystallography or geodynamics) it is forced to be anti-symmetric while the strain remains symmetric. k whilst aj ak is symmetric. Then Tij = lip ljq Tpq = lip ljq Tqp = ljq lip Tqp = Tji . a × a = 0 because [a × a]i = and ijk ijk aj ak is anti-symmetric in j . The properties of symmetry and anti-symmetry are invariant under a change of frame: that is. 1 (∂ui /∂xj + ∂uj /∂xi ) is clearly symmetric.Note: if Aij is a symmetric second rank tensor then the matrix corresponding to A is symmetric. proof for second rank tensors: T = LT LT using T T = T . For example. suppose that Tij is symmetric. A = AT . so that Tij is also symmetric. For example. Example: for any vector a. Inertia tensors are always symmetric. 2002 . i. they are truly tensor properties. =⇒ T T = (LT LT )T = LT T LT = LT LT = T 57 © R. (Alternative.) Symmetry and anti-symmetry occur frequently in practical applications.

It is not necessarily the case that it must also be symmetric in k . which we showed was zero above. but without loss of generality (s) we may assume that it is. Then Rij = cijkl Skl + cijkl Skl = cijkl Skl because the second term is the contraction of an anti-symmetric tensor with a symmetric one. ! is a vector as it is a contraction of two tensors. Hence we can ignore any anti-symmetric part of cijkl . j (for otherwise. any anti-symmetric tensor Aij can be expressed in terms of a vector (sometimes known as the dual vector ) such that Aij = Proof: define ijk ωk . 2 Aij = 1 (Tij − Tji ) 2 are symmetric and anti-symmetric respectively.Decomposition into Symmetric and Anti-Symmetric Parts Any second rank tensor Tij can be uniquely expressed as the sum of a symmetric and an anti-symmetric tensor. 2002 . This definition of ! actually corresponds to setting  0 ω3 −ω2   A = −ω3 0 ω1  . E. 2 klm lm Then ijk ωk = 1 A 2 ijk klm lm =1 (δ δ − δim δjl )Alm 2 il jm 1 (Aij − Aji ) = Aij =2 as required. Exercise: prove that S and A are tensors. Furthermore. by the following argument. ω2 −ω1 0  Example: suppose that two symmetric second rank tensors Rij and Sij are linearly related. ! ! by ωk = 1 A . l and a part cijkl which is anti-symmetric. It is clear that cijkl must be symmetric in i. 58 (s) (a) (s) © R. l. Decompose cijkl into a part cijkl (a) which is symmetric in k . for Tij = Sij + Aij where Sij = 1 (Tij + Tji ). Then there must be a relationship between them of the form Rij = cijkl Skl . Hunt. Rij would not be).

λ2 .  λ1 0 0   =  0 λ2 0  0 0 λ3 because. e2 . T LT = T e 1 e 2 e 3 = So   T = LT LT =   e1 T e2 T e3 T    λ 1 e 1 λ2 e 2   λ3 e 3  λ1 e 1 λ 2 e 2 λ3 e 3 . for example. λ1 0 There is another way of seeing that T =  0 λ2 0 0 (because T is a tensor. e3 be the eigenvectors (arranged as a right-handed set of orthonormal vectors). The equation T e1 = λ1 e1 is true in any frame λ3 In particular it is true in the frame with {e1 . Hence in matrix notation. e3 }. Change frame to one in which the coordinate axes are aligned with {e1 . the top LHS entry is given by e1 . 2002 . e2 and e3 (measured relative to the first frame). Because T is symmetric. What is T ? Recall that LT = e1 e2 e3 .7 Diagonalization of Symmetric Second Rank Tensors Suppose Tij is a symmetric second rank tensor.   0 0 .e. λ2 and λ3 are known as the principal values of T . Hunt.3. if we transform T to that frame. it has components given by   λ1 0 0   T =  0 λ2 0  . and the principal axes are the eigenvectors. and the top RHS entry is e 1 . i. the three columns of LT are the vectors e1 .. The values λ1 . and the Cartesian coordinate axes of the corresponding frame are known as the principal axes. e2 . e1 a vector and λ1 a scalar). E. λ3 e 3 . λ1 e1 . we know that there are 3 real eigenvalues and that we can find 3 corresponding eigenvectors which are orthogonal and of unit length. e2 . We will see that in fact the principal values are just the eigenvalues of the matrix corresponding to T . λ3 be the eigenvalues and e1 . Let λ1 . 0 0 λ3 This process is known as diagonalization. e3 } 59 © R. We shall show that there exists a frame such that.

Rank 2: The most general isotropic second rank tensor is λδij where λ is any scalar. = Tijk.8 Isotropic Tensors An isotropic tensor is one whose components are the same in all frames. We can classify isotropic tensors up to rank four as follows: Rank 0: All scalars are isotropic. So we recover the “simple version” of the conductivity law. Note: the three principal values are invariants of T .e. Hence σij = λδij and Ji = σij Ej = λδij Ej = λEi . J = λE.. Hunt. and T has components T . The eigenvalues are properties of the tensor. But. .e. i. as proved below. What is the physical significance of an isotropic tensor? Consider the conductivity tensor σij in an isotropic medium. 2002 .. 0)T . measured in this frame. Rank 1: There are no non-zero isotropic vectors. 0. not of the coordinate system. we expect that σij will be isotropic too. when we diagonalize T we will obtain the same values of λ coordinate axes.. As the medium is the same in all directions. Rank 3: The most general isotropic third rank tensor is λ Rank 4: The most general isotropic fourth rank tensor is λδij δkl + µδik δjl + νδil δjk where λ. 0)T . since the tensor transformation law states that T = T for tensors of rank zero. 60 ijk . 0. ν are scalars. 3. Tijk.. Similarly for the other columns. i. That is. e1 is just (1. µ.. © R. E. as we might expect. whatever frame we start from. so     1 λ1 T 0 =  0  0 0 which shows that the first column of T is (λ1 .

Isotropic Second Rank Tensors Consider a general tensor T of rank two. Hunt. Its components should then be unaltered by a rotation of 90◦ about the 3-axis. 0 0 1 we see that  −1 0  0 0 0 1 Using the matrix  T11  T21 T31 formulation of the transformation law for tensors. e2 = −e1 . Therefore all off-diagonal elements of T are zero. T31 = T32 = 0. i. T32 −T31 T33 But. with components Tij with respect to some set of axes {e1 . Similarly.     0 1 0 T11 T12 T13 0 T12 T13     T22 T23  = −1 0 0 T21 T22 T23  1 0 0 1 T31 T32 T33 0 T32 T33   T22 −T21 T23   = −T12 T11 −T13  .  The matrix of this rotation is  0 1 0   L = −1 0 0 . we have: T11 = T22 . considering a rotation of 90◦ about the 2-axis. E. In summary. we find that T11 = T33 and that T12 = T32 = 0. Thus   λ 0 0   T = 0 λ 0 . comparing matrix entries. T21 = T23 = 0. e2 . Hence. and all diagonal elements are equal. 2002 . say λ. 61 © R.e. with respect to new axes e1 = e2 . we have shown that any isotropic second rank tensor must be equal to λδij for some scalar λ. T13 = T23 = −T13 T31 = T32 = −T31 so that so that T13 = T23 = 0. Tij = λδij . e 3 = e3 . e3 }. Tij = Tij . because T is isotropic.. 0 0 λ or in suffix notation. Suppose that T is isotropic.

Hunt. div. we have σij = σij (x). we start with scalars and vectors. We can rewrite the definitions of grad.9 Tensor Differential Operators A tensor field is a tensor which depends on the location x. ∇)v]i = uj Laplace’s equation ∇2 Φ = 0 becomes ∂2Φ =0 ∂xi ∂xi in suffix notation. v are vectors then we define the vector (u . Similarly. For example: (i) Temperature is a scalar field (a tensor field of rank zero). (note that we only use Cartesian coordinates here). because T = T (x). [(u . [∇ 2 F ]i = ∂ 2 Fi ∂xj ∂xj ∂vi . We are interested here in calculating the derivatives of tensor fields. and curl using suffix notation. Grad: Div: Curl: [∇Φ]i = ∇.3. In particular. is a tensor field of rank one. E. ∇)v = In suffix notation. ∂xi 62 © R. We sometimes find it useful to use the differential operator ∂i defined by ∂i = ∂ . (ii) Any vector field F(x). such as a gravitational force field. because it is a vector function of position! (iii) In a conducting material where the conductivity varies with location.F= [∇ × F]i = ∂Φ ∂xi ∂F1 ∂F2 ∂F3 ∂Fi + + = ∂x1 ∂x2 ∂x3 ∂xi ijk ∂Fk ∂xj There is another notation worth knowing: if u. x is itself a vector field. 2002 . a tensor field of rank two. ∂xj u1 ∂ ∂ ∂ + u2 + u3 ∂x1 ∂x2 ∂x3 v.

F and ∇ × F are scalars or vectors (as appropriate). i. E. consider the derivative of the conductivity tensor. ∂xj = lij This is true for any quantity T . to show that if F is a vector field then ∇ × F is a vector field: [∇ × F]i = ijk ∂j Fk pqr ljs ∂s lkt Ft = lip ljq lkr [ . ∇ . Then ∂i T = = ∂T ∂x1 ∂T ∂x2 ∂T ∂x3 ∂T = + + ∂xi ∂x1 ∂xi ∂x2 ∂xi ∂x3 ∂xi ∂T ∂xj ∂xj ∂xi ∂T = lij ∂j T. we can just state that ∇ × F is a contraction of the tensor outer product Tijklm = (because [∇ × F]i = Tijkjk ). ∂i σjk . so ∂i = lij ∂j . ijk ∂l Fm As an example of a tensor field of rank three. ∂xi ∂xi ∂xi (This looks obvious but has to be proved very carefully!) Now let T be some quantity (perhaps a scalar or a tensor of some rank). ∂i transforms like a vector. 63 © R. ∇ . [∇ × F ] i = ijk ∂j Fk . We know that xj = lij xi (from x = LT x ) so that ∂xj ∂x ∂ = (lkj xk ) = lkj k = lkj δik = lij . It turns out that ∂i is in fact a tensor of rank one.. ∂ and F are tensors] = lip (ljq ljs )(lkr lkt ) = lip δqs δrt = lip pqr ∂s Ft pqr ∂s Ft pqr ∂q Fr = lip [∇ × F]p . 2002 . Hunt. Alternatively. F = ∂i Fi . as required. This result allows us to prove that ∇Φ.e. For example. This cannot be written using ∇.Then [∇Φ]i = ∂i Φ. and is hence a tensor of rank one.