484 GENERAL TENSORS
‘Therfore the Jacobian matrix J ofthe inverse transformation is given by
ax! ax!
= ax! ax? “2(x?)? ass
Helge l= 1Gy Gi
ax’ ox ty
Now from (a), we get
Adj J 1
serio 7 2a
requires n numbers called its components . Scalars and
‘object called a tensor of order n , whose specification
noes at aide comegat ta ra In
ised te ule ortho et
wh 8.12 ZEROTH - ORDER TENSORS ( OR
Suppose ¢ is a function ofthe coordinates |OF Store tis oad wo eda wt
coordinates system (5g? om
7a See
ont 4 @
Favation () represents the following neq be,
go. gi
i! . 2 es
A ie TaytA’ +. Sat
a? - eee a,
AU s agrAl FTA rn EAD
A at Pay
which may be written in matrix form as
si ax! ax! ox! |rat
a ax" J
=! as! og! Teal
i 2: ox,
ax! x? ax ao
pCa, eee ag |
ax’ ax? ox"
ax ax’ oF |] ys
2[Note that a superscript is used to indicate contravariant components whereas a subscript is used to indicate
‘covariant components . :
Equation (iv) represents the following n equations :
which may be written in matrix form as
ox! ox?
ax! ax!
xtFor n = 3, wehave 37 (= 9) quantities in which case the matrix
x ze ox.
x
COVARIANT TENSOR OF ORDER TWO
‘A set of n? quantities Ajj in a coordinate system (x', x7, ..
components
coordinate system (X!, X?
iy ft
ax?
An An
An
rd
ox’ ax
‘xt axt
ax! ax?
a
ataiaxt oy
oe oF 2
ax” Ox" ~ax®
form (ii) reduces to
‘ans Al aa x! as? ax?
ox* az" tax! ast
x mt 2 and
x AY am ae ax’ ax’ ox e
ex", ox? ax ax | ™
ox’ ub) O85 Ome On
at Ea a PL ays ax? xt.
+x") is said to form the
of a covariant tensor of order two if it is related to another set of n? quantities AP* ina
x") by the transformation equationsgz
|
>
=
>
Ino
re XL
ox"
ox'
Ox! ax? ox?
lo
@
2
Pe, x
ox"
ox’
For n= 3, wehave 37(= 9) quantities in which case the matrix form (v) reduces to
_ ee A ee Ae
Au An Ao Ox Ox Ox
afm OR Oe OM ee ck
Ay RS ORS ax ax ax a An An
x in i ax! x? ax?
A: A A. = ie
a ee ax ax ag JLAn An As
MIXED TENSOR OF ORDER TWO
ax
ox
ax'
lo @ Jo
le. x [se
k
”
Ox
ax"
Sx! ax!
ax? ax?
Ox? ax?
ae 8x"
ax? ox?
a
ex’ ax
Asset of n? quantities Aj in a coordinate system (x19 ocx) isd fom te
‘components of a mixed tensor of order two if it is related to another set of n
coordinate system (X", x”
nie a? an)
ax! axa)VECTOR AND TENSOR ANALYSIS
Let the c
* the coordinate transformation be! = (x! x2, ox") 3212.00
be the: it
% nding com, i
Ete ing components ofthe covariant vector in the coordinate system
), then the laws of transformation is
e @
Differentiating both sides of equation () wt. %%, we get
Lee
4
2
*
1
53 Ai
ax? 8x axt axfax?
ax! ax) aay, _atxl
ax? ax4 2%) axPax®
oA
‘Due to the presence of the second term on the RLH.S. of equation (ii), “5 7} does not transform according
oAy
to the transformation law of the contravariant vector . Hence “57 isnot a tensor.
8.19 RANK OF A TENSOR
-The rank of a tensor is determined by the number of indices attached to it. As a matter of fact the
umber of dimensions ofa tensor wien raised to the power of rank, el the munber of omponsts of
the tensor and hence the components ofa matrix that represents the tensor, As uch atesar ofa is
four ~ dimensional space has 4" components . Consequently the rank of tensor gives the numberof
the mode of changes of physical quantity when passing from one system to aoter Sem rh
clear from this discussion that a quamty that does nt ch
rank
= (dimension )
=(4)' 2 4x4x4x4x4 = 1024 Ans.VECTOR AND TeNsoR ANALYsIs
501
where @%!@2.. Fp NY.. Yr zara, prilaccte
Ba Bq 8183... 8, DB ise ae
. a amet te hae cpr (p+q) and (15) is again a tensor
of contravariant order p +p and covariant order (q +s)
EXAMPLE (15): Show that if AJ and By aretwo fensors then their outer product A! By isa
tensor of order three .
; SOLUTION: Mek Che artnyy vb AG and B, be the coresponding components ofthe even
; tensors in the coordinate system (', x +X"), then their transformation laws are :
which shows that Cj, = A} Bx is a mixed tensor of order 3 i.e. contravariant order 1 and covariant
order?2.
21 CONTRACTION ,
4 aii ne ee i as
is two
indie i called contraction. By such procs esr is fomed wow nk
tie alfa ch ass fac anag A Lae
than the rank . com
indices ) cannot be contracted .GENERAL ‘TENSORS
510
ax"
‘Multiplying both sides of this equation by ox wrt
vax?s ax? axt axis Beh oe
oe aah oF ule
ty, Bat Ox. _ Ok Ox d a
Gt oxk ax” Ox! Ox?” Ox" xt ax
axe oxi axis axis ox® iris a
a ida
ax? ax? oat ox! axl: axis ax* inianir
“Ox an ORT aEH axe” ORY Ox™
P, ax?
zPiPanPr gt ox sk
yaya, t Om” BRM |xT “Ox ag
ax’ ax"
Replacing the index m by t, we get
ax ax) axis
st Bx Oxt Ox" oe axe
which shows tat A 2" ju k i Attar ofthe indicated type and oder
hae
824 SYMMETRIC AND SKEW - SYMMETRIC TENSORS
SYMMETRIC TENSOR
A tensor is called symmetric w.r. two contravariant or two covariant indices if its components
remain unaltered upon interchange of the indices . Thus the tensor Ase +
epee as ton im and ip if
Me altace Imi
“ipa ode byt edd
For example, if AUER a Jit
the tensor is symmetric wat.t,
Similarly if AUK = 15 did ec nalts J
symmetric wr. all pairs of contravarant indices oral pairs of
SKEW - SYMMETRIC TENSOR
A tensor is called skew ~ symmetric wrt, twoVECTOR AND TENSOR ANALYSIS
ijk ik
me Ad
For example, if A bie
the tensor is skew — symmetrie w.r. the contravariant indices
i and j. Similarly, if Ajgn = — AUX , the tensor is skew — symmetric wart. the covariant indices m
and n
Ifa tensor is skew — symmetric ws. all pairs of contravariant indices or all pars of covariant
indices , itis called skew — symmetric .
THEOREM (8.19): Show that every tensor can be expressed as the sum of two tensors , one of which
is symmetric and the other skew — symmetric in a pair of contravariant or
covariant indices .
PROOF: We prove this result for the contravariant tensor A‘). We have
eee Leatealy statically
= Busch
put BY = (a+ as!) = 5c al +A4) = Bil issymmenic,
and cl = 2¢A Ad! = 3 as! aH) = cil in skew — symmenic.
By similar reasoning the result is seen to be true for any tensor .
THEOREM (8.20): Prove that in an n— dimensional space , a symmetric second order tensor Ai
has atmost *2*1) gitterent components. OCR MON
Since Ajj isa covariant tensor of order 2, it hasGENERAL TENSORS
S12
(A pstnla Manes
TAny Abt An, An 7 Any
Since Ajj is symmetric, therefore A wz
‘Thus the number of independent components of A, |; with unequal indices Is
tensor Aj) has atmost ae 1) itterent components.
Ajj is a covariant tensor of order 2 It has the followifig * eomponents
we Ain
PROOF: Since
Au Ais
Ay Anw
Ant Anges.
Since A jj is skew ~symmettie ; we have
AUS Aji, i.) =1,2,3,..
Hence Ajj = —
or 2Ai;=.0 of Ai = 0. forall
=1,2,.
Thus Ay) = An =.= Ann =0
So the number of independent components on the leading diagonal is zero. Thus the number of
n-n
components of distinct indices is n?-n. But due to skew ~ symmetry , this number is reduced to z
Hence maximum number of independent components of Ajj is 0+
8.25 INVARIANCE OF SYMMETRIC AND eee ecy
PROPERTIES OF A TENSOR
THEOREM (8.22): If.a tensor is symmetric ( skew ~ symmetric )
two covariant indices in Sne coordinate
(skew ~ symmetric) w.r.t, these indices in
PROOF: Lawl te
(x59
at
xk
Ba: apBae EPR ac
qritaw te = aeVECTOR AND TENSOR ANALYsis
515
@
Gi)
Cy (iv)
From equations (ii), (ii) , and ™, we get
As z J x? axl
APBg:-302, = 2 oR? 41 ax! ont ax? ax)
7 qi ax" ax 8 xt e 39°
2x? axi ax
or 7 (A'Bjx-3C
Ox" gg ggt (ABU 3Cjx)
= 0 [using equation () }
or APBg, = 37,
which shows that the form of the tensor equation remains the same in all coordinate systems .
8.27 THE LINE ELEMENT AND METRIC TENSOR
‘We know that in a three ~ dimensional rectangular coordinate system (x,y,z), the square of
the line element ds? is given by
ds? = dx?+dy?#d77 @ :
Let us transform equation (i) in a general curvilinear coordinate system (u',u?,u?), in which the
position vector of the point is given by ¥ = 7 (u',u?,u’). ~*~
Ee eae
then dt = SFrau's Saute Saw? ‘
* |B yduleGadul+ asd? og ant
where = 2 eth tangent et tthe corns eres wis
varie = ee
= (a :du'+@,du?+a5du’). (Grange ju
- 3 cau" Badetaate By. Fal
eee Bravran'+as sea
ae aioe‘GENERAL TENSORS
is called the fundamental quadratic form or metric form . The quantities g ; are called
metric coefficients . The space in which ds? is given by equation (ii) is called three ~ dimensional
Euclidean space and is denoted by E?.
Ifthe curvilinear coordinate system is orthogonal , then Bi 70 for ij.
Generalizing this concept , the square of the line element ds® in an n ~ dimensional coordinate system
(x!',x?, ....,x") is given by the fundamental or metric form
non
ds?= YD aiydx'axd
i=t je
( , using the summation convention
ds? = gyjdx'de, (i,j =1,2,....0) Gi),
As this extension is due to Riemann , the equation (iii) is called Riemannian metric and any space in
which d s? is given by equation (ii) is called an n — dimensional Riemannian space and is denoted by
R" .-Note that the square of the line element ds? in equation (ii) isa scalar invariant .
In the particular case of unit tangent vectors , we have is Bonen i.
Thus equation (ii) takes the form E
ds? = (dx!)?+(dx?)?4....+(dx")? = dx'ax! (iv)
Equation i is aed an Euctalan metre and such pace i called an n~ dimensional Euclidean sp 4
wished E”. 2 U) opa ariett 20
METRIC TENSOR —
‘The quantities g ,; are the components of a covariant
fundamental tensor .
THEOREM (8.25): If ds? = g,, dx'dx! isan
tensor of rank 2.
PROO! Since ds? is an invariant,
adxPdx! = gidxlaxdVECTOR AND TENSOR ANALYSIS 517
NOTE: Since the metic tensor is symmetric ie. g,, = g;. therefore the number of independent
eee ay ean aca
THEOREM (8.26): Prove that for an orthogonal curvilinear coordinate system ,
Bu-8u- 0 By= en 9, 8,78, = 0.
PROOF: By definition
diy" Bik 0)
Thus 8, = By, = @1.2=|@,||a2|c0s90° = 0
Similarly, g,, = 8 = @2-a3=|@2||a3]cos90° = 0
and 8) = By = &s- 1 =| a3] |@,|cos90°= 0
8.28 METRIC TENSOR IN CYLINDERICAL AND SPHERICAL POLAR
COORDINATES
‘We know that in Cartesian coordinates the square of the differential of arc length s is given by
ds? = dx?+dy?+dz2? Ge
CYLINDERICAL POLAR COORDINATES
In cylinderical polar coordinates , this becomes
ds? = dr?+17do?+dz?
Ix'=r, x7=0, az
then ds? = (dx!)?+17(dx?)?+(dx?)? @
Also ds? = gijdx'dx!
= By (dx')? +8, (dx?)?+g,,(dx?)? +g, dx'dx?
+B ydx'dx?+g,,dx7dx'+g,,dx?7dx? ‘ oe
+85dx?dx!+g,.dx?dx? Gi
Comparing equstions (fi) and (ili) , we get 3 .
Bun pool Fy fey = 1 =
Ba 781% By=Bn=~0, By = By = 0.
In matrix form , the metric tensor can be written
(ei e629)
ee eee
SPHERICAL POLAR COORDINATES _
ds? = dr2+r7d674r7sin270d6?518
ae 3 =p, then
‘i Be ca Mere ax’)? (iv)
ast = (axt)? 4282?) +r?sin?6 (dx
‘Comparing equations sot we get
eit ee | By 7 esin?@,
een
a= en 7 Or oe Gta s!
st tie or ne merits cnbe wen
Bu Sn Sa Nene et
wo hee oS
By 82 Bx. o 0 r?sin?@.
In general for orthogonal coordinates, 8 =0 for i#j.
Sy Bn Ss
: vr be. f +
EXAMPLE (23): (i) Express the determinant gle | ea ee,
gy Bn Bx ae
in terms ofthe elements in the second row and their corresponding cofactors -
Gi) Show that g,, Gl = g where c'l is the cofactor of g,, in g and
where the summation is over j only.
SOLUTION:
( The cofactor of gij is the determinant obtained from g vena
which gj appears and (i) associating the sign (— 1)!
‘ i.
Cofactor of g,, = (-1)?*"
conceinig =(-1)%3| Su 82
ae fal
Denote these cofactors by G", G™ and
determinants
£6" +8074," = g