Upload
pticicaaa
View
218
Download
0
Embed Size (px)
Citation preview
8/14/2019 Introduction to Groups
1/164
Introduction
toGroups, Invariantsand
ParticlesFrank W. K. Firk, Emeritus Professor of Physics, Yale University
2000
8/14/2019 Introduction to Groups
2/164
CONTENTS 2
1. INTRODUCTION 3
2. GALOIS GROUPS 8
3. SOME ALGEBRAIC INVARIANTS 20
4. SOME INVARIANTS IN PHYSICS 29
5. GROUPS CONCRETE AND ABSTRACT 43
6. LIES DIFFERENTIAL EQUATION, INFINITESIMAL
ROTATIONS, AND ANGULAR MOMENTUM 57
7. LIES CONTINUOUS TRANSFORMATION GROUPS 68
8. PROPERTIES OF n-VARIABLE r-PARAMETER
LIE GROUPS 78
9. MATRIX REPRESENTATIONS OF GROUPS 83
10. SOME LIE GROUPS OF TRANSFORMATIONS 94
11. THE GROUP STRUCTURE OF LORENTZ
TRANSFORMATIONS 107
12. ISOSPIN 115
13. LIE GROUPS AND THE STRUCTURE OF MATTER 128
14. LIE GROUPS AND THE CONSERVATION LAWS
OF THE PHYSICAL UNIVERSE 159
15. BIBLIOGRAPHY 163
8/14/2019 Introduction to Groups
3/164
1 3
INT RODUC TION
The noti on of geom etric al symmetr y in Art a nd in Natu re is a
fam iliar one. In Moder n Phy sics, this noti on ha s evo lved to in clude
sym metri es of an a bstra ct ki nd. These new symme tries play an e ssential
par t in the t heori es of the micro struc ture of ma tter. The basic sym metri es
fou nd in Natu re se em to orig inate in t he ma thema tical stru cture of t he la ws
the mselvesla ws th at go vern the m otion s of the g alaxi es on the onehan d and the motio ns of quar ks in nucl eons on th e oth er.
In the N ewton ian e ra, t he la ws of Natu re we re de duced from a sm all
num ber o f imperfec t obs ervat ions by a small numb er of reno wned
sci entists an d mat hemat ician s. I t was not until the Einst einia n era ,
how ever, that the signi fican ce of the symme tries asso ciated wit h the laws
was full y app reciated. The disco very of sp ace-time symmet ries has led to
the wide ly-he ld be lief that the l aws o f Nat ure c an be deri ved f rom
sym metry , or invar iance , pri ncipl es. Our i ncomp lete knowledge of th e
fun damen tal intera ctions mea ns th at we are not y et in a po sitio n to confi rm
thi s belief. We t heref ore u se ar gumen ts ba sed o n emp irica lly established
laws and rest ricte d sym metry prin ciple s to guide us i n our sear ch fo r the
fun damen tal symmet ries. Fre quent ly, i t is impor tant to un derst and w hy
the symm etry of a system is obser ved t o be broke n.
8/14/2019 Introduction to Groups
4/164
In Geome try, an ob ject with a def inite shap e, si ze, location, a nd 4
ori entat ion c onstitutes a st ate w hose symme try p roper ties, or i nvari ants,
are to b e stu died. Any tran sform ation that leav es th e sta te un chang ed in
for m is calle d a s ymmet ry tr ansformati on. The g reate r the numb er of
sym metry tran sform ations tha t a s tate can u nderg o, th e hig her i ts
sym metry . If the numbe r of conditions that define th e sta te is redu ced
the n the symm etry of th e sta te is incr eased. Fo r exa mple, an o bject
cha racte rized by o blateness alone is s ymmet ric u nder all transformat ions
exc ept a chan ge of shap e.
In descr ibing the symme try o f a sta te of the most gener al kind (n ot
simply g eomet ric), the algebraic struc ture of th e set of s ymmet ry op erato rs
mus t be given; it is no t suf ficie nt to give the numbe r of opera tions, and
not hing else. The law of c ombin ation of the opera tors must be st ated. It
is the alg ebrai c gro up th at fu lly charac terizes th e sym metry of t he ge neral
sta te.
The The ory o f Gro ups ca me ab out u nexpe ctedly. G alois show ed
tha t an equat ion o f deg ree n , whe re n is an inte ger g reate r tha n or equal to
fiv e can not, in ge neral, be solve d by algebraic means . In the cours e of this
gre at wo rk, h e dev eloped the idea s of Lagra nge, Ruffi ni, and Ab el an d
int roduc ed th e con cept of a gro up. Galoi s discussed the func tional
rel ationships amon g the root s of an eq uation, an d sho wed t hat t he
rel ationships have symm etrie s associated wi th th em un der p ermut ations of
the root s.
8/14/2019 Introduction to Groups
5/164
The oper ators that tran sform one funct ional rela tions hip i nto 5
ano ther are e lemen ts of a se t tha t is chara cteri stic of th e equ ation ; the set
of opera tors is ca lled the G alois grou p of the e quati on.
In the 1 850s , Cay ley showed that ever y finite g roup is is omorp hic
to a cer tain permu tatio n gro up. The g eomet rical symm etries of cryst als
are desc ribed in t erms of fi nite group s. T hese symme tries are discu ssed in
man y sta ndard work s (see bib liography) and there fore, they will not be
discussed in this book.
In the b rief perio d bet ween 1924 and 1 928, Quant um Me chanics
was deve loped. Alm ost i mmedi ately, it was r ecogn ized by Weyl, and by
Wig ner, that certa in pa rts o f Gro up Th eory could be u sed a s a p owerf ul
ana lytic al to ol in Quan tum P hysic s. T heir ideas have been deve loped over
the deca des in man y are as th at ra nge f rom t he Th eory of So lids to Pa rticl e
Phy sics.
The esse ntial role play ed by grou ps th at ar e cha racte rized by
par amete rs th at va ry co ntinu ously in a give n ran ge wa s first em phasi zed
by Wigne r. T hese group s are know n as Lie Gro ups. They have becom e
increasingly impor tant in ma ny br anche s of conte mpora ry ph ysics ,
par ticul arly Nuclear an d Par ticle Phys ics. Fift y yea rs af ter G alois had
int roduc ed th e con cept of a group in t he Th eory of Eq uations, L ie
int roduc ed th e con cept of a continuous tran sform ation grou p in the T heory
of Diffe renti al Eq uations. Lies theo ry un ified many of t he di sconn ected
met hods of so lving diff erent ial e quations t hat h ad ev olved over a pe riod of
8/14/2019 Introduction to Groups
6/164
two hund red y ears. Inf inite simal unit ary t ransformat ions play a key role in 6
discussi ons o f the fund ament al co nserv ation laws of P hysic s.
In Class ical Dynam ics, the i nvari ance of th e equ ations of motio n of a
par ticle , or system of parti cles, unde r the Gali lean trans forma tion is a basic
par t of every day r elati vity. The sear ch fo r the tran sform ation that leav es
Max well s equ ations of Elect romag netism unc hange d in form (inva riant )
und er a linea r tra nsfor mation of the s pace-time coord inates, le d to the
discover y of the L orent z tra nsfor matio n. T he fu ndame ntal impor tance of
thi s tra nsfor matio n, an d its rela ted i nvari ants, cann ot be over state d.
This int roduc tion to Gr oup T heory , wit h its emph asis on Li e Gro ups
and thei r app licat ion t o the stud y of symme tries of t he fu ndame ntal
con stituents of ma tter, has its o rigin in a one- semester c ourse that I ta ught
at Yale University for more than ten y ears. The cour se wa s dev eloped for
Sen iors, and advan ced J unior s, ma joring in the P hysic al Sc ience s. T he
stu dents had gener ally completed the c ore c ourse s for thei r maj ors, and
had take n int ermed iate level cour ses in Lin ear A lgebr a, Re al an d Com plex
Ana lysis , Ord inary Line ar Di fferential Equa tions, and some of t he Sp ecial
Fun ctions of Physics. Group Theo ry wa s not a ma thema tical requ ireme nt
for a de gree in th e Phy sical Scie nces. The majo rity of ex istin g
und ergra duate text books on G roup Theor y and its appli cations in Phys ics
ten d to be ei ther highl y qua litat ive o r hig hly m athem atical. The pu rpose of
thi s int roduc tion is to stee r a m iddle cour se th at pr ovides the stud ent w ith
a s ound mathe matical ba sis f or st udying the symm etry prope rties of t he
8/14/2019 Introduction to Groups
7/164
fun damen tal p artic les. It i s not gene rally appr eciated by Phys icist s tha t 7
con tinuo us tr ansformati on gr oups originated in t he Th eory of Differential
Equ ations. T he in finit esima l gen erato rs of Lie Group s the refor e have
forms that involve differential operators and their commutators, and these
operators and their algebraic properties have found, and continue to find, a
natural place in the development of Quantum Physics.
Guilford, CT.
June, 2000
8/14/2019 Introduction to Groups
8/164
2 8
GALOIS GROUPS
In the early 19th - century, Abel proved that it is not possible to solve the
general polynomial equation of degree greater than four by algebraic means.
He attempted to characterize all equations that can be solved by radicals.
Abel did not solve this fundamental problem. The problem was taken up and
solved by one of the greatest innovators in Mathematics, namely, Galois.
2.1. Solving cubic equations
The main ideas of the Galois procedure in the Theory of Equations,
and their relationship to later developments in Mathematics and Physics, can
be introduced in a plausible way by considering the standard problem of
solving a cubic equation.
Consider solutions of the general cubic equation
Ax3 + 3Bx2 + 3Cx + D = 0, where A - D are rational constants.
If the substitution y = Ax + B is made, the equation becomes
y3 + 3Hy + G = 0
where
H = AC - B2
and
G = A2D - 3ABC + 2B3.
8/14/2019 Introduction to Groups
9/164
The cubic has three real roots if G2 + 4H3 < 0 and two imaginary roots if G2 9
+ 4H3 > 0. (See any standard work on the Theory of Equations).
If all the roots are real, a trigonometrical method can be used to obtain
the solutions, as follows:
the Fourier series of cos3u is
cos3u = (3/4)cosu + (1/4)cos3u.
Putting
y = scosu in the equation y3 + 3Hy + G = 0
(s > 0),
gives
cos3u + (3H/s2)cosu + G/s3 = 0.
Comparing the Fourier series with this equation leads to
s = 2 (-H)
and
cos3u = -4G/s3.
If v is any value of u satisfying cos3u = -4G/s3, the general solution is
3u = 2n 3v, where n is an integer.
Three different values of cosu are given by
u = v, and 2/3 v.
The three solutions of the given cubic equation are then
scosv, and scos(2/3 v).
Consider solutions of the equation
8/14/2019 Introduction to Groups
10/164
x3 - 3x + 1 = 0. 10
In this case,
H = -1 and G2 + 4H3 = -3.
All the roots are therefore real, and they are given by solving
cos3u = -4G/s3 = -4(1/8) = -1/2
or,
3u = cos-1(-1/2).
The values of u are therefore 2/9, 4/9, and 8/9, and the roots are
x1 = 2cos(2/9), x2 = 2cos(4/9), and x
3 = 2cos(8/9).
2.2. Symmetries of the roots
The roots x1, x2, and x3 exhibit a simple pattern. Relationships among
them can be readily found by writing them in the complex form 2cos = e i +
e-i where = 2/9 so that
x1 = ei + e -i ,
x2 = e2i + e-2i ,
and
x3 = e4i + e-4i .
Squaring these values gives
x12 = x2 + 2,
x22 = x3 + 2,
and
8/14/2019 Introduction to Groups
11/164
x32 = x1 + 2. 11
The relationships among the roots have the functional form f(x1,x2,x3) = 0.
Other relationships exist; for example, by considering the sum of the roots we
find
x1 + x22 + x2 - 2 = 0
x2 + x32 + x3 - 2 = 0,
and
x3 + x12 + x1 - 2 = 0.
Transformations from one root to another can be made by doubling-the-
angle, .
The functional relationships among the roots have an algebraic
symmetry associated with them under interchanges (substitutions) of the
roots. If is the operator that changes f(x1,x2,x3) into f(x2,x3,x1) then
f(x1,x2,x3) f(x2,x3,x1),
2f(x1,x2,x3) f(x3,x1,x2),
and
3f(x1,x2,x3) f(x1,x2,x3).
The operator 3 = I, is the identity.
In the present case,
(x12 - x2 - 2) = (x2
2 - x3 - 2) = 0,
8/14/2019 Introduction to Groups
12/164
and
2(x12 - x2 - 2) = (x3
2 - x1 - 2) = 0. 12
2.3. The Galois group of an equation.
The set of operators {I, , 2} introduced above, is called the Galois
group of the equation x3 - 3x + 1 = 0. (It will be shown later that it is
isomorphic to the cyclic group, C3).
The elements of a Galois group are operators that interchange the
roots of an equation in such a way that the transformed functional
relationships are true relationships. For example, if the equation
x1 + x22 + x2 - 2 = 0
is valid, then so is
(x1 + x22 + x2 - 2 ) = x2 + x3
2 + x3 - 2 = 0.
True functional relationships are polynomials with rational coefficients.
2.4. Algebraic fields
We now consider the Galois procedure in a more general way. An
algebraic solution of the general nth - degree polynomial
aoxn + a1x
n-1 + ... an = 0
is given in terms of the coefficients ai using a finite number of operations (+,-
,,,). The term "solution by radicals" is sometimes used because the
operation of extracting a square root is included in the process. If an infinite
number of operations is allowed, solutions of the general polynomial can be
8/14/2019 Introduction to Groups
13/164
obtained using transcendental functions. The coefficients ai necessarily belong 13
to afieldwhich is closed under the rational operations. If the field is the set
of rational numbers, Q, we need to know whether or not the solutions of a
given equation belong to Q. For example, if
x2 - 3 = 0
we see that the coefficient -3 belongs to Q, whereas the roots of the equation,
xi = 3, do not. It is therefore necessary to extendQ to Q', (say) by
adjoining numbers of the form a3 to Q, where a is in Q.
In discussing the cubic equation x3
- 3x + 1 = 0 in 2.2, we found
certain functions of the roots f(x1,x2,x3) = 0 that are symmetric under
permutations of the roots. The symmetry operators formed the Galois group
of the equation.
For a general polynomial:
xn + a1xn-1 + a2x
n-2 + .. an = 0,
functional relations of the roots are given in terms of the coefficients in the
standard way
x1 + x2 + x3 .. .. + xn = -a1
x1x2 + x1x3 + .. x2x3 + x2x4 + ..+ xn-1xn = a2
x1x2x3 + x2x3x4 + .. .. + xn-2xn-1xn = -a3
.
.
x1x2x3 .. .. xn-1xn = an.
8/14/2019 Introduction to Groups
14/164
8/14/2019 Introduction to Groups
15/164
It can be solved by writing x3 = y, y2 = 3 or 15
x = (3)1/3.
To solve the equation, it is necessary to calculate square and cube roots
not sixth roots. The equation x6 = 3 is said to be compound (it can be
broken down into simpler equations), whereas x2 = 3 is said to be atomic.
The atomic properties of the Galois group of an equation reveal
the atomic nature of the equation, itself. (In Chapter 5, it will be seen that a
group is atomic ("simple") if it contains no proper invariant subgroups).
The determination of the Galois groups associated with an arbitrary
polynomial with unknown roots is far from straightforward. We can gain
some insight into the Galois method, however, by studying the group
structure of the quartic
x4 + a2x2 + a4 = 0
with known roots
x1 = ((-a2 + )/2)1/2 , x2 = -x1,
x3 = ((-a2 - )/2)1/2 , x4 = -x3,
where
= (a 22 - 4a4)
1/2.
The field F of the quartic equation contains the rationals Q, and the
rational expressions formed from the coefficients a2 and a4.
The relations
x1 + x2 = x3 + x4 = 0
8/14/2019 Introduction to Groups
16/164
are in the field F. 16
Only eight of the 4! possible permutations of the roots leave these
relations invariant in F; they are
x1 x2 x3 x4 x1 x2 x3 x4 x1 x2 x3 x4{ P1 = , P2 = , P3 = ,
x1 x2 x3 x4 x1 x2 x4 x3 x2 x1 x3 x4x1 x2 x3 x4 x1 x2 x3 x4 x1 x2 x3 x4
P4 = , P5 = , P6 =x2 x1 x4 x3 x3 x4 x1 x2 x3 x4 x2 x1x1 x2 x3 x4 x1 x2 x3 x4
P7 = , P8 = }.x4 x3 x1 x2 x4 x3 x2 x1
The set {P1,...P8} is the Galois group of the quartic in F. It is a subgroup of
the full set of twentyfour permutations. We can form an infinite number of
true relations among the roots in F. If we extend the field F by adjoining
irrational expressions of the coefficients, other true relations among the roots
can be formed in the extended field, F'. Consider, for example, the extended
field formed by adjoining (= (a22 - 4a4)) to F so that the relation
x12 - x3
2 = is in F'.
We have met the relations
x1 = -x2 and x3 = -x4
so that
x12 = x2
2 and x32 = x4
2.
Another relation in F' is therefore
8/14/2019 Introduction to Groups
17/164
x22 - x4
2 = . 17
The permutations that leave these relations true in F' are then
{P1, P2, P3, P4}.
This set is the Galois group of the quartic in F'. It is a subgroup of the set
{P1,...P8}.
If we extend the field F' by adjoining the irrational expression
((-a2 - )/2)1/2 to form the field F'' then the relation
x3 - x4 = 2((-a2 - )/2)1/2
is in F''.
This relation is invariant under the two permutations
{P1, P3}.
This set is the Galois group of the quartic in F''. It is a subgroup of the set
{P1, P2, P3, P4}.
If, finally, we extend the field F'' by adjoining the irrational
((-a2 + )/2)1/2 to form the field F''' then the relation
x1 - x2 = 2((-a2 - )/2)1/2 is in F'''.
This relation is invariant under the identity transformation, P1 , alone; it is
the Galois group of the quartic in F''.
The full group, and the subgroups, associated with the quartic equation
are of order 24, 8, 4, 2, and 1. (The order of a group is the number of
distinct elements that it contains). In 5.4, we shall prove that the order of a
subgroup is always an integral divisor of the order of the full group. The
8/14/2019 Introduction to Groups
18/164
order of the full group divided by the order of a subgroup is called the index 18
of the subgroup.
Galois introduced the idea of a normal or invariant subgroup: if H is a
normal subgroup of G then
HG - GH = [H,G] = 0.
(H commutes with every element of G, see 5.5).
Normal subgroups are also called either invariant or self-conjugate subgroups.
A normal subgroup H is maximal if no other subgroup of G contains H.
2.5. Solvability of polynomial equations
Galois defined the group of a given polynomial equation to be either
the symmetric group, Sn, or a subgroup of Sn, (see 5.6). The Galois method
therefore involves the following steps:
1. The determination of the Galois group, Ga, of the equation.
2. The choice of a maximal subgroup of Hmax(1). In the above case, {P1, ...P8}
is a maximal subgroup of Ga = S4.
3. The choice of a maximal subgroup of Hmax(1) from step 2.
In the above case, {P1,..P4} = Hmax(2) is a maximal subgroup of Hmax(1).
The process is continued until Hmax = {P1} = {I}.
The groups Ga, Hmax(1), ..,Hmax(k) = I, form a composition series. The
composition indices are given by the ratios of the successive orders of the
groups:
8/14/2019 Introduction to Groups
19/164
gn/h(1), h(1)/h(2), ...h(k-1) /1.
The composition indices of the symmetric groups Sn for n = 2 to 7 are found
to be:
n Composition Indices
2 2
3 2, 3
4 2, 3, 2, 2
5 2, 60
6 2, 360
7 2, 2520
We shall state, without proof, Galois' theorem:
A polynomial equation can be solved algebraically if and only if its
group is solvable.
Galois defined a solvable group as one in which the composition indices are
all prime numbers. Furthermore, he showed that if n > 4, the sequence of
maximal normal subgroups is Sn, An, I, where An is the Alternating Group
with (n!)/2 elements. The composition indices are then 2 and (n!)/2. For n >
4, however, (n!)/2 is not prime, therefore the groups Sn are not solvable for n
> 4. Using Galois' Theorem, we see that it is therefore not possible to solve,
algebraically, a general polynomial equation of degree n > 4.
8/14/2019 Introduction to Groups
20/164
3 20
SOME ALGEBRAIC INVARIANTS
Although algebraic invariants first appeared in the works of Lagrange and
Gauss in connection with the Theory of Numbers, the study of algebraic
invariants as an independent branch of Mathematics did not begin until the
work of Boole in 1841. Before discussing this work, it will be convenient to
introduce matrix versions of real bilinear forms, B, defined by
m nB = aijxiyj
I=1 j=1
where
x = [x1,x2,...xm], an m-vector,
y = [y1,y2,...yn], an n-vector,
and aij are real coefficients. The square brackets denote a
column vector.
In matrix notation, the bilinear form is
B = xTAy
where
a11 . . . a1n
. . . .
A = . . . . .
. . . .
am1. . . amn
The scalar product of two n-vectors is seen to be a special case of a
bilinear form in which A = I.
8/14/2019 Introduction to Groups
21/164
Ifx = y, the bilinear form becomes a quadratic form, Q: 21
Q = xTAx.
3.1. Invariants of binary quadratic forms
Boole began by considering the properties of the binary
quadratic form
Q(x,y) = ax2 + 2hxy + by2
under a linear transformation of the coordinates
x' = Mx
where
x = [x,y],
x' = [x',y'],
and
| i j | M = .
| k l |
The matrix M transforms an orthogonal coordinate system into an
oblique coordinate system in which the new x'- axis has a slope (k/i), and the
new y'- axis has a slope (l/j), as shown:
8/14/2019 Introduction to Groups
22/164
y 22
y[i+j,k+l]
[j,l]x'
[0,1] [1,1]x
[i,k]
[0,0] [1,0] x
Fig. 1. The transformation of a unit square underM.
The transformation is linear, therefore the new function Q'(x',y') is a
binary quadratic:
Q'(x',y') = a'x'2 + 2h'x'y' + b'y'2.
The original function can be written
Q(x,y) = xTDx
where
| a h | D = ,
| h b |
and the determinant ofD is
detD = ab - h2, called the discriminant of Q.
The transformed function can be written
8/14/2019 Introduction to Groups
23/164
Q'(x',y') = x'TD'x' 23
where
a' h'
D' = ,h' b'
and
detD' = a'b' - h'2, the discriminant of Q'.
Now,
Q'(x',y') = (Mx)TD'Mx
= xTMTD'Mx
and this is equal to Q(x,y) if
MTD'M = D.
The invariance of the form Q(x,y) under the coordinate transformation M
therefore leads to the relation
(detM)2
detD' = detD
because
detMT = detM.
The explicit form of this equation involving determinants is
(il - jk)2(a'b' - h'2) = (ab - h2).
The discriminant (ab - h2) of Q is said to be an invariant
of the transformation because it is equal to the discriminant (a'b' - h'2) of Q',
apart from a factor (il - jk)2 that depends on the transformation itself, and not
on the arguments a,b,h of the function Q.
8/14/2019 Introduction to Groups
24/164
3.2. General algebraic invariants 24
The study of general algebraic invariants is an important branch of
Mathematics.
A binary form in two variables is
f(x1,x2) = aox1n + a1x1
n-1x2 + ...anx2n
= aix1n-ix2
i
If there are three or four variables, we speak of ternary forms or quaternary
forms.
A binary form is transformed under the linear transformation M as
follows
f(x1,x2) => f'(x1',x2') = ao'x1'n + a1'x1'
n-1x2' + ..
The coefficients
ao, a1, a2,.. ao', a1', a2' ..
and the roots of the equation
f(x1,x2) = 0
differ from the roots of the equation
f'(x1',x2') = 0.
Any function I(ao,a1,...an) of the coefficients of f that satisfies
rwI(ao',a1',...an') = I(ao,a1,...an)
is said to be an invariantof f if the quantity r depends only on the
transformation matrix M, and not on the coefficients ai of the function being
transformed. The degree of the invariant is the degree of the coefficients, and
8/14/2019 Introduction to Groups
25/164
the exponent w is called the weight. In the example discussed above, the 25
degree is two, and the weight is two.
Any function, C, of the coefficients andthe variables of a form f that is
invariant under the transformation M, except for a multiplicative factor that is
a power of the discriminant ofM, is said to be a covariant of f. For binary
forms, C therefore satisfies
rwC(ao',a1',...an'; x1',x2') = C(ao,a1,...an; x1,x2).
It is found that the Jacobian of two binary quadratic forms, f(x1,x2) and
g(x1,x2), namely the determinant
f/x1 f/x2
g/x1 g/x2
where f/x1 is the partial derivative of f with respect to x1 etc., is a
simultaneous covariant of weight one of the two forms.
The determinant
2f/x12 2f/x1x2
, 2g/x2x1
2g/x22
called the Hessian of the binary form f, is found to be a covariant of weight
two. A full discussion of the general problem of algebraic invariants is outside
8/14/2019 Introduction to Groups
26/164
the scope of this book. The following example will, however, illustrate the 26
method of finding an invariant in a particular case.
Example:
To show that
(aoa2 - a12)(a1a3 - a2
2) - (aoa3 - a1a2)2/4
is an invariant of the binary cubic
f(x,y) = aox3 + 3a1x
2y + 3a2xy2 + a3y
3
under a linear transformation of the coordinates.
The cubic may be written
f(x,y) = (aox2+2a1xy+a2y
2)x + (a1x2+2a2xy+a3y
2)y
= xTDx
where
x = [x,y],
and
aox + a1y a1x + a2y D = .
a1x + a2y a2x + a3y
Let x be transformed to x': x' = Mx, where
i j M =
k l
8/14/2019 Introduction to Groups
27/164
then 27
f(x,y) = f'(x',y')
if
D = MTD'M.
Taking determinants, we obtain
detD = (detM)2detD',
an invariant of f(x,y) under the transformation M.
In this case, D is a function of x and y. To emphasize this point, put
detD = (x,y)and
detD'= '(x',y')
so that
(x,y) = (detM)2'(x',y'
= (aox + a1y)(a2x + a3y) - (a1x + a2y)2
= (aoa2 - a12)x2 + (aoa3 - a1a2)xy + (a1a3 - a2
2)y2
= xTEx,
where
(aoa2 - a12 ) (aoa3 - a1a2)/2
E = .
(aoa3 - a1a2)/2 (a1a3 - a22 )
Also, we have
'(x',y') = x'TE'x'
8/14/2019 Introduction to Groups
28/164
= xTMTE'Mx 28
therefore
xTEx = (detM)2xTMTE'Mx
so that
E = (detM)2MTE'M.
Taking determinants, we obtain
detE = (detM)4detE'
= (aoa2 - a12)(a1a3 - a2
2) - (aoa3 -a1a2)2/4
= invariant of the binary cubic f(x,y) under the transformation
x' = Mx.
8/14/2019 Introduction to Groups
29/164
4 29
SOM E INV ARIAN TS IN PHYS ICS
4.1 . Gal ilean inva rianc e.
Eve nts o f fin ite e xtens ion a nd du ratio n are part of t he ph ysica l
wor ld. It will be conv enient to intro duce the n otion ofide al ev ents th at
hav e nei ther exten sion nor d urati on. Ideal even ts ma y be repre sente d as
mat hemat ical point s in a spa ce-ti me ge ometr y. A part icula r eve nt, E, i s
des cribed by the f our c ompon ents [t,x, y,z] where t is the time of th e eve nt,
and x,y, z, ar e its thre e spa tial coord inates. The tim e and spac e coo rdina tes
are refe rred to ar bitra rily chose n ori gins. The spat ial mesh n eed n ot be
Car tesia n.
Let an e vent E[t, x], r ecord ed by an o bserv er O at th e ori gin o f an x-
axis, be reco rded as th e eve nt E'[t ',x'] by a seco nd ob serve r O', movi ng at
con stant spee d V a long the x -axis . We supp ose t hat t heir clocks are
syn chron ized at t = t' = 0 w hen t hey c oinci de at a co mmon origin, x = x' =
0.
At time t, we writ e the plau sible equa tions
t' = t
and
x' = x - V t,
whe re Vt is t he di stanc e tra velle d by O' in a ti me t. The se eq uations ca n
be writt en
E' = GE
8/14/2019 Introduction to Groups
30/164
whe re 30
1 0 G = .
-V 1
G is the opera tor o f the Gali lean trans forma tion.
The inve rse e quations a re
t = t'
and
x = x ' + V t'
or
E = G-1E'
whe re G-1 is the inver se Ga lilea n ope rator . (I t und oes t he ef fect ofG).
If we mu ltipl y t a nd t' by t he co nstan ts k and k ', re spect ively , whe re
k a nd k' have dime nsions of velocity t hen a ll terms h ave d imensions of
length.
In space-spac e, we have the Pytha gorea n for m x 2 + y2 = r2, a n
inv ariant und er ro tatio ns. We ar e the refor e led to a sk th e que stion: is
(kt )2 + x2 in variant un der t he op erato r G in spac e-tim e? C alcul ation give s
(kt)2 + x2 = (k't ')2 + x'2 + 2Vx't ' + V 2t'2
= (k't' )2 + x'2 o nly if V = 0.
We see, there fore, that Gali lean space-time is n ot Py thago rean in its
algebraic for m. W e not e, ho wever , the key role played by acc elera tion in
Galilean -Newtonian phys ics:
8/14/2019 Introduction to Groups
31/164
The velo citie s of the e vents acco rding to O and O' ar e obt ained by 31
dif feren tiati ng th e equ ation x' = -Vt + x w ith r espect to time, giving
v' = -V + v,
a p lausi ble r esult, bas ed up on ou r exp erien ce.
Dif feren tiati ng v' with resp ect t o tim e giv es
d v'/dt ' = a ' = d v/dt = a
whe re a and a ' are the accel erati ons i n the two frame s of refer ence. The
classica l acceleration is inv arian tun der t he Ga lilea n tra nsfor matio n. I f the
rel ationship v' = v V is used to desc ribe the m otion of a puls e of light ,mov ing in emp ty sp ace a t v = c 3 x 108 m/ s, it does not fit t he fa cts. All
stu dies of ve ry hi gh sp eed p artic les t hat e mit e lectr omagn etic radiation
sho w tha t v' = c f or all va lues of th e rel ative spee d, V.
4.2 . Lor entz invar iance and Einst ein's spac e-tim e
sym metry .
It was E instein, above all others , who adv anced our under stand ing o f
the true natu re of spac e-tim e and rela tive motio n. W e sha ll se e tha t he
mad e use of a symm etry argum ent t o find the chan ges t hat m ust b e mad e
to the G alilean tr ansformati on if it i s to accou nt fo r the rela tive motio n of
rap idly movin g obj ects and o f bea ms of light. H e rec ognized an
inconsis tency in t he Ga lilea n-Newtonian equ ations, ba sed a s the y are , on
eve ryday expe rience. H ere, we sh all restrict th e discussi on to non-
acc elerating, or s o cal led inerti al, f rames
8/14/2019 Introduction to Groups
32/164
We have seen that the c lassi cal equations r elati ng th e eve nts E an d 32
E' a re E' = GE, a nd th e inv erse E = G-1E'
whe re
1 0 1 0 G = an d G-1 = . V 1 V 1
The se eq uations ar e con necte d by the s ubstitutio n V V; this is an
algebraic sta temen t of the N ewton ian p rinci ple o f rel ativi ty. Einst ein
incorpor ated this princ iple in his the ory. He a lso r etained th e lin earit y of
the classical equa tions in t he ab sence of a ny ev idence to the c ontra ry.
(Eq uispa ced i nterv als of tim e and dist ance in on e ine rtial fram e rem ain
equ ispac ed in any other iner tial frame ). H e the refor e sym metri zedth e
spa ce-ti me eq uations as foll ows:
t' 1 -V t= .
x' -V 1 x
Not e, ho wever , the inco nsist ency in th e dim ensio ns of the time- equat ion
tha t has now been intro duced :
t' = t Vx .
The term Vx h as dimensions o f [L] 2/[T], an d not [T]. Thi s can be
cor recte d by intro ducing the inva riant spee d of light , ca postu late in
Ein stein's th eory that is consist ent w ith e xperi ment:
ct' = ct Vx /c
so that all terms now h ave d imensions of length.
8/14/2019 Introduction to Groups
33/164
Ein stein went furt her, and i ntrod uced a dim ensio nless quan tity 33
instead of th e sca ling facto r of unity that appe ars i n the Gali lean equat ions
of space-time . Th is fa ctor must be co nsist ent w ith a ll observa tions. Th e
equ ations the n bec ome
ct ' = ct x
x ' = -ct + x, where =V/ c.
The se ca n be writt en
E' = LE,
whe re
- L = , and E = [ct,x ]
-
L is the opera tor o f the Lore ntz t ransformat ion.
The inve rse e quation is
E = L-1E'
whe re
L-1 = .
This is the i nvers e Lor entz trans forma tion, obta ined from L by chan ging
(o r ,V V); it h as th e effect o f und oing the t ransf ormat ion L.
We can t heref ore w rite
LL-1 = I
or
8/14/2019 Introduction to Groups
34/164
- 1 0 34= .
- 0 1
Equ ating elem ents gives
222 = 1
the refor e,
= 1/(1 2) ( taking the posi tive root) .
4.3 . The inva riant inte rval.
Pre viously, i t was show n tha t the spac e-tim e of Galileo an d New ton
is not P ythag orean in f orm. We n ow as k the ques tion: is E insteinian spac e-
tim e Pyt hagor ean i n for m? D irect calculati on le ads t o
( ct)2 + (x)2 = 2(1 + 2)(c t')2 + 42x'c t'
+2(1 + 2)x'2
(c t')2 + (x')2 if > 0.
Not e, ho wever , tha t the dif feren ce of squa res is an
inv ariant und er L:
(ct)2 (x )2 = (ct')2 (x ')2
bec ause
2(1 2) = 1.
Spa ce-ti me is said to b e pse udo-E uclid ean.
The nega tive sign that chara cteri zes L orent z inv ariance ca n be
included in t he th eory in a gener al wa y as follo ws.
We intro duce two k inds of 4- vecto rs
8/14/2019 Introduction to Groups
35/164
x = [x0, x1, x2, x3], a con trava riantve ctor, 35
and
x = [x0, x1, x2, x3], a cov arian tve ctor, wher e
x = [x0,-x1,-x2,-x3].
The scalar pr oduct of t he ve ctors is d efine d as
x Tx = (x0, x1, x2, x3)[x0,-x1,-x2,-x3]
= (x 0)2 (( x1)2 + (x2)2 + (x3)2)
The even t 4-v ector is
E
= [ct, x, y, z] a nd th e cov ariant for m is
E = [ct,- x,-y, -z]
so that the L orent z inv arian t sca lar p roduc t is
ETE = (ct)2 (x2 + y2 + z2).
The 4-ve ctor x tr ansforms a s follows:
x' = Lx
whe re
- 0 0- 0 0
L = .0 0 1 00 0 0 1
This is the o perat or of the Loren tz tr ansformati on if the motio n of O' is
along th e x-a xis o f O's fram e of refer ence.
Imp ortan t con seque nces of th e Lor entz trans forma tion are t hat
int ervals of time measu red i n two diff erent iner tial frame s are not the s ame
but are relat ed by the equat ion
8/14/2019 Introduction to Groups
36/164
t' = t 36
whe re t i s an inter val m easur ed on a cl ock a t res t in O's f rame, and
distance s are give n by
l' = l/
whe re l is a l ength meas ured on a ruler at r est i n O's fram e.
4.4 . The ener gy-mo mentu m inv arian t.
A d iffer ential time int erval, dt, cann ot be used in a Lore ntz-i nvari ant
way in k inematics. We must use t he pr oper time diffe renti al interva l, d,
def ined by
( cdt)2 - dx2 = (cdt' )2 - dx'2 (c d)2.
The Newt onian 3-ve locit y is
vN = [dx/d t, dy /dt, dz/dt],
and this must be r eplac ed by the 4-vel ocity
V = [d(ct )/d, d x/d, d y/d, d z/d]
= [ d(ct)/dt, dx/dt , dy/ dt, d z/dt] dt/d
= [c,vN] .
The scalar pr oduct is t hen
VV = (c)2 - (vN)
2
= (c)2(1 - (vN/c)2)
= c2.
(In form ing t he sc alar produ ct, t he tr anspo se is unde rstoo d).
The magn itude of t he 4- velocity is V = c, th e inv ariant spe ed of light.
8/14/2019 Introduction to Groups
37/164
In Class ical Mechanics, the conce pt of mome ntum is im porta nt be cause 37
of its r ole a s an invar iant in an isol ated system. W e the refor e int roduc e the
con cept of 4- momen tum i n Rel ativi stic Mechanics in or der t o find
pos sible Lore ntz i nvari ants invol ving this new q uantity. The c ontra variant
4-m oment um is defined a s:
P = mV
whe re m is th e mas s of the p artic le. ( It is a Lo rentz scalar, t he ma ss
mea sured in t he fr ame i n whi ch th e par ticle is a t res t).
The scalar pr oduct is
P P = (mc)2.
Now ,
P = [mc, mvN]
the refor e,
P P = (mc)2 (mvN)
2.
Wri ting
M = m, the r elati visti c mas s, we obta in
PP = (Mc)2 (M vN)
2 = (mc)2.
Multiply ing t hroug hout by c2 gi ves
M2c4 M2vN2c2 = m2c4.
The quan tity Mc2 ha s dim ensio ns of ener gy; w e the refor e wri te
E = Mc2
the tota l ene rgy o f a f reely movi ng pa rticl e.
8/14/2019 Introduction to Groups
38/164
This leads to the fun damen tal i nvari ant o f dyn amics 38
c2PP = E2 (p c)2 = Eo2
whe re
Eo = mc2 is the rest energ y of the p artic le, and
p is its rel ativi stic 3- momen tum.
The tota l ene rgy c an be writ ten:
E = Eo = Eo + T,
whe re
T = Eo
( 1) ,
the rela tivis tic k ineti c ene rgy.
The magn itude of t he 4- momen tum i s a L orent z inv ariant
P = mc.
The 4- m oment um tr ansforms a s follows:
P' = LP.
For rela tive motio n alo ng th e x-a xis, this equat ion is equ ivale nt to the
equ ations
E' = E cpx
and
cpx = -E + cpx .
Usi ng th e Pla nck-E instein eq uations E = h an d
E = pxc f or ph otons , the ener gy eq uation bec omes
' =
8/14/2019 Introduction to Groups
39/164
= (1 ) 39
= (1 )/( 1 2)1/2
= [(1 )/(1 + )]1/2 .
This is the r elati visti c Dop pler shift for the f reque ncy', measu red i n an
inertial fram e (pr imed) in t erms of th e fre quenc y me asure d in anoth er
inertial fram e (un prime d).
4.5 . The freq uency -wavenumbe r inv arian t
Par ticle -Wave dual ity, one o f the most prof ound
discover ies i n Phy sics, has its o rigins in Loren tz in variance. It w as
pro posed by d eBrog lie i n the earl y 192 0's. He u sed t he fo llowing
arg ument .
The disp lacement o f a w ave c an be writ ten
y(t ,r) = Acos (t - kr)
whe re = 2 (t he an gular freq uency ), k = 2/ (t he wa venum ber),
and r = [x, y , z] (the position v ector ). T he ph ase (t kr) c an be
wri tten ((/c) ct kr), and t his h as th e for m of a Lor entz invar iant
obt ained from the 4-vectors
E[ct , r], and K [/c, k]
whe re E is the event 4-ve ctor, and K is the "freq uency -wavenumbe r" 4-
vec tor.
deB roglie not ed th at th e 4-m oment um P is conn ected to t he ev ent 4 -
vec tor E th rough the 4-vel ocity V, a nd th e fre quenc y-wavenumb er 4-
vec tor K is conn ected to t he ev ent 4 -vector E th rough the Loren tz
8/14/2019 Introduction to Groups
40/164
inv ariant pha se of a wa ve ((/c) ct k r). He t heref ore p ropos ed th at a 40
dir ect c onnec tion must exist betw een P an d K; it is illu strat ed
in the f ollow ing d iagram:
E[ct ,r]
( Einst ein) PP=in v. EK=in v. (d eBrog lie)
P [E/ c,p] K[/c,k]
(deB roglie)
Fig . 2. The coup ling betwe en P an d K vi a E.
deB roglie pro posed that the conn ection is the
simplest poss ible, name ly, P an d K ar e pro porti onal to ea ch ot her. He
rea lized that ther e cou ld be only one value for the c onsta nt of
pro porti onali ty if the Planck-Einstein resu lt for pho tons E = h/2 is but a
spe cial case of a gener al re sultit must be h /2, w here h is Plancks
con stant . Th erefo re, d eBrog lie propos ed th at
P K
or
P = (h/2)K.
Equ ating the elements o f the 4-ve ctors give s
E = (h/2)
and
p = (h/2)k .
8/14/2019 Introduction to Groups
41/164
In these rema rkabl e equ ations, ou r not ions of pa rticl es an d wav es ar e 41
for ever merge d. T he sm allne ss of the value of P lanck's co nstan t pre vents
us from obser ving the d ualit y dir ectly; how ever, it i s clearly obser ved a t
the mole cular, ato mic, nuclear, a nd pa rticl e lev el.
4.6 . deB rogli e's i nvari ant.
The inva riant form ed fr om th e fre quenc y-wavenumb er 4- vecto r is
KK = (/c, k)[/c, - k]
= (/c)2 k2 = (o/c)2, w here o is the prope r
ang ular frequ ency.
Thi s inv arian t is the w ave v ersion of Einst ein's
ene rgy-m oment um in varia nt; it giv es th e dispersion re latio n
o2 = 2 (k c)2.
The rati o /k is th e pha se ve locit y of the w ave, v.
For a wa ve-packet, the group velo city vG is d/dk ; it can b e obt ained by
dif feren tiati ng th e dispersion eq uation as follo ws:
d kc2dk = 0
the refor e,
vG = d/dk = kc2/.
The deBr oglie inva riant invo lving the produ ct of the phase and group
velocity is t heref ore
vvG = (/k) .(kc2/) = c2.
Thi s is the w ave-equivalent of Ei nstein's f amous
8/14/2019 Introduction to Groups
42/164
E = M c2. 42
We see t hat
vvG = c2 = E/M
or,
vG = E/Mv = Ek/M = p/M = vN, t he pa rticl e
velocity.
This res ult p layed an i mport ant p art i n the deve lopment of Wave
Mec hanics.
We shall find in l ater chapt ers, that Loren tz tr ansformati ons f orm a
gro up, a nd th at in variance p rinci ples are r elated dir ectly to s ymmet ry
tra nsfor matio ns an d the ir as socia ted g roups .
8/14/2019 Introduction to Groups
43/164
5 43
GROUPS CONCRETE AND ABSTRACT
5.1 Some concrete examples
The elements of the set {1, i}, where i = -1, are the roots of the
equation x4 = 1, the fourth roots of unity. They have the following special
properties:
1. The product of any two elements of the set (including the same two
elements) is always an element of the set. (The elements obey closure).
2. The order of combining pairs in the triple product of any elements
of the set does not matter. (The elements obey associativity).
3. A unique element of the set exists such that the product of any
element of the set and the unique element (called the identity) is equal to the
element itself. (An identity element exists).
4. For each element of the set, a corresponding element exists such
that the product of the element and its corresponding element (called the
inverse) is equal to the identity. (An inverse element exists).
The set of elements {1, i} with these four properties is said to form
a GROUP.
In this example, the law of composition of the group is multiplication; this
need not be the case. For example, the set of integers Z = {.., -2, -1, 0, 1, 2,
...} forms a group if the law of composition is addition. In this group, the
identity element is zero, and the inverse of each integer is the integer with the
same magnitude but with opposite sign.
8/14/2019 Introduction to Groups
44/164
In a different vein, we consider the set of 44 matrices: 44
1 0 0 0 0 0 0 1 0 0 1 0 0 1 0 0{M} = 0 1 0 0 , 1 0 0 0 , 0 0 0 1 , 0 0 1 0 .
0 0 1 0 0 1 0 0 1 0 0 0 0 0 0 1
0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 0
If the law of composition is matrix multiplication , then {M} is found to obey:
1 --closure
and
2 --associativity,
and to contain:
3 --an identity, diag(1, 1, 1, 1),
and
4 --inverses.
The set {M} forms a group under matrix multilication.
5.2. Abstract groups
The examples given above illustrate the generality of the group
concept. In the first example, the group elements are real and imaginary
numbers, in the second, they are positive and negative integers, and in the
third, they are matrices that represent linear operators (see later discussion).
Cayley, in the mid-19th century, first emphasized this generality, and he
introduced the concept of an abstract group, Gn which is a collection of n
distinct elements (...gi...) for which a law of composition is given. If n is finite,
the group is said to be a group of order n. The collection of elements must
obey the four rules:
8/14/2019 Introduction to Groups
45/164
1. If gi, gj G then gn = gjgi G gi, gj G (closure) 45
2. gk(gjgi) = (gkgj)gi [leaving out the composition symbol] (associativity)
3. e G such that gie = egi = gi gi G (an identity exists)
4. If gi G then gi-1 G such that gi
-1gi = gigi-1 = e (an inverse exists).
For finite groups, the group structure is given by listing all
compositions of pairs of elements in a group table, as follows:
e . gi gj . (1st symbol, or operation, in pair)e . . . .. . . . .
gi . . g igi g igj .
gj . gjgi gjgj .
gk . gkgi gkgj ...
If gjgi = gigj gi, gj G, then G is said to be a commutative or abelian
group. The group table of an abelian group is symmetric under reflection in
the diagonal.
A group of elements that has the same structure as an abstract group is
a realization of the group.
5.3 The dihedral group, D3
The set of operations that leaves an equilateral triangle invariant under
rotations in the plane about its center, and under reflections in the three
planes through the vertices, perpendicular to the opposite sides, forms a
8/14/2019 Introduction to Groups
46/164
group of six elements. A study of the structure of this group (called the 46
dihedral group, D3) illustrates the typical group-theoretical approach.
The geometric operations that leave the triangle invariant are:
Rotations about the z-axis (anticlockwise rotations are positive)
Rz(0) (123) (123) = e, the identity
Rz(2/3)(123) (312) = a
Rz(4/3)(123) (231) = a2
and reflections in the planes I, II, and III:
RI (123) (123) = b
RII (123) (321) = c
RIII (123) (213) = d
This set of operators is D3 = {e, a, a2, b, c, d}.
Positive rotations are in an anticlockwise sense and the inverse rotations are in
a clockwise sense., so that the inverse of e, a, a2 are
e-1 = e, a-1 = a2, and (a2)-1 = a.
The inverses of the reflection operators are the operators themselves:
b-1 = b, c -1 = c, and d-1 = d.
We therefore see that the set D3 forms a group. The group
multiplication table is:
e a a2 b c de e a a2 b c da a a2 e d b ca2 a2 e a c d bb b c d e a a2
c c d b a2 e ad d b c a a2 e
8/14/2019 Introduction to Groups
47/164
In reading the table, we follow the rule that the first operation is written on 47
the right: for example, ca2 = b. A feature of the group D3 is that it can be
subdivided into sets of either rotations involving {e, a, a2
} or reflections
involving {b, c, d}. The set {e, a, a2} forms a group called the cyclic group
of order three, C3. A group is cyclic if all the elements of the group are
powers of a single element. The cyclic group of order n, Cn, is
Cn = {e, a, a2, a3, .....,a
n-1},
where n is the smallest integer such that an = e, the identity. Since
aka
n-k= a
n= e,
an inverse an-k
exists. All cyclic groups are abelian.
The group D3 can be broken down into a part that is a group C3, and a
part that is the product of one of the remaining elements and the elements of
C3. For example, we can write
D3 = C3 + bC3 , b C3
= {e, a, a2} + {b, ba, ba2}
= {e, a, a2} + {b, c, d}
= cC3 = dC3.
This decomposition is a special case of an important theorem known as
Lagranges theorem. (Lagrange had considered permutations of roots of
equations before Cauchy and Galois).
5.4 Lagranges theorem
8/14/2019 Introduction to Groups
48/164
The order m of a subgroup Hm of a finite group Gn of order n is a 48
factor (an integral divisor) of n.
Let
Gn = {g1=e, g2, g3, ...gn} be a group of order n, and let
Hm = {h1=e, h2, h3, ...hm} be a subgroup of Gn of order m.
If we take any element gkof Gn which is not in Hm, we can form the set of
elements
{gkh1, gkh2, gkh3, ...gkhm} gkHm.
This is called the left cosetof Hm with respect to gk. We note the important
facts that all the elements of gkhj, j=1 to m are distinct, and that none of the
elements gkhjbelongs to Hm.
Every element gkthat belongs to Gn but does not belong to Hm
belongs to some coset gkHm so that Gn forms the union of Hm and a number
of distinct (non-overlapping) cosets. (There are (n - m) such distinct cosets).
Each coset has m different elements and therefore the order n of Gn is
divisible by m, hence n = Km, where the integer K is called the index of the
subgroup Hm under the group Gn. We therefore write
Gn = g1Hm + gj2Hm + gk3Hm + ....gKHm
where
8/14/2019 Introduction to Groups
49/164
gj2 Gn Hm, 49
gk3 Gn Hm, gj2Hm
.
gnK Gn Hm, gj2Hm, gk3Hm, ...gn-1, K-1Hm.
The subscripts 2, 3, 4, ..K are the indices of the group.
As an example, consider the permutations of three objects 1, 2, 3 ( the
group S3) and let Hm = C3 = {123, 312, 231}, the cyclic group of order
three. The elements of S3 that are not in H3 are {132, 213, 321}. Choosing
gk= 132, we obtain
gkH3 = {132, 321, 213},
and therefore
S3 = C3 + gk2C3 ,K = 2.
This is the result obtained in the decomposition of the group D3, if we make
the substitutions e = 123, a = 312, a2 = 231, b = 132, c = 321, and d = 213.
The groups D3 and S3 are said to be isomorphic. Isomorphic groups have
the same group multiplication table. Isomorphism is a special case of
homomorphism that involves a many-to-one correspondence.
5.5 Conjugate classes and invariant subgroups
8/14/2019 Introduction to Groups
50/164
If there exists an element v Gn such that two elements a, b Gn are 50
related by vav-1 = b, then b is said to be conjugate to a. A finite group can
be separated into sets that are conjugate to each other.
The class of Gn is defined as the set of conjugates of an element a
Gn. The element itself belongs to this set. If a is conjugate to b, the class
conjugate to a and the class conjugate to b are the same. If a is not conjugate
to b, these classes have no common elements. Gn can be decomposed into
classes because each element of Gn belongs to a class.
An element of Gn that commutes with all elements of Gn forms a class
by itself.
The elements of an abelian group are such that
bab -1 = a for all a, b Gn,
so that
ba = ab.
If Hm is a subgroup of Gn, we can form the set
{aea-1, ah2a-1, ....ahma
-1} = aHma-1 where a Gn .
Now, aHma-1 is another subgroup of Hm in Gn. Different subgroups may be
found by choosing different elements a of Gn. If, for all values of a Gn
aHma-1 = Hm
8/14/2019 Introduction to Groups
51/164
(all conjugate subgroups of Hm in Gn are identical to Hm), 51
then Hm is said to be an invariant subgroup in Gn.
Alternatively, Hm is an invariant in Gn if the left- and right-cosets
formed with any a Gn are equal, i. e. ahi = hka.
An invariant subgroup Hm of Gn commutes with all elements of Gn.
Furthermore, if hi Hm then all elements ahia-1 Hm so that Hm is an
invariant subgroup of Gn
if it contains elements of Gn
in complete classes.
Every group Gn contains two trivial invariant subgroups, Hm = Gn and
Hm = e. A group with no proper (non-trivail) invariant subgroups is said to
be simple (atomic). If none of the proper invariant subgroups of a group is
abelian, the group is said to be semisimple.
An invariant subgroup Hm and its cosets form a group under
multiplication called thefactor group (written Gn/Hm) of Hm in Gn.
These formal aspects of Group Theory can be illustrated by considering
the following example:
The group D3 = {e, a, a2, b, c, d} ~ S3 = {123, 312, 231, 132, 321, 213}.
C3 is a subgroup of S3: C3 = H3 = {e, a, a2} = {123, 312, 231}.
Now,
bH3 = {132, 321, 213} = H3b
8/14/2019 Introduction to Groups
52/164
8/14/2019 Introduction to Groups
53/164
If we now consider Hm = {e, b} an abelian subgroup, we find 53
aHm = {a,d}, Hma = {a.c},
a2Hm = {a2,c}, Hma2 = {a2, d}, etc.
All left and right cosets are not equal: Hm = {e, b} is therefore not an
invariant subgroup of S3. We can therefore write
S3 = {e, b} + {a, d} + {a2, c}
= Hm + aHm + a2Hm.
Applying Lagranges theorem to S3 gives the orders of the possible
subgroups: they are
order 1: {e}
order 2: {e, d}; {e, c}: {e, d}
order 3: {e, a, a2} (abelian and invariant)
order 6: S3.
5.6 Permutations
A permutation of the set {1, 2, 3, ....,n} of n distinct elements is an
ordered arrangement of the n elements. If the order is changed then the
permutation is changed. the number of permutations of n distinct elements is
n!
We begin with a familiar example: the permutations of three distinct
objects labelled 1, 2, 3. There are six possible arrangements; they are
123, 312, 231, 132, 321, 213.
8/14/2019 Introduction to Groups
54/164
These arrangements can be written conveniently in matrix form: 54
1 2 3 1 2 3 1 2 31 = , 2 = , 3 = ,
1 2 3 3 1 2 2 3 1
1 2 3 1 2 3 1 2 34 = , 5 = , 6 = .
1 3 2 3 2 1 2 1 3
The product of two permutations is the result of performing one arrangement
after another. We then find
23 = 1
and
32 = 1
whereas
45 = 3
and
54 = 2.
The permutations 1, 2, 3 commute in pairs (they correspond to the
rotations of the dihedral group) whereas the permutations do not commute
(they correspond to the reflections).
A general product of permutations can be written
s1 s2 . . .sn 1 2 . . n 1 2 . . n= .
t1 t2 . . .tn s1 s2 . . sn t1 t2 . . tn
The permutations are found to have the following properties:
1. The product of two permutations of the set {1, 2, 3, ...} is itself a
permutation of the set. (Closure)
8/14/2019 Introduction to Groups
55/164
2. The product obeys associativity: 55
(kj)i = k(ji), (not generally commutative).
3. An identity permutation exists.
4. An inverse permutation exists:
s1 s2 . . . sn
-1 = | |1 2 . . . n
such that -1 = -1 = identity permutation.
The set of permutations therefore forms a group
5.7 Cayleys theorem:
Every finite group is isomorphic to a certain permutation group.
Let Gn ={g1, g2, g3, . . .gn} be a finite group of order n. We choose any
element gi in Gn, and we form the products that belong to Gn:
gig1, gig2, gig3, . . . gign.
These products are the n-elements of Gn
rearranged. The permutation i,
associated with gi is therefore
g1 g2 . . gn i = .
gig1 gig2 . . gign
If the permutation j associated with gj is
g1 g2 . . gn
j = ,gjg1 gjg2 . . gjgn
where gi gj, then
g1 g2 . . gn ji = .
(gjgi)gi (gjgi)g2 . . (gjgi)gn
8/14/2019 Introduction to Groups
56/164
This is the permutation that corresponds to the element gjgi of Gn. 56
There is a direct correspondence between the elements of Gn and the n-
permutations {1,
2, . . .n}. The group of permutations is a subgroup of
the full symmetric group of order n! that contains all the permutations of the
elements g1, g2, . . gn.
Cayleys theorem is important not only in the theory of finite groups
but also in those quantum systems in which the indistinguishability of the
fundamental particles means that certain quantities must be invariant under
the exchange or permutation of the particles.
8/14/2019 Introduction to Groups
57/164
6 57
LIES DIFFERENTIAL EQUATION, INFINITESIMAL ROTATIONS
AND ANGULAR MOMENTUM OPERATORS
Although the field of continuous transformation groups (Lie groups)
has its origin in the theory of differential equations, we shall introduce the
subject using geometrical ideas.
6.1 Coordinate and vector rotations
A 3-vector v = [vx, vy, vz] transforms into v = [vx, vy, vz] under a
general coordinate rotation R about the origin of an orthogonal coordinatesystem as follows:
v = R v,
where
i.i j.i k.i
R
=i.j j.j k.j i.k j.k k.kcosii . .
= cosij . . cosik . coskk
where i,j, k, i,j, k are orthogonal unit vectors, along the axes, before and
after the transformation, and the cosiis are direction cosines.
The simplest case involves rotations in the x-y plane:
vx = cosii cosjivxvy cosij cosjj vy
8/14/2019 Introduction to Groups
58/164
= cos sin = R c()v 58-sin cos
where R c() is the coordinate rotation operator. If the vector is rotated in a
fixed coordinate system, we have so that
v = R v()v,where
R v() = cos -sin .sin cos
6.2 Lies differential equation
The main features of Lies Theory of Continuous Transformation
Groups can best be introduced by discussing the properties of the rotation
operator R v() when the angle of rotation is an infinitesimal. In general,
R v() transforms a point P[x, y] in the plane into a new point P[x, y]:
P = R v()P. Let the angle of rotation be sufficiently small for us to put
cos() 1 and sin() , in which case, we have
R v() = 1 1
and
x = x.1 - y = x - y
y = x + y.1 = x + y
8/14/2019 Introduction to Groups
59/164
Let the corresponding changes x x and y y be written 59
x = x + x and y = y +y
so that
x = -y and y = x.
We note that
R v() = 1 0 + 0 -1 0 1 1 0
= I + iwhere
i = 0 -1 = Rv(/2).
1 0
Lie introduced another important way to interpret the operator
i = R v(/2), that involves the derivative ofR v() evaluated at the identity
value of the parameter, = 0:
dRv
()/d
=
sin cos
=
0 -1
= i
=0 cos sin 1 0 = 0
so that
R v() = I + dR v()/d ., = 0
a quantity that differs from the identity I by a term that involves the
infinitesimal, : this is an infinitesimal transformation.
Lie was concerned with Differential Equations and not Geometry. He
was therefore motivated to discover the key equation
8/14/2019 Introduction to Groups
60/164
dR v()/d = 0 -1cos -sin 601 0 sin cos
= iR v() .This isLies differential equation .
Integrating between = 0 and = , we obtain
R v()
dR v()/R v() = id I 0
so that
ln(R v()/I) = i,or
R v() = Iei
, the solution of Lies equation.
Previously, we obtained
R v() = Icos + isin.
We have, therefore
Iei
= Icos + isin .This is an independent proof of the famous Cotes-Euler equation.
We introduce an operator of the form
O = g(x, y, /x, /y),
and ask the question: does
x = Of(x, y; ) ?
Lie answered the question in the affirmative; he found
x = O(x) = (x/y y/x)x = y
and
8/14/2019 Introduction to Groups
61/164
y = O(y) = (x/y y/x)y = x . 61
Putting x = x1 and y = x2, we obtain
xi = Xxi , i = 1, 2
where
X = O = (x1/x2 x2/x1), the generator of rotations in the plane.
6.3 Exponentiation of infinitesimal rotations
We have seen that
R v() = ei,
and therefore
R v() = I + i, for an infinitesimal rotation,
Performing two infinitesimalrotations in succession, we have
R v2() = (I + i)2
= I + 2i to first order,
= R v(2).
ApplyingR v() n-times gives
R vn() = R v(n) = e
in= e
i
= R v() (as n and 0, the
product n).
This result agrees, as it should, with the exact solution of Lies differential
equation.
8/14/2019 Introduction to Groups
62/164
A finite rotation can be built up by exponentiation of infinitesimal 62
rotations, each one being close to the identity. In general, this approach has
the advantage that the infinitesimal form of a transformation can often be
found in a straightforward way, whereas the finite form is often intractable.
6.4 Infinitesimal rotations and angular momentum operators
In Classical Mechanics, the angular momentum of a mass m, moving in
the plane about the origin of a cartesian reference frame with a momentum p
is
Lz = rp = rpsinnz
where nz is a unit vector normal to the plane, and is the angle between r
and p.
In component form, we have
Lzcl = xpy ypx, where px and py are the cartesian
components ofp.
The transition between Classical and Quantum Mechanics is made by
replacing
px by i(h/2)/x (a differential operator)
and py by i(h/2)/y (a differential operator),where h
is Plancks constant.
We can therefore write the quantum operator as
LzQ = i(h/2)(x/y y/x) = i(h/2)X
and therefore
8/14/2019 Introduction to Groups
63/164
X = iLzQ/(h/2), 63
and
xi = Xxi = (2iLzQ/h)xi, i = 1,2.
Let an arbitrary, continuous, differentiable function f(x, y) be
transformed under the infinitesimal changes
x = x - y
and
y = y + x .
Using Taylors theorem, we can write
f(x, y) = f(x + x, y + y)
= f(x y, y + x)
= f(x, y) + ((f/x)x + ((f/y)y)
= f(x, y) + (y(/x) + x(/y))f(x, y)
= I + 2iLz
/h)f(x, y)
= e2iLz/h
f(x, y)
= R v(2Lz/h) f(x, y).
The invatriance of length under rotations follows at once from this result:
If f(x, y) = x2 + y2 then
f/x = 2x and f/y = 2y, and thereforef(x, y) = f(x, y) + 2xx + 2yy
= f(x, y) - 2x(y) + 2y(x)
= f(x, y) = x2 + y2 = invariant.
8/14/2019 Introduction to Groups
64/164
This is the only form that leads to the invariance of length under rotations. 64
6.5 3-dimensional rotations
Consider three successive counterclockwise rotations about the x, y,
and z axes through angles , , and , respectively:
zz y
about xy y
x x, x
z y z y, y about y
x x xz z
y y about z
x x xThe total transformation is
R c(, , ) = R c()R c()R c()
coscos cossinsin + sincos cossincos + sinsin= sincos sinsinsin + coscos sinsincos + sinsin
sin cossin coscos
For infinitesimal rotations, the total rotation matrix is, to 1st-order in the s:
1 R c(, , ) = 1 . 1
8/14/2019 Introduction to Groups
65/164
The infinitesimal form can be written as follows: 65
1 0 1 0 1 0 0 R c(, , ) = 1 0 0 1 0 0 1
0 0 1 0 1 0 1
= I + Y3I + Y2I + Y1
where
0 0 0 0 0 -1 0 1 0
Y1 = 0 0 1 , Y2 = 0 0 0 , Y3 = -1 0 0 .0 -1 0 1 0 0 0 0 0
To 1st-order in the s, we have
R c(, , ) = I + Y1 + Y2 + Y3 .
6.6 Algebra of the angular momentum operators
The algebraic properties of the Ys are important. For example, we find
that their commutators are:
0 0 0 0 0 -1 0 0 -1 0 0 0[Y1, Y2] = | 0 0 1|| 0 0 0| | 0 0 0|| 0 0 1|
0 -1 0 1 0 0 1 0 0 0 -1 0
= Y3 ,
[Y1, Y3] = Y2 ,
and
[Y2
, Y3
] = -Y1
.
These relations define the algebra of the Ys. In general, we have
[Yj, Yk] = Yl = jklYl ,
8/14/2019 Introduction to Groups
66/164
8/14/2019 Introduction to Groups
67/164
x = x - x = y - z 67
y = y - y = -x + z
z = z - z = x - y .
Introducing the classical angular momentum operators: Licl, we find that
these small changes can be written 3
xi = kXkxi k = 1
For example, if i = 1
x1 = x = (z/y - y/z)x
+ (-z/x + x/z)x
+ (y/x - x/y)x = -z + y .
Extending Lies method to three dimensions, the infinitesimal form
of the rotation operator is readily shown to be
3R c(, , ) = I + (R c/i)| i .
i= 1 All is = 0
8/14/2019 Introduction to Groups
68/164
7 68
LIES CO NTINUOUS T RANSF ORMAT ION G ROUPS
In the p revio us ch apter , we discu ssed the p roper ties of infinit esima l
rot ations in 2- an d 3-d imensions, and we fo und t hat t hey a re re lated
dir ectly to t he an gular mome ntum opera tors of Qu antum Mech anics .
Imp ortan t al gebra ic pr opert ies of the matr ix re prese ntati ons o f the
ope rator s also wer e int roduc ed. In th is chapter , we shall cons ider the
sub ject in ge neral term s.
Let xi, i = 1 to n be a set o f n v ariables. They may be co nside red t o
be the c oordi nates of a poin t in an n- dimensiona l vec tor s pace, Vn. A set
of equat ions invol ving the x is is ob tained by the t ransformat ions
x i = fi(x1, x2, . ..xn: a1, a2, . ...ar), i = 1 to n
in which the set a1, a2, . ..ar co ntains r-indepe ndent para meter s. T he se t Ta,
of trans forma tions maps x x . We shall write
x = f(x; a) or x = Tax
for the set o f fun ctions.
It is assumed that the funct ions fi ar e differen tiabl e wit h res pect to
the xs and t he a s to any r equir ed or der. Thes e fun ctions nec essarily
dep end o n the esse ntial para meter s, a. Thi s mea ns th at no two
tra nsfor matio ns with di fferent nu mbers of p arame ters are t he sa me. r is
the smallest numbe r req uired to c harac teriz e the tran sform ation,
com pletely.
The set of fu nctio ns fi fo rms a fini te co ntinu ous g roup if:
8/14/2019 Introduction to Groups
69/164
1. The result of two s uccessive trans forma tions x x x is equivalent 69
to a single t ransf ormat ion x x :
x = f( x; b) = f(f( x; a); b)
= f(x ; c)
= f(x ; (a; b))
whe re c is th e set of p arame ters
c = (a ; b) , = 1 to r,
and
2. To e very trans forma tion there corr espon ds a uniqu e inv erse that
belongs to th e set :
a such that x = f (x; a) = f(x; a)
We have
TaTa-1 = Ta
-1Ta = I, the i dentity.
We shall see that 1) is a hi ghly restr ictiv e req uirem ent.
The tran sform ation x = f(x; a0) is the iden tity. Without loss of
gen erali ty, w e can take a0 = 0. T he es sential po int o f Lie s th eory of
con tinuo us tr ansformati on gr oups is to cons ider that part of th e gro up th at
is close to t he id entit y, an d not to c onsider th e gro up as a wh ole.
Suc cessi ve in finit esima l cha nges can b e use d to build up t he finite chang e.
7.1 One- param eter group s
Con sider the trans forma tion x x unde r a f inite chan ge in a si ngle
par amete r a, and t hen a chan ge x + dx . T here are t wo pa ths f rom x
x + dx ; the y are as s hown:
8/14/2019 Introduction to Groups
70/164
x 70an infi nitesimal
aa , a f inite para meter chan ge
x + dx a + daa di fferential
x (a = 0)
We have
x + d x = f(x; a + d a)
= f (f(x; a); a) = f(x ; a)
The 1st- order Taylor ex pansi on is
dx = f (x; a)/a a u( x) a a = 0
The lie group cond ition s the n dem and
a + da = (a; a).
But
(a; 0) = a, (b = 0)
the refor e
a + da = a + (a; b)/b a b = 0
so thatd a = (a; b)/b a
b = 0
or
a = A(a)da.
The refor e
dx = u (x)A (a)da ,
leading to
8/14/2019 Introduction to Groups
71/164
8/14/2019 Introduction to Groups
72/164
and
x2 = 2x2a. 72
In the l imit, thes e equ ations giv e
dx1/x1 = dx2/2x2 = da.
The se ar e the diff erent ial e quations t hat c orres pond to th e infinite simal
equ ations abo ve.
Int egrat ing, we ha ve
x1 a x2 a
dx 1/x1 =
da and
dx 2/2x2 = da ,
x1 0 x2 0
so that
lnx1 - ln x1 = a = ln (x1/x1)
and
ln (x2/x2) = 2a = 2ln(x1/x1)
or
U = (x2/x12) = U = ( x2/x1
2) .
Put ting V = lnx1, w e obt ain
V = V + a and U = U, th e tra nslat ion g roup.
7.2 Det ermin ation of t he fi nite equat ions from the i nfini tesim al
for ms
Let the finit e equ ations of a one -para meter grou p G(1) be
x1 = (x1, x2 ; a)
and
x2 = (x1, x2 ; a),
8/14/2019 Introduction to Groups
73/164
and let the i dentity co rresp ond t o a = 0. 73
We consi der t he tr ansformati on of f(x1, x2) t o f(x1, x2). We expan d
f(x1, x2) in a Macla urin series in the p arame ter a (at defin ite v alues of x 1
and x2):
f(x1, x2) = f (0) + f (0)a + f (0) a2 /2! + ...
whe re
f(0) = f(x1, x2)| a=0 = f(x1, x2),
and
f(0) = (df(x1, x2)/ da| a=0
= {(f/x1)( dx1/d a) + (f /x2)( dx2/d a)}| a= 0
= {(f/x1)u(x1, x2) + (f/x2)v(x1, x2)}|a=0
the refor e
f (0) = {( u(/x1) + v(/x2))f }|a=0
= Xf(x1, x2).Con tinui ng in this way, we h ave
f (0) = {d2f(x1, x2)/ da2}|a=0 = X
2f(x1, x2), etc....
The func tion f(x1, x2) can b e exp anded in t he se ries
f(x1, x2) = f (0) + af (0) + ( a2/2! )f( 0) + ...
= f(x1, x2) + aXf + (a2/2! ) X2f + .. .
Xn
f is the symb ol for ope ratin g n-t imes in su ccess ion o f f w ith X.
The fini te eq uations of the group are there fore
x1 = x1 + aXx1 + (a2/2! ) X2x1 + ...
andx2 = x2 + aXx2 + (a
2/2! ) X2x2 + = ...
8/14/2019 Introduction to Groups
74/164
If x1 a nd x2 ar e def inite valu es to whic h x1an d x2 r educe for the i dentity 74
a=0 , the n the se eq uations ar e the seri es so lutio ns of the diffe renti al
equ ations
dx1/u (x1, x2) = d x2/v (x1, x2) = d a.
The grou p is referred t o as the g roupXf.
For exam ple, let
Xf = (x1/x1 + x2/x2)f
the n
x1 = x1 + aXx1 + (a2/2! ) X2f . ..
= x1 + a(x1/x1 + x2/x2)x1 + ...
= x1 + ax1 + (a2/2! )(x 1/x1 + x2/x2)x1 +
= x1 + ax1 + (a2/2! )x 1 + ...
= x1(1 + a + a2 /2! + ...)
= x1ea
.
Also, we find
x2 = x2ea.
Put ting b = ea, w e hav e
x1 = bx 1, a nd x2 = bx2.
The fini te gr oup i s the grou p of magni ficat ions.
IfX = (x/y - y/x ) we find, for examp le, that t he finite group is t he
gro up of 2-dimensional rotat ions.
8/14/2019 Introduction to Groups
75/164
7.3 Inv arian t fun ction s of a gro up 75
Let
Xf = (u/x1 + v/x2)f defin e a o ne-paramet er
gro up, a nd le t a=0 give the identity. A fu nction F(x1, x2) is ter med a n
inv arian tun der t he tr ansformati on gr oup G (1) if
F( x1, x2) = F (x1, x2)
for all values of the p arame ter, a.
The func tion F(x1, x2) can b e exp anded as a seri es in a:
F(x1, x2) = F(x 1, x2) + aXF + (a2
/2! ) X(XF) + . ..
If
F(x1, x2) = F (x1, x2) = in variant fo r all valu es of a,
it is necessa ry fo r
XF = 0,
and this mean s tha t
{u( x1, x2)/x1 + v(x 1, x2)/x2}F = 0 .
Con seque ntly,
F(x1, x2) = co nstan t
is a solution of
dx1/u( x 1, x2) = dx2/v( x 1, x2) .
This equ ation has one s oluti on th at de pends on o ne ar bitra ry co nstan t, an d
the refor e G(1) ha s onl y one basic inv arian t, an d all othe r pos sible inva riant s
can be g iven in te rms o f the basic inv ariant.
For exam ple, we no w rec onsider th e the inva riant s of rotat ions:
8/14/2019 Introduction to Groups
76/164
The infi nitesimal trans forma tions are given by 76
Xf = (x 1/x2 - x2/x1),
and the diffe renti al eq uation tha t giv es th e inv ariant fun ction F of the
gro up is obta ined by so lving the chara cteri stic diffe renti al eq uations
dx1/x2 = d, a nd dx 2/x1 = -d,
so that
dx1/x2 + dx2/x1 = 0.
The solu tion of th is equation is
x12
+ x22
= cons tant,
and ther efore the invar iant funct ion is
F (x1, x2) = x12 + x2
2.
All func tions of x 12 + x2
2 ar e the refor e inv ariants of the 2-dimensio nal
rot ation grou p.
This met hod c an be gene raliz ed. A gro up G (1) in n-va riabl es de fined
by the e quation
xi = (x1, x2, x3, . ..xn; a ), i = 1 to n ,
is equivalent to a uniq ue in finit esima l tra nsfor matio n
Xf = u1(x1, x2, x3, . ..xn)f /x1 + ... un(x1, x2, x3, . ..xn)f /xn .
If a is the g roup param eter then the i nfini tesimal tr ansformati on is
x i = xi + u i(x1, x2, . ..xn)a (i = 1 t o n),
the n, if E(x1, x2, . ..xn) is a f unction th at ca n be diffe renti ated n-times with
res pect to it s arg ument s, we have
8/14/2019 Introduction to Groups
77/164
E (x1, x2, ...xn) = E (x1, x2, . ..xn) + aXE + (a2/2! ) X2E + . 77
Let (x1, x2, . ..xn) b e the coor dinates of a po int in n-s pace and l et a be a
par amete r, in depen dent of th e xis. As a var ies, the p oint (x1, x2, . ..xn) w ill
des cribe a tr ajectory, start ing from t he in itial poin t (x1, x2, . ..xn). A
nec essary and sufficien t con ditio n tha t F(x 1, x2, . ..xn) b e an invar iant
fun ction is t hat XF = 0. A cur ve F = 0 i s a t rajectory and t heref ore a n
inv ariant cur ve if
XF(x1, x2, x3, . ..xn) = 0.
8/14/2019 Introduction to Groups
78/164
8 78
PROPERTIES OF n-VARIABLE, r-PARAMETER LIE GROUPS
The change of an n-variable function F(x) produced by the
infinitesimal transformations associated with r-essential parameters is:n
dF = (F/xi)dxii = 1
wherer
dxi = ui(x)a , the Lie form. = 1
The parameters are independent of the xis therefore we can write r n
dF = a{ ui(x)(/xi)F} =1 i = 1 r
= aX F = 1
where the infinitesimal generators of the group are n
X ui(x)(/xi) , = 1 to r. i = 1
The operator
r I + Xa = 1
differs infinitesimally from the identity.
The generators X have algebraic properties of basic importance in the
Theory of Lie Groups. The Xs are differential operators. The problem is
therefore one of obtaining the algebraic structure of differential operators.
This problem has its origin in the work of Poisson (1807); he
introduced the following ideas:
The two expressions
X1f = (u11/x1 + u12/x2)fand
8/14/2019 Introduction to Groups
79/164
X2f = (u21/x1 + u22/x2)f 79
where the coefficients ui are functions of the variables x1, x2, and f(x1, x2)
is an arbitrary differentiable function of the two variables, are termed
linear differential operators.
The product in the order X2 followed by X1 is defined as
X1X2f = (u11/x1 + u12/x2)(u21f/x1 + u22f/x2)
The product in the reverse order is defined as
X2X1f = (u21/x1 + u22/x2)(u11f/x1 + u12f/x2).
The difference is
X1X2f - X2X1f = X1u21f/x1 + X1u22f/x2
- X2u11f/x1 - X2u12f/x2.
= (X1u21 - X2u11)f/x1 + (X1u22 - X2u12)f/x2
[X1, X2]f.
This quantity is called the Poisson operatoror the commutatorof the
operators X1f and X2f.
The method can be generalized to include = 1 to r essential parameters
and i = 1 to n variables. The ath-linear operator is then
Xa = uiaf/xi n
= uiaf/xi , ( a sum over repeated indices). i = 1
Lies differential equations have the formxi/a = uik(x)Ak(a) , i = 1 to n, = 1 to r.
Lie showed that
(ck/a)uik = 0
8/14/2019 Introduction to Groups
80/164
in which 80
ujui/xj - ujui/xj = ck(a)uik(x),
so that the cks are constants. Furthermore, the commutators can be
written
[X, X] = ( ckujk)/xj
= ckXk.
The commutators are linear combinations of the Xks. (Recall the earlier
discussion of the angular momentum operators and their commutators).
The cks are called the structure constants of the group. They have the
properties
ck = -ck ,
cc + cc + cc = 0.
Lie made the remarkable discovery that, given these structure constants,
the functions that satisfy
xi/a = uikAk(a) can be found.
(Proofs of all the above important statements, together with proofs of
Lies three fundamental theorems, are given in Eisenharts
standard workContinuous Groups of Transformations, Dover Publications,
1961).
8.1 The rank of a groupLet A be an operator that is a linear combination of the generators
of a group, Xi:
A = iXi (sum over i),
8/14/2019 Introduction to Groups
81/164
and let 81 X = xjXj .
The rank of the group is defined as the minimum number of commuting,
linearly independent operators of the form A.
We therefore require all solutions of
[A, X] = 0.For example, consider the orthogonal group, O+(3); here
A = iXi i = 1 to 3,and
X = xjXj j = 1 to 3
so that[A, X] = ixj[Xi, Xj] i, j = 1 to 3
= ixjijkXk .
The elements of the sets of generators are linearly independent, therefore
ixjijk = 0 (sum over i, j,, k = 1, 2, 3)
This equation represents the equations
-2 1 0 x1 03 0 -2x2 = 0 .0 -3 2 x3 0
The determinant of is zero, therefore a non-trivial solution of the xjs
exists. The solution is given by
xj = j (j = 1, 2, 3)
so that
A = X .
O+(3) is a group of rank one.
8.2 The Casimir operator of O+(3)
8/14/2019 Introduction to Groups
82/164
The generators of the rotation group O+(3) are the operators. Yks, 82
discussed previously. They are directly related to the angular momentum
operators, Jk:
Jk = -i(h/2)Yk(k = 1, 2, 3).
The matrix representations of the Yks are
0 0 0 0 0 -1 0 1 0
Y1 = 0 0 1, Y2 = 0 0 0 , Y3 = -1 0 0 .0 -1 0 1 0 0 0 0 0
The square of the total angular momentum, J is
3J2 = Ji2
1
= (h/2)2 (Y12 + Y2
2 + Y32)
= (h/2)2(-2I).
Schurs lemma states that an operator that is a constant multiple ofI
commutes with all matrix irreps of a group, so that
[Jk, J2] = 0 , k = 1,2 ,3.
The operator J2 with this property is called the Casimir operator of the
group O+(3).
In general, the set of operators {Ci} in which the elements commute
with the elements of the set of irreps of a given group, forms the set of
Casimir operators of the group. All Casimir operators are constant multiples
of the unit matrix:
Ci = aiI.
The constants ai are characteristic of a particular representation of a group.
8/14/2019 Introduction to Groups
83/164
9 83
MAT RIX R EPRES ENTATIONS OF GR OUPS
Mat rix r epres entat ions of li near opera tors are i mport ant i n Lin ear
Algebra; we s hall see t hat t hey a re eq ually impo rtant in G roup Theor y.
If a gro up of m m matri ces
Dn(m) = {D1
(m)(g1),. ..Dk(m)(gk), ...Dn
(m)(gn)}
can be f ound in wh ich e ach e lement is associated with the corre spond ing
element gkof a gr oup o f ord er n
Gn = {g1,...gk,....gn},
and the matri ces o bey
Dj(m)(gj)Di
(m)(gi) = Dji(m)(gjgi),
and
D1(m)(g1) = I, t he id entit y,
the n the matr ices Dk(m)(gk) a re sa id to form an m -dime nsional
rep resen tatio n of Gn. If th e ass ociation is one -to-one we have an
isomorph ism and th e rep resen tatio n is said to be fai thful .
The subj ect o f Gro up Re prese ntations f orms a ver y lar ge bra nch o f
Gro up Th eory. The re ar e man y sta ndard work s on this topic (see the
bib liography), eac h one cont ainin g num erous definitio ns, lemmas and
the orems . He re, a rath er br ief accoun t is given of s ome o f the more
imp ortan t res ults. The read er sh ould delve into the deepe r asp ects of th e
sub ject as th e nee d ari ses. The subje ct wi ll be intr oduce d by consi derin g
8/14/2019 Introduction to Groups
84/164
rep resen tatio ns of the rotat ion g roups , and thei r cor respo nding cycl ic 84
gro ups.
9.1 The 3-di mensi onal repre senta tion of ro tatio ns in the plane
The rota tion of a vecto r thr ough an an gle in the plane is
cha racte rized by t he 2 x 2 m atrix
c os -sin R v() = .
s in cos
The grou p of symme try t ransformat ions that leave s an equil atera l
tri angle inva riant unde r rot ations in the p lane is of orde r thr ee, a nd ea ch
element of th e gro up is of d imension t wo
Gn ~ R 3(2) = {R (0), R (2 /3), R (4 /3)}
= 1 0 , -1 /2 -3/2 , -1/2 3 /2 .0 1 3/2 -1/ 2 , -3 /2 -1 /2
{ 123, 312, 231} = C3.
The se ma trice s for m a 2 -dime nsional re prese ntati on of C3 .
A 3 -dime nsional re prese ntati on of C3 ca n be obtained a s follows:
Con sider an e quila teral tria ngle located in the plane and let t he
coo rdina tes o f the thre e ver tices P1[x, y], P2[x , y] , and P3[x , y] be
wri tten as a 3-vector P13 = [P1, P2, P3], in no rmal order . We intr oduce
3 3 matri x ope rator s Di(3) th at ch ange the o rder of th e elements ofP13,
cyc lically. The identity is
P13 = D1(3) P13, w here D1
(3) = diag (1, 1 , 1).
8/14/2019 Introduction to Groups
85/164
The rear range ment 85
P13 P23[P3, P1, P2] i s giv en by
P23 = D2(3) P13,
whe re
0 0 1 D2
(3) = 1 0 0 ,0 1 0
and the rearr angem ent
P13P33[P2, P3, P1] i s giv en by
P33 = D3(3) P13
whe re
0 1 0 D3
(3) = 0 0 1 .1 0 0
The set of ma trice s {Di(3) } = {D1
(3) , D2(3) , D3
(3) } i s said to form a 3-
dim ensio nal r epresentat ion of the original 2 -dime nsional re prese ntati on
{R 3(2) }. The elements Di
(3) ha ve th e sam e gro up mu ltipl icati on ta ble a s
tha t ass ociated wi th C3.
9.2 The m-di mensi onal repre senta tion of sy mmetr y
tra nsfor matio ns in d-di mensi ons
Con sider the case in wh ich a grou p of order n
Gn = {g1, g2, . ..gk, . ..gn}
is repre sente d by
8/14/2019 Introduction to Groups
86/164
R n(m) = {R 1
(m) , R 2(m) , . ....R n
(m) 86
whe re
R
n
(m)
~ Gn,
and R k(m) is an m m matri x rep resen tatio n of gk. Let P1d be a ve ctor in
d-d imensional spac e, wr itten in n ormal orde r:
P1d = [P1, P2, . ..Pd],
and let
P 1m = [P1d, P2d, . ...Pmd]
be an m- vecto r, wr itten in n ormal orde r, in whic h the comp onent s are each
d-v ector s. I ntrod uce t he m m matri x ope rator Dk(m)(gk) s uch t hat
P 1m = D1(m)(g1)P 1m
P 2m = D2(m)(g2)P 1m
.
.
P km = Dk(m)(gk)P 1m , k = 1 to m , the numb er of
symme try o perat ions,
whe re P km is the kth ( cycli c) pe rmuta tion ofP 1m , and Dk(m)(gk)
is call ed
the m-d imensional repr esent ation of g k.
8/14/2019 Introduction to Groups
87/164
Infinite ly ma ny re prese ntati ons o f a g iven repre senta tion can b e 87
fou nd, f or, i fS is a ma trix repre senta tion, and M is any defin ite m atrix
wit h an inver se, w e can form T(x) = MS(x)M-1, x G. Sin ce
T(xy ) = MS(xy )M-1 = MS(x)S(y)M-1 = MS(x)M-1MS(y)M-1
= T(x)T(y),
T is a re prese ntation of G. The n ew re prese ntation si mply invol ves a
cha nge o f var iable in t he co rresp ondin g sub stitutions. Re prese ntati ons
rel ated in th e man ner o fS an d T ar e equ ivale nt, and a re no t reg arded as
dif feren t rep resen tatio ns. All repres entat ions that are e quivalent to S ar e
equ ivale nt to each othe r, an d the y for m an infin ite c lass. Two equi valent
rep resen tatio ns wi ll be writ ten S ~ T.
9.3 Dir ect s ums
IfS is a re prese ntati on of dime nsion s, a nd T is a re prese ntation of
dim ensio n t o f a g roup G, th e mat rix
S(g) 0 P = , (g G)
0 T(g)
of dimension s + t is c alled the dir ect s um of the matri ces S(g) and T(g),
wri tten P = ST. There fore, give n two repr esent ations (th ey ca n be the
sam e), we can obta in a third by a dding them dire ctly. Alt ernat ively , let P
be a rep resen tatio n of dimension s + t ; we suppo se th at, f or al l x G, the
mat rix P(x) is o f the form
A(x) 0
0 B(x)
8/14/2019 Introduction to Groups
88/164
whe re A(x) and B(x) are s s and t t matri ces, respe ctively. (The 0s 88
are s t and t s zero matri ces). Def ine t he ma trice s S an d T as foll ows:
S(x) A(x) and T(x) B(x), x G.
Sin ce, b y the grou p pro perty , P(xy ) = P(x)P(y),
A(xy ) 0 A(x) 0 A(y) 0=
0 B(xy ) 0 B(x) 0 B(y)
A(x)A(y) 0= .
0 B(x)B(y)
The refor e, S(xy ) = S(x)S(y) and T(xy ) = T(x)T(y), so that S an d T ar e
rep resen tatio ns. The r epres entat ion P is said to b e dec ompos able, w ith
com ponen ts S an d T. A rep resen tation is indecompos able if it cann ot be
dec ompos ed.
If a com ponen t of a dec ompos able repre senta tion is it self
dec ompos able, we c an co ntinu e in this manne r to decom pose any
rep resen tatio n int o a f inite numb er of inde compo sable comp onent s. ( It
sho uld b e not ed th at th e pro perty of i ndeco mposa blity depe nds o n the fiel d
of the r epres entat ion; the r eal field must somet imes be ex tende d to the
com plex field to c heck for i ndeco mposa bilit y).
A w eaker form of d ecomp osabi lity arises whe n we consi der a
mat rix o f the form
A(x) 0 P(x) =
E(x) B(x)
8/14/2019 Introduction to Groups
89/164
whe re A(x), and B(x) are matri ces o f dimensio ns s s and t t 89
res pectively and E(x) is a matr ix th at de pends on x , and 0 is the s t zero
mat rix. The matri x P, a nd an y equ ivale nt fo rm, i s said to be red ucibl e.
An irr educi blerep resen tatio n is one t ha