DEGENERATE REPRESENTATIONS OF THE SYMPLECTIC GROUPS …streaming.ictp.it/preprints/P/66/077.pdf ·...
Transcript of DEGENERATE REPRESENTATIONS OF THE SYMPLECTIC GROUPS …streaming.ictp.it/preprints/P/66/077.pdf ·...
i (*> i * # i -m —t
INTERNATIONAL ATOMIC ENERGY AGENCY
INTERNATIONAL CENTRE FOR THEORETICAL
PHYSICS
DEGENERATE REPRESENTATIONSOF THE SYMPLECTIC GROUPS
I. THE COMPACT GROUP Sp(n)
P. PAJASAND
R.
1966PIAZZA OBERDAN
TRIESTE
IC/66/77
INTERNATIONAL ATOMIC ENERGY AGENCY
INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS
DEGENERATE REPRESENTATIONS OF THE SYMPLECTIC GROUPS
I. THE COMPACT GROUP Sp(n)t
P. PAJAS*
and
R. RACZKA**
TRIESTE
June 1966
' To be submitted to Proceedings of the Royal Society (London)
* Present address: Nuclear Research Institute, Rez, Czechoslovakia
** On leave of absence from Institute of Nuclear Research, Warsaw, Poland
DEGENERATE REPRESENTATIONS OF THE SYMPLECTIC GROUPS
I. THE COMPACT GROUP Sp(n)
I. INTRODUCTION
Many attempts have been made in the last few years to understand
the properties of physical systems such as elementary particles, the
hydrogen atom, nuclei etc., using the theory of representations of
the underlying symmetry group. The main effort was devoted to the rota-
tion and unitary groups, while the class of symplectic groups did not receive
much attention. This, may be due to the peculiar property of these groups
of conserving an anti-symmetric bilinear form.
Some interest in symplectic groups was raised by remarks of LIPKIN
[1] on possible applications of the group Sp(n#R) to systems of bosons which
do not conserve the number of particles. BUDINI [2] has pointed out that,
using Sp(6, 6) as a higher symmetry group.it is possible to obtain a mass
formula for elementary particles without symmetry breaking. The quest-
ions of symplectic symmetry of hadrons and of the embedding of the harmonic
oscillator in the symplectic group have been discussed in [3] . On the other
hand, in a series of papers [4], the theory of the degenerate representations
of the rotation and unitary (both compact and non-compact)groups, have
been developed. In this work we present the extension of that approach to
the unitary symplectic, groups, i. e., those which conserve both symmetric
and anti-symmetric bilinear forms.
In general, the irreducible unitary representations of a semi-simple
Lie group G are realized as mappings of a Hilbert space yj. (/Cj into itself,
the domain of corresponding functions being some homogeneous space X
of the type
X - 7 G O (1-1)
when G is a closed subgroup of G .
GEL'FAND [5] has proved the important theorem which states that the
number of independent invariant operators in the enveloping algebra acting
- 1 -
in the Hilbert space of functions 'iv ( X) with domain X is equal to the
rank of the space X (and is therefore independent of the rank of the
fundamental group G ), Since we are primarily interested in construction
of representations characterized by the minimum number of invariants, we
can use this theorem to select an appropriate domain A , namely that of
rank one.
In order to select the proper invariant operator we can use the theo-
rem of HELGASON ([6] , Chap. 4, p. 397) according to which the ring of
invariant operators in the algebra (rL of the group Gr > realized on the
space of rank one, is generated by the Laplace-Beltrami operator.
A (X) - -=L= \ 2^ (X) vlfT Pp o- 2»
on X . Here a, M X) is defined by
where q, (X) is the metric tensor on the space X and |£H- I «st [ i f iWn
The operator (1. 2) is actually equal to the second order
Casimir operator Q = %;.• Z ^ ( ??'<£. (y^j of the group
G , provided that the metric tensor Ckj<? ^X) °^ n e space /\ is induced
by the Cartan metric tensor (1. . of the algebra S\ of u , (see [6] , Chap.
X, p. 451).
Then the problem of construction of the most degenerate irreducible
unitary representations is reduced to the problem of determining eigen-
functions and eigenvalues of the Laplace-Beltrami operator on the appro-
priate symmetric space A~ &/&aof rank one.
We select a suitable domain X and solve the eigenproblem of the
Laplace-Beltrami operator on it in Section II. Section III is then devoted
to the study of the properties of the most degenerate representation of the
group Sp(n) obtained in this way. In Section IV we discuss some aspects
of the determination of the series of less degenerate representations of
Sp(n) characterized by two independent numbers. Thus in the present
- 2 -
• : - • ? • ; ! * • • ft *
paper we shall deal only with the case of the compact group Sp(n). We
shall, however, use the results obtained here in forthcoming papers, in
which we would like to solve the following problems:
(i) The construction of a representation space for (most) degenerate
irreducible unitary representations of the non-compact unitary symmetric
group Sp(p, q) determined by a discrete or a continuous invariant.
(ii) The decomposition of the tensor product of two representations of
Sp(p, q) group into irreducible components and the decomposition of the
irreducible unitary representations of Sp(p, q) with respect to compact and/
or non-compact subgroups.
(iii) The investigation and construction of infinite-dimensional represent-
ations of the compact group Sp(n).
n. CONSTRUCTION OF THE REPRESENTATION SPACE
According to GEL'FAND's theorem [5] , the properties of the ir-
reducible unitary representations of a group Cr realized on a Hilbert
space Ti (X) are determined by the geometrical properties of a domain ) \
of functions -V (X) € <3( iX), the domain X being some homogeneous
space.
Symmetric spaces of the type (1.1) with a compact stability group £ro
have been classified by E. Cartan, whereas those with non-compact sta-
bility group have been listed by ROSENFELD [7] . We reproduce in Table I
Cartan1 s list of symmetric spaces [6] for the fundamental group Gr of the
symplectic type. There are also collected the spaces from Rosenfeld's
list, together with their ranks arid dimensions.
- 3 -
Table I
Homogeneous spaces connected with symplectic groups.
Cartan's List
Go - compact.
p n/uu)
Alin)
Rank of A
a
71
~ »
Dimensionof/
71 (W)
Rosenfeld's List
Go - non-compact
X - Gr/&. Rank of X
pn
/
Dimensionof X
X Cti'-O1'*-}
We see that the only suitable candidate for a space of rank one on
which the compact group Sp(n) acts transitively is the space
(2.1)
This space is known to be a quaternionic protective space ([6] ,[7]). But
it is rather difficult to construct a convenient and simple geometrical model
for it. Fortunately, we may use for our purposes the space
(2.2)
which is evidently closely related to the space (2.1). Furthermore, the
space X i s isomorphic to the unitary sphere in the n-dimensional quater-
nionic unitary spaceit
t defined by the equation
%(2.3)
It has been proved by CHEVALLEY [8] and HSIEN-CHUNG [9] that the
group Sp(n) <g>Sp(l) acts transitively on (2. 3) and that its stability group is
Sp(n-l)g>Sp(l).
As is well known, the non-commutative algebra of quaternions GL is
defined as an algebra of dimension 4 over the field g? of real numbers with
a base composed of four elements 1, i, j , k whose multiplication table is
t
I
i
11
i
1 6L' 4
- i {<
J -i
Uk
t
7
-I
(2.4)
Then any quaternion 0 € ( ^ may be expressed either in the form
- 5 -
where X{ (^» V"A)axe real numbers or
I ' ^ + ZzJ (2.5')
where Z{ fl= 1,1) a re complex numbers.
The quaternionic conjugation is the mapping
*t*i£ (2.6)
of Q. into itself. For a detailed treatment of properties of the body of
quaternions as well as for the questions about the relation of symplectic
groups to the vector spaces over the body of quaternions see , for example,
the book of CHEVALLEY[8] .
It is important that the n-dimensional quaternionic unitary sphere
(2. 3) is homeomorphic to the usual sphere in 4n-dimensional Euclidean
space (f\ . Because of it, its properties are rather simple.
Now let us introduce an inner co-ordinate system on the sphere (2. 3).
Let us suppose that we have defined a co-ordinate system on the quater-
nionic unitary sphere (2. 3) of dimension j> < 7t . Let these co-ordinates
be denoted by ^ ( ii - ^ / P ) • Then the co-ordinate system on the
"sphere" of dimension •[> + 1 will be defined by
(2.8)
Now, starting from
= (e^ux^ + e^W^j) (2.9)
we get the co-ordinate system for an arbitrary dimension of the quaternionic
unitary sphere (2. 3) using recursive formulas (2. 7) and (2. 8). This choice
is convenient because there appears in brackets in (2. 8) and (2.9) a general
expression for a quaternion of modulus equal to one. The ranges of varia-
bles Vfc , Vk , $K and 0 k must be chosen so that the co-ordinates (2.7)
cover the space ^ *(•«--1 only once. In this way we introduce on the
space X ^ " 1
- 6 -
n variables € [O,27) C W * - I , ...,
n
n
" •
and (n-1) 2, . . . , 1* )
i . e . , ^n-1 variables altogether.
The metric tensor Qj& (X* ) induced by the metric tensor
of the quaternionic unitary space Q f x ' is given by the symmetric part of
the tensor defined by
f
where
and
ftIn our parametrizatxon the metric tensor <X^» (^ / is diagonal and
therefore the Laplace-Beltrami operator (1. 3) can be represented in the
form:
(2.12)
k(<*-*\~)
where
/ J I• /
and A(X n" '" ) is the Laplace-Beltrami operator on the quaternionic unitary
sphere embedded in the space Q; . For n=l we have
-7-
To find the basis functions for the Hilbert space c* (X) on which the r e -
presentations of the Sp(n) group may be realized, we have to solve the
equation
)
where SL stands for the set of variables {^v ••• > 51* n r . Representr
ing solutions V (SlM) of (2.15) in the form
we obtain the set of ordinary differential equations of second order
(2.19)
(2. 20)
(2. 21)
General solutions of equations (2.19) and (2. 20) a re given in terms of hyper-
geometrical functions as follows
(2.22)
(2.23)
- 8 -
where the eigenvalues wn and h^ are integers and the spectrum of
remaining eigenvalues is given by
(2,24)
and
(2,25)
with positive integers J^ and L .
These solutions are square integrable with respect to the measure
if the following restrictions on eigenvalues are satisfied::
and
L ^ °.V--> L "? I
(
(2.27)
The solutions (2.22) and (2.23) are expressible in terms of the usual d-
functions of the theory of angular momenta [10] . Then the eigenfunctions
of (2,15) are explicitly given by
^
rr<J
where
(2.28)
(2. 29)
and
(2. 30)
- 9 -
The normalization constant NA is then
N * T i n ( V v n 6£k.H)(Ll<+-,2li-'f) . (2.31)
The functions (2, 28) with a given value L = Ln are square integrable with
respect to the left-invariant measure
(2.32)
on the domain X . The explicit expression for the measure djd. is
C
wL L I (2.33)Hence, the set of functions J — (J»i- ) span the Hilbert space
0** i ) c
ji [ A /with the scalar product related to the left invariant measure
(2. 30) by
(2.34)
for any V /C £ OV \A ) . In fact, the space "% (X J i s a
representation space of the group Sp(l) (50 Sp(n) which occurs as a fundament-
al group of the space (2. 2). However, a closer study of the properties of
the Lie algebras of Sp(l) and Sp(n) groups reveals that we can realize ir-
reducible unitary representations of Sp(n) on certain subspaces of the
Hilbert space
HI. THE MOST DEGENERATE REPRESENTATIONS OF Sp(n)
1° Structure of the Lie algebra
The group Sp(l) (g)Sp(n) which acts on the manifold {2. 2) is a direct
product of two simple groups. Therefore its Lie algebra (R decomposes in-
to two commuting subalgebras which we call VS} and i)\^ , respectively.
-10-
The algebra dv^ of Sp(n) Is then formed by the K (a* + 1) generators
which have the symmetry properties
K- H<> <-V< — V--^.: (3-2)The commutation relations of these generators are
( 3 - 4 )
We have collected in Appendix I the explicit expressions for the generators
(3.1) as linear differential operators in quatemionic and complex variables
as well as their connection with the generators of the group Ci.C2.iy.
The algebra (R ± of Sp(l) is generated by the three operators Wg
(X = 1* 2, 3) which have on the manifold A the form
(3-6)
There is a close relation between the operators U.^ and 1 / ^ and UL^k
and XTjck . It is easiest to see from expressions {AI# 22) of Appendix I
which define the generators in te rms of complex variables. The meaning
of the tilde in eqttations (3. 6) then consists in the substitution z* *=z> *•*,
for only K=- - lf •••, - K- , while remaining variables a r e unchanged
( z k - ^ z k and z K - ^ z k for k = X} ••- ? ^ ) .
The commutation relations for the generators (3. 6) are:
j w ; ft-*i«.^ 0.7)
-11-
Throughout this paper we often use, instead of (3.1) and (3. 6), the set of
generators of the complex extension of the real Lie algebras (xK and *Kt
These are especially convenient when dealing with the basis functions (2. 23}
and can be normalized in such a way that they form the Weyl's standard
basis. We define these operators by
( 3 - 9 )
- ' i r t \ (3.10)
and
In Appendix I we give their commutation relations and their explicit form
x in the parametriz-
ation (2.10).
2 Properties of the generators.
As we are using the quaternionic unitary sphere (2. 3) instead of the
quaternionic projective space (2.1) we must be aware of the fact that the
irreducible unitary representations of the group Sp(l) (g)Sp(n) are directly
realized on the space ** ^ h /spanned by functions (2. 28). Nevertheless,
the space "K- (X ) should be reducible with respect to the action of the
group Sp(n). To show this, we use the formulas for the action of generators
of algebras ^ K and KL of Sp(n) and Sp(l) on the basis functions (2. 28),
respectively. They are collected in Appendix II and one can easily see
that the generators have the following properties:
(i) The generators H {p^l, . ^ n ) form the Cartan subalgebra of (k K
and are diagonal in parametrization (2.10). They have the eigenvalues
Mp * r*f. + wtf, (3.12)
-12-
•r ft -s ; . : « r t . ! S > H i ! - . 't & ••'- ii. '.:•-*
(ii) T h e g e n e r a t o r s *~j£?(Prl>'••? a ) c o n s e r v e a l l n u m b e r s !_;. a n d J?'L
a n d a l s o t h e v a l u e of
KA- _— (3.13)lp - mp p,
(iii) All remaining generators conserve the value of
The last property is simply a consequence of the fact that the generator wz
of Sp(l) which has the eigenvalue (3.14) commutes with the algebra K *.
of Sp(n). Therefore, the space A- ( / ) spanned by functions (2.28)
decomposes into subspaces ov ^- C * / o f simultaneous eigenfunctions
of the Laplace-Beltrami operator /X fX *" ) and of the generator *Wj
of Sp(l).
Now the value of M~ is restricted by the conditions (2. 26) and (2.27)
so, that
/ M l ^ L (3-15)
The structure and properties of subspaces <p- ^~ (/ Jstrongly depends
on the value of M . In the case when JM \ - L the subspaces <J[+L(/( j
are irreducible under the action of the group Sp(n) and therefore they can
be considered as representation spaces for a unitary irreducible represent-
ation of Sp(n). Because these representations are characterized by a single
number L , we call them most degenerate representations. They will be
treated in detail in this section. In the case when I Mi ^ L the situation
is not so simple. The space J\. u- {.A J is in this case reducible
with respect to the action of the algebra of Sp(n) and to obtain its irreduc-
ible components one needs further investigation. We have devoted Section
IV to these questions.
3° Unitarity and irreducibility of the most degenerate representations
of Sp(n).
The condition H - + L reduces the two sets of equations
(2.26) and (2.27) to
-13-
- vnf) . tp (p*^...,*) (3.16)
and
V * +#p - / Cp'2y -,*•/3 (3.17)
These are strict conditions on the eigenvalues and they select from the
set of eigenfunctions (2. 28) of the Laplace-Beltrami operator the subset of
functions
(3.18)
Here we have introduced the notation
and
^'i^^^i (3.20)
In the special case / * d we have
We have also put L = L^ .
In the considerations which follow a key role is played jjy the second order
invariant operator !„ which is proportional to the second order Casimir
operator of the group Sp(n). We have found the following connection of
this operator with the Laplace-Beltrami operator A (X "~ j and the^ til
second order invariant operator ~LA of the Sp(l) group which enters the
direct product Sp(l) H Sp(n):
-14-
(3.22)
Here we define
l r = t HZ - 2 [E., f^ + f E ] (3. 23)OC>O
and
ii* (3-24)
while /\ is defined by equation (2.12).
The functions (3.18) are simultaneous eigenf unctions of the operators/\ and jT^ and therefore also of J ^ , the eigenvalues being
and L(L+£n} ^respectively. They span a Hilbert space
*1"' ) d e f l n e d by ^ e scalar product (2.31) with the left invariantmeasure (2.33).
a) Unitarity
The space *K> L ( X / creates a representation space for the
group Sp(n) because for any generator Z ^ £ R^ and any f e %L (A
we have
We shall denote representations of the group Sp(n) related to this space by
DL (Sp(n)) or simply D . They are realized by associating to any
S ( ) t X i n ^ " ' ) uch thatelement Q € Sp(n) an operator X in nL\.^"' ) such that
(3.26)
-15-
for any f { -><L) - ^ ^ " * V>/ ">o ^ ' /W/ ( ^ - J of
JVL^' J« Here i i, is a point of the manifold X and a^JZ is itsleft translation by the element oT1 of Sp(n). - Then unitarity of re-
presentations DL follows immediately from the left invariance of themeasure d,jx(SLj .
b) Irreducibility.
From the explicit form {3.18) of the eigenfunctions I u+ we see
that the structure of the Hilbert space ^> L (. X / i s relatively simple.
Namely, we can decompose >i L (X ) in the direct sum of subspaces as
follows:
where the summation over L^.^and fll goes through
and
(3.28)
(3.29)
/ +respectively. Each of the spaces %/ ^+(X "'') forms a represent-
ation space for an irreducible unitary representation of the Sp(n-l) sub-
group of Sp(n). We see that any representation of the maximal subgroup
Sp(n-1) occurs only once in the decomposition (3. 23). This is illustrated
diagrammatically in Fig. 1.
ftl,
To each point ( L ^ > $K) in the diagram Pig. l.a) there corresponds a
diagram Fig. l.b) which gives the possible values of *Hn and ttiK for a given
* * •
Now, to prove the irreducibility of representations D L of Sp(n) it is
sufficient to prove that, starting from any point in diagrams a) and b) of Fig, 1,
we can by successive application of the generators o? the algebra (\«. of
Sp(n) reach any other point in these diagrams.
It is simple to prove this for Sp(l) (\ - l) because the algebra of Sp(l)
is formed by the three generators H«g» and H ; the former act as step
operators while the latter is diagonal. The presence of the "stopping"
factor / ( X ? m T i T in the formula (All. 3) of Appendix II for the action
of the generators E -o'e* assures us that starting from some point of Fig. 1,
diagram b) we can reach any other point on the diagranxand only these points.
This means that the irreducibility of representation Du ^ 1_ = J?1) of Sp(l)
is proved. Similarly one proves the irreducibility of the most degenerate
representations for the Sp(2) and Sp(3) groups. Now let us assume that we
have proved the irreducibility of D L for the group Sp(n-l). It means that
the spaces d< i ^+ ( * ) are irreducible with respect to the action of the
generators of the algebra MV^^L of Sp(n-l).
Now, to prove the irreducibility of /V L (_ A ) space with respect
to the actions of the algebra ^ ^ of Sp(n) it is sufficient to consider only
the action of generators E + ae? > ^^v-^n- i a n d ^ ^e1 + ^v-i '
(See also Appendix II). The corresponding formulas are formulas (All, 11-
14) and (All. 3) of Appendix II. We see that the considered generators con-
serve the conditions (3.16) and (3.17), Moreover, the stopping factors are
always combined in such a way that relations
4 > O ; Ln.L >oand
/ rnn + m» U 4 ; I *.., + v^n., I ^ L ^
are fulfilled.
This procedure enables us to prove the irredueiMlity of representations
X> L of Sp(n) by full induction.
-17-
4° Properties of the most degenerate representations of Sp(n).
/up L / w l/n-f\The basis functions (3.18) of the Hilbert space A\.,, (A Jare
characterized by n numbers L* and n numbers Hjf O'-V-/*?)
The former are related to the eigenvalues of the set of the second order4 (i)
invariant operators -ly> of the chain of subgroups
p p p (3.30)
These eigenvalues are
Numbers M/" are then eigenvalues of the generators -H& of the Cartan
subgroup of Sp(n).
Therefore, the set of commuting operators is explicitly formed by
hl~2<n (2.32)
operators:
The number (3. 32) is reasonably small compared to the corresponding
number in the case of a non-degenerate series of representations of Sp(n),
where it is
This may be. of particular interest from the point of view of physical
applications, because usually we would like to have the smallest possible
number of invariants for characterization of a given physical state. The
fact that all Cartan subgroup generators are diagonal,which is due to the
parametrization we have employed, is also convenient for it makes it
possible to relate each of these generators to an additive conservation law.
Another property which is a direct consequence of the choice of the
parametrization of the domain X is the pattern of the decomposition
-18-
of the Hilbert space dV L ( . / ) into the aubspaces o\-/_ ^ (A
on which the subgroup Sp(n-l) acts irreducibly. One can easily find a
parametrization in which the representation 2)L of Sp(n) decomposes in
a similar way with respect to a subgroup Sp(n-k) with arbitrary k (*" ^ £
In Appendix III we give the detailed calculation of the highest weight
of the representation of Sp(n). From it follows, that the representation
T>u corresponds to the representation D(L, (?,...,£?) in the notation used
in 111] , or it can be represented by a one-row Young tableaux (see [12] )
and therefore may be interpreted as a fully symmetrical representation
of Sp(n).
More detailed examples for the lowest representations of Sp(2) and
Sp(3) are given in Tables I and II of Appendix IV.
IV. THE LESS DEGENERATE REPRESENTATIONS OF Sp(n) AND THE
REPRESENTATIONS OF THE GROUP Sp(l) H Sp(n),
1° The less degenerate representations of Sp(nX
The condition
| M " | - wwsi < L
imposes a restriction on the eigenvalues, which can be conventionally
written as
But from (2.27) it follows that
|H"| 4 L -il ( 4 - 2 )
where now k may be one of the run
We can again represent the possible values of L^y and ** by points in
the net of Fig. 2
- 1 9 -
Fig. 2
From the formula (AII#13) of Appendix II we see that the operator J-Y and
therefore also the second order invariant operator J» of Spfn) are not
diagonal on the space <A M_ ( / ) . Generally the operators xi and
Tn when acting on functions f2. 28) conserve, besides all the numbers L^
and ^A ( ^ V ' i i ) > also the numbers M,,4" (/> =• ^ •• ,n) and M
but generally they change the values of MA 0>= I_, , -., ^ ) . The last
numbers are eigenvalues of operators f> which do not belong to the set
of commuting operators of the algebra (jln of Spfn). Therefore, it is
necessary to diagonalize the operators -i- and -Ln on the subs paces
3^ ufu ( X "") ' This leads to the usual procedure of diagonalizationHA i"~
of matrices by which the generators are represented on a given space.
In order to see the structure of the space <jl ^-{A /let us
first investigate the points for which Ln_t •= O . Then we have all other
numbers equal to zero except / h and M^ . The value of M^ is fixed
by
<4 -4 )
There fore the condition (4. 4) defines the function Yul i"''°J *'°'"'
The eigenvalue of X will then be
(4.5)
-20-
This means that functions with Lr)_i » 0 and different values of -U in
- (Y *"') belong to different irreducible representations of Sp(n).
We denote the eigenvalue of X, by
we get
t = I M I ; I H - I
and then from (4. 5) and (4.1)
., L (4.6)
Now, taking into account the generators of the algebra 31^ we see that
to the same representation as the point ( L , O ) in diagram
must also belong the points ( L i 4-} "I j . Proceeding in this way we get
the following decomposition of the space ^f [Y n / into the sub-
spaces which are irreducible under the action of Sp(n):
L
This is illustrated diagrammatically in Fig. 3
M0L
Fig. 3
Of course, the set of functions which span the space t
by the diagonalization of the operator Jf in all subspaces
Therefore, the function £f M// fi> * ^
operators:
obtained
kkyis the simultaneous eigenfunction
of the set of
and is generally expressed as a certain linear combination of functions
-21-
pwhich span the space <?v H * n-
On the spaces <Kr ^-(X "') one can introduce the irreducible
unitary representations of the Sp(n). Unitarity of these representations
can be proved in the same way as it was in the case of the most degenerate
representations.
The highest weight of a representation 3>£- realized on the space
~ K (X J * s c a l c u l a ted *n Appendix III. Then the representation
£. can be denoted according to the notation of [11] by
,<>(4.9)
which corresponds to the Young tableaux with two rows defined by the
symmetry scheme
-— {~*vl
JThe representations realized on spaces and
are equivalent because they have the same highest weight.
The diagram Fig. 4
T
(4.10)
where
Fig. 4
- 2 2 -
corresponds to the pattern of decomposition of the representation T)^
of Sp(n) in respect to the irreducible representations of the maximal sub-
group Sp(l) H Sp(n).
In Appendix IV we give detailed examples of lowest representationsf '
of this series (which we call less degenerate) of the groups Sp(2) and Sp(3).
2° Representations of Sp(l)(x)Sp(n).
Finally, we should remark that the full set of functions (2. 28) span a
Hilbert space % (X^^ ) which acts as a carrier space for irreducible
unitary representations of the group Sp(l)<3£>Sp(n).
on L / y l ( .H -3 \
The corresponding decomposition of the Hilbert space av \ A /
into subspaces is then given by
/ / — ( "* * W•'•/l^i ^l( Wj, VJ^ * • / #
(4.11)
where the summation is restricted by conditions (2.26) and (2.27). There
is no need to repeat the same kind of consideration as above to prove the
irreducibility and unitarity of the representations DL of s'p(l)<g>Sp(n) on
this space. Since Sp(l) GDSp(n) is a direct product and we have treated
Sp{n) in the previous sections, it is only necessary to use generators from
both (j V and £K algebras of Sp(l) and Sp(n), respectively, to prove the
irreducibility of representation D . Representations DU of Sp(l)®
Sp(n) decompose into the representations of the subgroups Sp(l) and Sp(n)
according to the formula
D . ,(4.12)
as may be easily verified.
- 2 3 -
V. CONCLUSION
The main results can be summarized as follows:
(i) A set of harmonic functions which span the representation space
for the most degenerate unitary irreducible representations of the compact
symplectic group Sp(n) has been found. These representations are charcter-
ized by a single number which is related to the eigenvalue of the second
order Casimir operator of Sp(n). They can be also described by a one-
row Young tableaux.
(ii) Besides this, the series of unitary irreducible representations of
Sp(n) characterized by two independent numbers L and L has been
obtained. These numbers are related to the eigenvalues of the Laplace-
Beltrami operator A (X " 7 on the guaternionic unitary sphere X
and of the second order Casimir operator of Sp(n), respectively. To these
representations corresponds the Young tableaux with two rows.
(iii) The number of operators in the maximal set of commuting operators,
the eigenvalues of which characterize the basis functions of the represent-
ation space, is in the case of the most degenerate representations
A/,-in (5.1)
and in the case of the less degenerate representations it is
^ , \ n ^ (5-2)
These numbers are sufficiently small when compared to the corresponding
number in the case of non-degenerate representations of Sp(n);
This underlines the importance of the degenerate representations for
physical applications.
(iv) In the parametrization introduced On the quaternionic unitary sphere
A , the generators n^ which form the Cartan subalgebra are all
diagonal which is again useful for applications.
-24-
,.i. i,. ".,.: • ,
(v) Finally, the patterns of the decomposition of the given irreducible
representation of Sp(n) with respect to the maximal compact subgroup Sp(l)
63 Sp(n-l) of it,occur due to the parametrization.
ACKNOWLEDGEMENTS
The authors would like to thank the IAEA and Professors Abdus
Salam and P. Budini for the hospitality extended to them at the International
Centre for Theoretical Physics in Trieste. It is a pleasure to thank
Dr. J. Fischer for many helpful suggestions and Drs. J. Niederle and
A. Bartl for interesting discussions as well as Dr. R. Anderson for kindly
reading the manuscript.
-25 -
APPENDIX I
Algebra of Spfn) and of Spfl)
In order to obtain the algebras of Spfn) and Sp(l) groups, let us
consider an infinitesimal symplectic transformation in the unitary n-dimen-
sional quatemionic space Q. *
A"
The necessary and sufficient condition for the n x n matrix of quatemionic
elements A to be symplectic is
A A4 - A4A - X (AI.2)t r
(i'
where
We point out that the dagger here represents the quatemionic conjugate defined
in Section JII, and the transposition of matrix A . Representing any element
of A in the form
and considering the real parameters a s t , & s t 1 cs(- and d. . ^.S;t = j v . . .
as infinitesimally small quantities, we obtain,imposing condition (AI. 2)7
(AI. 5)
(AI.6)
or _
A = ^ ^Let us now define four formal.linearly independent quaterBionic quantities
- 26 -
W l ^ H ' i - H • - ! • • • * ^
' 7 )
, •<
and define the representation of the group Sp(n) as a transformation in the
space of functions of these variables determined by
for any A € Sp(n). Defining the formal derivatives by
J (AI. 9 )
we can expand the right-hand side of equation (AI.8) into a Taylor series.
Then the generators of the Sp(n) group are obtained if we consider the one-
parameter subgroups of Sp(n). This procedure leads to the following set
of generators:
- 27 -
(AI.10)
Here we have introduced the notation that for example
represents the expression
t +
The commutation relations (3.3) - (3.5) of the generators w f and U^,
which are given in Section III, are easily verified using the definition (AI.9)
of quaternionic derivatives, or expressing generators UJ^ and T££ in
variables <£, # £,. #* and jt* as follows:
j
* ^ 7
- 2 8 -
Now we may express these generators through those of U(2n) given in [4]
as
* *A (AI.12)
We see that
Because the algebra %, of the Sp(n) group is the compact real form of the
algebra of (the complex group un ,we can easily find the Weyl's standard
basis K* , E tae > £*lp +F » -i$'$%°t ^*\ by using the generators (AI.10),
This relation is given by formulas (3.8) - (3.11) of Section HI.
The commutation relations of these generators are
J - /v t f i e , ^ (Ai.i5)
where oi> are roots and N^A = ^ .^ ^ 0 only if ot + £ is also a root.
We obtain a Dynkin diagram
-29-
of the group Ch tfwe put
• (AI.16)
where &i > **" ? **- are unit Vectors which form an orthonormal basis in
the root space.
Generators £ + * play the role of raising and lowering operators
when acting on the basis functions (2. 28) of the space #i (X :) while the
generators H^ form a Cartan subalgebra. In the following, we present
explicit expressions for these generators as linear differential operators
in the parametrization of the manifold X ^ *~X we have introduced in
Section II. Similarly, the three generators of CL which correspond
to the Sp(l) group are defined by
W ' " (AI.17)
where »
* (AI.18)
In parametrizationization of X "*v"i we now have
and for t < Jt we have
-30-
(AI.21)
and
E * 2 *
t 9
- 3 1 -
Here
A(AI.23)
and
Oat OJL fr>\*I-I
-X<l
— c<s<>(AI.24)
The algebra of Sp'(l) in Sp'fl) <g>Sp(n) is then formed by three generators
(AI.25)
and
(AI.26)
It is rather easy to show that these generators commute with the algebra of
Sp(n).
- 3 2 -
APPENDIX II
Actions of generators on basic functions.
To prove irreducibility of the representations D (X ) and
' we need explicit formulas for the action of different generators
of the algebras Rj_ and <RK of Sp(l) and Sp{n), respectively. However,
we do not need to know this action for all generators of Sp(n) because if we
know, say, the action of E ^ + g» , Ej, 7g» and Eyg1 for any
b= a} ..., *\ , we can use the commutation relations (AH5) to get, for example,
\g + £• for any k< t> • For this reason we give in this Appendix the
formulas for the action of the generators E+<f +•£ , Et » _•£» and E+^
on the basis functions (2. 26).
As may easily be seen from the explicit expressions of the genera-
tors in Appendix I. the generators ,Et^ do not change either values
Lt ((c = 3,...,ti) or ^ K (K= ij • -•."), while £.5* . ? and £4^ _£» change
in these sets of eigenvalues only L,,,^, x^ and JJ*,_j_ . All the generators,
except those of H ^ which form the Cartan subalgebra, change the values of
*v\ . Tvt i.. Av\. . andi«. , . Therefore, we label the eigenfunctions in what
follows only by the eigenvalues Lp=L , Z^.j * i! $.a - -t , fj>-i £ £ ,
t __ , —_— -t—/ . Thus we have:
Hp
LJ VL'L')fi£i' r -^ v ^ ' / V ' ( A I L 2 )LJ Vnp 1 / |_
L,L-A tl\ 4 t—• r v L>L>)£'£;
^ {An.4}
-33-
{AH. 5)
7
( t ± ,* y,LliteL
-34 -
Here (AH. 6)
(All . 7)
(AH. 8)
^C
The coefficients df and d* are the same as C* and Cj , respectively,
if one substitutes Jt*, J and K' for i , m and * , respectively.
In the case of the representations 3> ( X ^ ' *) of Sp(n), when
L'.L'+t!
and
we have
This reduces expressions (AH. 5) and (AIL 6) to the following form:
4
(An.
(AH. 12)
-35 -
The action of the invariant operator I , connected with the Sp(l) component
of Sp(l) Ef Sp(n) is given by the formula
z: y
where
(All. 13)
c r { A I L
-36-
APPENDIX in
Calculation of the highest weight.
In a representation X> of a semi-simple Lie group <3r of rank n.,
in which the representives I>(H;Jof the Cartan subgroup generators \ are
diagonal, the eigenstates and eigevalues of lifW;) are defined by
D0O4 «>*,•*,
The set of eigenvalues lv*\\ may be considered as a n-dimensional vector
in the so-called weight space [13] . A weight y£ is called higher than
another weight 'to ' if the first non-vanishing component 'ht - yv\ ' is a
positive number. The weight A which is higher than any other weight in
a given representation is called the highest weight. As is well known [14],
the highest weight fully characterizes the irreducible representation of a
compact group. We shaU denote its components by A .
It has been proved by Cartan that there exist v\. fundamental weights
A , . . . ; A and that any highest weight A is a linear combination
1 Acu) . 2)
where )• ^ are non-negative integers. Moreover, these fundamental
weights are uniquely defined by the root system of the algebra of
G; from the system ; io</ , . . . ,o( ] of simple roots,we can calculate
the Cartan matrix / A j : i by
A,. - ^ ^ ' * (Am.3)
and then the fundamental weights a r e given by
Uw (Am.4)
where A* i s the inve r se m a t r i x to j /^j ;] . Usually, an i r reduc ib le r e p r e s e n t -
ation of the group Q- i s denoted by ^(X^t •• >^^) . This notation
appea r s in column 2 of tab les II and III and throughout Sections IQ and IV.
-37-
Let us calculate the coefficients A L , ... , 1 ^ for the representations
of the group Sp(n). The eigenvalue of the generator
is clearly
* ^v.
/**tt
To this value corresponds
~ T
Hence, there is a highest possible value of X^.t given by
L- T
(Ain. 5)
(Ain. 6)
So the highest weight will be that with the highest value of J?K . From
Fig, 4 we see that the highest possible value of $^ is
(Am. 7)
(Am. 8)
(Am. 9)
But when (ADI. 7) and (Ain. 9) hold simultaneously, it is easy to see that
"2
So we have
L..., / \ ^ ) « (of ..v
Now in the case of group Sp(n) the Cartan matrix is given by
•2 ~4 - - • O O O
0
O
0oo O -2 X
(Ain.
(Ain.
(Am. 12)
-38-
if we use the definition of o(Jl) ( { a l,..., h) given in Appendix I (A1.16).
Then the set of fundamental weights of Sp(n) is
" * *" " ' • * (AIII.13)7\ C M " l
- ) •f" x °*- • &i -f •
This gives in the case of the representation Dh of Sp(n)
(AIH.14)
Therefore, we have obtained the resulting formula
(AIII.15)
as indicated in Sections IH and TV.
-39-
APPENDIX IV
Lowest representations of Sp(2) and Sp(3)
In this Appendix we consider the lowest representations of Sp(2)
and Sp(3) which have been obtained in this paper.
In Tables I and II we give
1° The comparison of our notation 3> of the representation of Sp(n)
and Sp(3); respectively with the notation based on the highest weight com-
ponents and the Young tableaux, as defined in [ 11] and [12] .
2° The decomposition in respect to the maximal compact subgroup.
This is given by diagrams in the last columns of both tables where at each
point in brackets stand the dimensionalities of the corresponding represent?
ations of the subgroup Sp(l) S Sp(2) and Sp(l) <g>Sp(l), respectively.
In Tables III and IV we give the list of the basis functions for the simplest
less degenerate representations of Sp{2) and Sp(3), respectively. For the
Sp(2) group we use the denotation Y \ x[ i for the original functions
of the set (2. 28) and ^- ] »,+ ^+ for the new orthonormal basis functions.
( ( eIn the case of Sp(3) this denotation is / H f f _ and
' 2 Y respectively.
We have selected from the class cf equivalent representations those
which have H - + L . In the last column of these tables is the number
of functions of the corresponding type, which differ by the eigenvalues M
(s = 1, 2 resp. 3).
-40-
Table II
Lowest representations of Sp(2)
D t
Z.D o
<
<
<
JL
D(XX.A2)
(1,0)
(2,0)
(0,1)
(3,0)
(1,1)
(4,0)
(2,1)
(0,2)
[10]
[20]
[11]
[30]
[21]
[40]
[31]
[22]
Dimension
4
10
5
^0
11
»
35*
14
Young tableaux
J
Diagram (^ 4 )
1!
(
>v fa i i
X / , '" fl
Aa
- 4 1 -
Table in
Lowest representations of Sp(3)
Dimension Young tableaux Diagram (*-j
(1,0,0) 1100] 6
(2,0,0) [200] 21ifi.J*)
D (0,1,0) [110] 1±
(3,0,0) [300] 56N^rSl
(1,1,0) [210]
n.<£,£)
(4,0,0) [400] 126
(2,1,0) [310] 189
(0, 2, 0) [220] 22J*.
Ci.D
-42-
Table IV
Basis functions for the representation spaces of the lowest less degenerate
representations of Sp(2)
Represent
L
D 3
1
List of basis functions
0 l / l j 0,0 yi- 0,0
0 ' O f o ~ I 0, Oj 0,0
f yi;°ii v i ' °« 1
1 y V'1 t y i i0'4
i Y^; - Y^O
No. of
tions
1
4
2
2
6
6
3
4
3
8
9
8
- 4 3 -
Table IV (Continued)
List of basis functionsNo. offunc-
tions
o yh,o,oyh,' o , 0
O,O; o,O
0
0(YH ; ; , t , , - Y,t;
Table V
Basis functions for the representation spaces of the lowest less degenerate
representations of Sp(3)
Represent.
0
0 1 C,
0 ' fij
' \7^0 ' 0,
o 1
o,a
List of basis functions
\ c,o,oto,ofo
f r vi ' V ' HjT, 3, h/j -1 o, i 1 Hi^o, Hf. A, o, -i j
A I fyz,'lioi/i'i y1^)0,^" )
No. of
func-tions
i-i
4
4
1
-44-
Table V (Continued)
List of basis functions No. of
functions
,1,0,0
1 O,o,Hi* ' 0,0,^0,0,4
^ . ^ 0,-1,0
- f . 1 V;2,<U
D
>,0
8
-f,
W^v, - Wi 0,0,4
M
-45-
FOOTNOTES
1) Note that sometimes in the set of invariant operators there appear
operators which are not elements of the enveloping algebra, (For
an explicit example see [4] ).
2) The rank of a space X - &{ % i s defined as the number of in-
variants of any two points x,o- £ X with respect to the action of
the fundamental group G" on A .
3) Here [a] means the integral part of a defined as usual.
4) We are limiting ourselves only to the case H~=+L because the
choice M *= — L leads only to the change of sign in the r . h. s.
of equation (3.16) which means changing the sign at any i/^ in lower
indices of functions (3.18). The corresponding representations
a re equivalent.
-46-
REFERENCES
[11 H. J. LIPKIN, Lie group for pedestrians (Amsterdam, 1965).
[2] P. BUDINI, private communication,
[3) H. BACRY, J. NUYTS and L. VAN HOVE, Nuovo Cimento 35,
510 (1965)j
R. HWA and J. NUYTS, Group embedding for the harmonic
oscillator. Institute for Advanced Study, Princeton, NJ, (1965)
[4J ' N. LIMIC, J. NIEDERLE and R. RACZKA, Discrete degenerate
representations of non-compact rotation groups, ICTP, Trieste,
preprint lC/66/2.
J. FISCHER and R, RACZKA, Discrete degenerate representations
of non-compact unitary groups, ICTP, Trieste, preprint IC/66/16.
[5] I.M. GEL'FAND, Am. Math. Soc. Transl. Ser. 2, 37, 31 (1964).
[6] S. HELGASON, Differential geometry and symmetric spaces.
Academic Press, N.Y. (1962).
[7] B.A. ROZENFELD, Dokl. Akad. Nauk SSSR 110, 23 (1956).
[8] C. CHEVALLEY, Theory of Lie groups - I , Princeton University
Press (1964).
[9] WANG HSIEN-CHUNG, Ann. Math. 55, 177 (1952).
[10J M. E. ROSE, Elementary theory of angular momentum, John Wiley
& Sons, N.Y. (1961).
[11] M. KONUMA, K. SHIMA and M. WADA, Suppl. to Progr. Theoret.
. (Kyoto) No. 28 (1963).
[12] M. HAMMERMESH* Group theory and its applications,; Addison-
Wesley Inc., London (1962).
-47-
[131 R.E. BEHRENDS, J. DREITLEWt C. FRONSDAL and B.W.LEE,
Rev. Mod. Phys.34, 1, (1962).
[14] E.B. DYNKIN. Usp. Math. Nauk 2, 4(20) 59 (1947).
-48-
Available from the Office of the Scientific Information and Documentation Officer,
International Centre for Theoretical Physics, Piazza Oberdan 6, TRIESTE, Italy
5394