1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \chapter{Highest weight representations}\label{c.highest}
3:
4: This chapter discusses highest weight representations, providing tools
5: for classifying many irreducible representations of interest.
6: We extend the ladder technique used in Section
7: \ref{section-reps-heis} for determining the unitary representations
8: of the oscillator algebra to some other small Lie algebras of interest,
9: and indicate how the ideas generalize further.
10:
11: The basic ingredient is a triangular decomposition, which exists for all
12: finite-dimensional semisimple Lie algebras, but also in other cases
13: of interest such as the oscillator algebra, the Heisenberg algebra
14: with the harmonic oscillator Hamiltonian adjoined.
15:
16: We look at detail at 4-dimensional Lie algebras with a nontrivial
17: triangular decomposition (among them the oscillator algebra
18: and $so(3)$), which behave almost like the oscillator algebra.
19: As a result, the analysis leading to Fock spaces generalizes without
20: problems, and we are able to classify all irreducible unitary
21: representations of the rotation group. Various related material
22: concerning $SO(3)$ and its universal covering group $SU(2)$ is also
23: included.
24:
25:
26: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
27: \section{Triangular decompositions}
28:
29: Let $\Lz$ be a Lie $*$-algebra. A {\bfi{triangular decomposition}}
30: \index{decomposition!triangular}
31: of $\Lz$ consists of Lie subalgebras $\Lz_-$, $\Lz_0$ and $\Lz_+$
32: of $\Lz$ satisfying the properties\footnote{
33: Note that the present concept of a triangular decomposition
34: is less demanding and hence more general than in the treatment
35: by \sca{Moody \& Pianzola } \cite{MooP}. Their additional restrictions
36: allow them to extend much of the finite-dimensional semisimple theory
37: outlined below to the infinite-dimensional case.
38: } % end footnote
39:
40: (T1)~~~$\Lz ~=~ \Lz_- \oplus \Lz_0 \oplus \Lz_+$,
41:
42: (T2)~~~$\Lz_0 \lp \Lz_\pm ~\subseteq~ \Lz_\pm$,
43:
44: (T3)~~~$\Lz_0^* = \Lz_0$,~~$\Lz_\pm^* = \Lz_\mp$,
45:
46: (T4)~~~$\Lz_0$ is abelian and contains the center $Z(\Lz)$.
47:
48: Triangular decompositions generalize the properties of annihilation
49: and creation operators in the oscillator algebra $os(1)$ to more
50: general Lie algebras.
51: The terminology derives from the following motivating examples.
52:
53: \begin{expls}
54: (i)
55: In the Lie algebra $\Lz = gl(n,\Cz) = \Cz^{n\times n}$, we can define
56: a triangular decomposition by defining $\Lz_0$ to be the Lie subalgebra
57: of diagonal matrices, $\Lz_+$ to be the Lie subalgebra of strictly
58: upper triangular matrices, and $\Lz_-$ to be the Lie subalgebra of
59: strictly lower triangular matrices. Verification of the axioms is
60: straightforward.
61:
62: (ii) The oscillator algebra $os(1)$ has a triangular decomposition,
63: given by
64: \[
65: \Lz_- = \Cz\, a\,, ~~~ \Lz_0 =\Cz\, 1 +\Cz\, n\,, ~~~
66: \Lz_+= \Cz\, a^*\,.
67: \]
68: \end{expls}
69:
70: A {\bfi{triangulated Lie algebra}}\index{Lie algebra!triangulated}
71: is a Lie $*$-algebra with a distinguished triangular decomposition.
72: We call the number $\rk\Lz:=\dim\Lz_0/Z(\Lz)$ the {\bfi{rank}},
73: and $\deg\Lz:=\dim\Lz_\pm$ the {\bfi{degree}} of the triangulated
74: Lie algebra $\Lz$. The elements of the dual space\footnote{Since in
75: the context of $*$-Lie algebras, the notation $V^*$ for the dual of
76: $V$ is ambiguous, we use in this section a prime to indicate the
77: dual.}
78: $\Lz_0'$ are called {\bfi{weights}}.
79: A {\bfi{highest weight representation}}
80: \index{representation!highest weight} is a representation $J$ of $\Lz$
81: on a vector space $\Vz$ with a distinguished element $1$, called the
82: {\bfi{ground state}}\footnote{In a quantum field theory context,
83: the ground state is referred to as the {\bfi{vacuum}}.}, such that
84:
85: (HW1)~~~$J(\alpha) 1 = 0$ for all $\alpha\in\Lz_-$, and
86:
87: (HW2)~~~$J(\alpha) 1 \in \Cz$ for all $\alpha\in\Lz_0$.
88:
89: The elements of $\Lz_-$ thus behave like annihilation operators.
90: The defining properties imply that
91: \[
92: w(\alpha):=J(\alpha)1\,, \for \alpha\in\Lz_0\,,
93: \]
94: defines a weight $w\in\Lz_0'$, called the {\bfi{highest weight}}
95: of the representation. A highest weight representation is irreducible
96: if and only if the elements
97: $a_1^*\dots a_k^*1$ with $a_1,\dots, a_k\in\Lz_-$ span a dense
98: subspace of $\Vz$. In an irreducible highest weight representation
99: with highest weight $w$, all Casimir elements $C$ of $\Lz$
100: have a fixed value $C(w)\in\Cz$.
101:
102: The {\bfi{spectrum}} of $\Lz$ is the set $\Sigma(\Lz)$ of weights
103: $w$ for which a unitary group \at{which group?} representation exists,
104: whose associated infinitesimal
105: representation is a highest weight representation of $\Lz$ with highest
106: weight $w$. The spectrum of $\Lz$
107: determines the possible spectra of each Casimir element $C$ in
108: arbitrary unitary representations of the universal covering group of
109: $\Lz$, since the possible eigenvalues are precisely the possible $C(w)$
110: where $w$ ranges over the spectrum of $\Lz$.
111:
112: Note that a weight $w$ belongs to the spectrum of $\Lz$ iff there is
113: a unitary (cf. Definition \ref{d.lie-rep}) highest weight
114: representation of $\Lz$ with highest weight $w$. In
115: this case, there is a Euclidean inner product on $\Vz$, and without
116: loss of generality, the ground state 1 may be assumed to be normalized.
117:
118: \bigskip
119: %%%%%%%%%%%%%%%%%%%%%%%%%%%
120: \bfi{The semisimple case.}
121: There are many examples of triangulated Lie algebras, related to
122: finite-dimensional \idx{semisimple} Lie algebras (see the outline below)
123: and to important classes of infinite-dimensional Lie algebras.
124:
125: We mention without proof (which can be found in many places, e.g.,
126: \sca{Fuchs \& Schweigert} \cite{fuchsandschweigert},
127: \sca{Fulton \& Harris} \cite{fultonharris}, \sca{Humphreys}
128: \cite{humphreys}, \sca{Jacobsen}
129: \cite{jacobsen}, \sca{Knapp} \cite{knapp}, \sca{Kirillov}
130: \cite{kirillovLiegroup}) a number of facts about
131: finite-dimensional semisimple Lie algebras.
132:
133: All finite-dimensional semisimple real Lie algebras have a triangular
134: decomposition, which is unique up to automorphisms. In this case,
135: $\Lz_0$ is a Cartan subalgebra (a maximal abelian subalgebra generated
136: by diagonal matrices in the adjoint representation, for some choice of
137: basis), which is unique up to conjugation, and the Lie algebra $\Lz$
138: decomposes as
139: \[
140: \Lz = \Lz_0 \oplus \bigoplus_{\alpha\in \Delta} \Lz_{\alpha}\,,
141: \]
142: where $\Delta\subset \Lz_{0}'$ is the set of {\bfi{roots}}.
143: The roots are nonzero elements of the dual of the Cartan subalgebra
144: such that the Cartan subalgebra acts diagonally on $\Lz_\alpha$:
145: \[
146: h \lp x = \alpha(h)x\,,~~~h\in
147: \Lz,~~x\in \Lz_\alpha\,.
148: \]
149: $\Lz_\alpha$ is always 1-dimensional; any nonzero element in
150: $\Lz_\alpha$ is called a {\bfi{root generator}}. For each root
151: $\alpha\in \Delta$ the negative $-\alpha$ is also a root:
152: for all $\alpha\in\Lz^*$, if $\Lz_\alpha \neq 0$, then
153: $\Lz_{-\alpha}\neq 0$. Therefore there exists a choice of ordering
154: such that $\Delta$ can be written as the union of the set of positive
155: roots $\Delta^+$ and the set of negative roots $\Delta^-$ and
156: $\Delta^-=-\Delta^+$, in such a way that the cone of nonnegative
157: linear combinations from $\Delta^+$ and $\Delta^-$ intersect in 0 only.
158: One defines
159: \[
160: \Lz_\pm = \bigoplus_{\alpha\in \Delta^{\pm}}\Lz_\alpha\,,
161: \]
162: and finds (using further properties of the roots) that the semisimple
163: Lie algebra is a triangulated Lie algebra.
164:
165: \begin{expl}
166: Take $\Lz = sl(n,\Cz)$, the Lie algebra of $n\times n$
167: matrices with trace zero. Let us we write $E_{ij}$ for the matrix that
168: is $1$ on the $(i,j)$-entry and zero everywhere else. Then the diagonal
169: matrices that have trace zero make up the Cartan subalgebra, which is
170: thus spanned by the matrices $E_{ii}-E_{i+1,i+1}$ for $1\leq i\leq
171: n-1$ so that the rank is $n-1$. We have for $h={\rm
172: diag}(h_1,\ldots,h_n)\in \Lz_0$ in the Cartan subalgebra and for
173: $E_{ij}$ with $i\neq j$
174: \[
175: h \lp E_{ij} = (h_i - h_j)E_{ij}\,.
176: \]
177: Hence the roots are of the form $\lambda_{i}-\lambda_{j}$ where
178: $\lambda_i$ reads off the $i$th diagonal entry of an element of the
179: Cartan subalgebra. We can choose a root $\lambda_i-\lambda_j$ to be
180: positive if $i< j$. Then $\Lz_+$ are the upper triangular matrices,
181: and $\Lz_-$ the lower triangular matrices. The positive root
182: generators are $E_{ij}$ with $i<j$.
183: \end{expl}
184:
185: Associated with each semisimple Lie algebra is a
186: {\bfi{weight lattice}}, which is a discrete additive subgroup of
187: $\Lz_0'$ and whose elements are called {\bfi{integral weights}}.
188: Additionally, there is a distinguished subset of the weight lattice,
189: which is closed under
190: addition and whose elements are called \bfi{dominant integral weights}.
191: In terms of these:
192:
193: (i) For each weight $w$, there is a Lie representation with $w$ as
194: highest weight.
195:
196: (ii) A highest weight representation is finite-dimensional if and only
197: if the highest weight is dominant and integral.
198:
199: (iii) For compact finite-dimensional Lie algebras, that is
200: finite-dimensional Lie algebras with a negative definite
201: Cartan--Killing form (these are automatically semisimple, see
202: Lemma \ref{lem.cartan}), a highest weight representation is
203: unitary if and only if it is finite-dimensional. The inner
204: product is then uniquely determined
205: by the requirement that the ground state 1 is normalized.
206:
207: (iv) The Lie algebra induces a unitary representation of the
208: universal covering group $\Gz$ if and only if $w$ is a dominant
209: integral weight. Thus the spectrum of $\Lz$ consists of all
210: dominant integral weights of $\Lz$.
211:
212: In the context of an integrable classical theory associated with $\Lz$,
213: (iv) is equivalent to the \bfi{Bohr--Sommerfeld quantization
214: condition}. (This folklore result is never stated in a precise form,
215: but see, e.g., \sca{Voros} \cite{Vor}, \sca{Kochetov} \cite{Koc}, and
216: \sca{Gadiyar} \cite{Gad}.)
217:
218:
219:
220: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
221: \section{Triangulated Lie algebras of rank and degree one}
222: \label{s.triang.rd1}
223:
224: We have seen that the oscillator algebra $os(1)$ has a triangular
225: decomposition of rank and degree 1.
226: A general triangulated $*$-Lie algebra of rank and degree 1 with
227: center $\Cz$ must be the direct sum of the algebras
228: \[
229: \Lz_-= \Cz a\,, ~~~ \Lz_+=\Cz a^*\,,~~~ \Lz_0=\Cz+\Cz h\,,
230: \]
231: where $h$ is a fixed element in $\Lz_0\setminus\Cz$.
232: The center $\Cz$ commutes with everything, but $h$ in
233: general does not, which is the case we consider here. Then
234: we may rescale $h$ to obtain
235: \[
236: a\lp h=ia\,.
237: \]
238: The operation $*$ then gives
239: \[
240: a^*\lp h=-ia^*\,.
241: \]
242: For the Lie product of $a$ and $a^*$ we introduce complex numbers $u$
243: and $v$ and write
244: \[
245: a\lp a^* = i(uh +v)\,,
246: \]
247: but noting that $(a\lp a^*)^*=-a\lp a^*$ we see that $u,v\in \Rz$.
248: It is easy to check that for all $u,v\in\Rz$ the Jacobi identities
249: are fulfilled and hence for all real numbers $u,v$ we have a
250: Lie $*$-algebra.
251:
252: For the two-parameter family of Lie $*$-algebras just defined there are
253: essentially four different cases;
254: \begin{enumerate}
255: \item
256: $u=v=0$.
257: This is the $*$-Lie algebra $iso(2)\oplus \Cz$.
258:
259: \item
260: $u=0$, $v=\pm1$.
261: If $u=0$ we can rescale the $a$ and $a^*$ as $a\mapsto \lambda a$ and
262: $a^*\mapsto \lambda^*a^*$ to get $v=\pm 1$. By complex conjugation of
263: the algebra we then can choose the sign of $v$ and we find the $*$-Lie algebra $os(1)$. For the oscillator algebra we have $h\sim a^*a$.
264:
265: \item
266: $u=1$ and $v=0$.
267: This $*$-Lie algebra is $so(2,1)\oplus \Cz$.
268: If $u$ and $v$ are both nonzero, we can redefine $h$ as $h \to
269: \alpha h+\beta$ for some $\alpha,\beta\in \Cz$ to obtain this case
270: or the next one.
271:
272: \item
273: $u=-1$ and $v=0$.
274: This is the $*$-Lie algebra $so(3)\oplus \Cz$.
275: \end{enumerate}
276:
277:
278: Note that the elements $a$ and $a^*$ are
279: abstract
280: vectors from the point of view of Lie algebras. That means that we
281: cannot say that $a^*$ is the conjugate of $a$; it is only in
282: Lie $*$-algebras, in the $*$-Poisson algebras and in their unitary
283: representations that we can say that $a^*$ is the Hermitian conjugate
284: of $a$. It is for these reasons that we have treated case 3 and case 4
285: separately. In a unitary representation we have $J(f\lp
286: g)=\frac{i}{\hbar }(J(f)J(g)-J(g)J(f))$, so that $J(f)^*=J(f^*)$ makes
287: sense.
288:
289: As alluded before $so(2,1)$ and $so(3)$ are isomorphic as complex Lie
290: algebras. If we define in $so(2,1)$ the elements $r=ia$, $s=ia^*$ we
291: obtain the relations
292: \[
293: h\lp r = -ir\,,~~~ h\lp s = is\,,~~~ r\lp s=-a\,,
294: \]
295: which defines case 4 of the list above:
296: $so(3)$. However, the map from $so(2,1)$ to $so(3)$ does not preserve
297: the $*$-operation, since $r^* = (ia)^* = -ia^* \neq s$. That means
298: that $so(2,1)$ and $so(3)$ are not isomorphic as Lie
299: $*$-algebras.
300:
301: Among the triangulated Lie algebras of rank and degree 1 listed above,
302: the most interesting cases for both classical and quantum mechanics are
303: $os(1)$ and $so(3)\oplus \Cz$.
304: As we have seen, the oscillator algebra $os(1)$ is related to the
305: harmonic oscillator. The algebra $so(3)\oplus \Cz$ involves
306: infinitesimal ordinary rotations and arises when dealing
307: with the spinning top, as explained in Chapter \ref{c.harmonic}.
308: The algebra $so(2,1)\oplus \Cz$ is less prominent in classical mechanics
309: although it arises in the analysis of the celestial 2-body
310: problem. The algebra $so(2,1)\oplus \Cz$ has important applications
311: to exactly solvable problems in quantum mechanics, and even appears
312: in so-called gauged supergravity theories.
313:
314:
315:
316: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
317: \section{Unitary representations of $SU(2)$ and $SO(3)$}\label{s.su2rep}
318:
319: We now discuss the unitary representations of the Lie groups $SU(2)$
320: and $SO(3)$.
321: The method presented below is often encountered in quantum physics
322: textbooks. In Section \ref{s.highest} we discuss the highest weight
323: representations of triangulated Lie algebras of rank and degree 1,
324: which shows a great similarity with the discussion here.
325:
326: Since the group $SU(2)$ is compact it has an invariant {\bfi{Haar
327: measure}} $d\mu(g)$. \at{details or references?}
328: Therefore we can integrate over the group in an
329: invariant way; invariance of the Haar measure means
330: $\sint f(hg)d\mu(g)=\sint f(g)d\mu(g)$ . If $SU(2)$ acts on a vector
331: space $V$ with an inner product $\<,\>_0$ we can integrate
332: over the group to get an invariant inner product;
333: \[
334: \< v,w\rangle = \int_{SU(2)} \< g\cdot v,g \cdot w\rangle_0 d\mu(g)\,,
335: \]
336: where we denoted the action of $g\in SU(2)$ on $v\in V$ by $g\cdot v$.
337: It is a direct consequence of the invariance of the Haar measure that
338: the inner product $\<,\rangle $ is $SU(2)$-invariant. Hence we have
339: realized $SU(2)$ by unitary matrices; every representation of $SU(2)$
340: is equivalent to a unitary representation.
341:
342: Since the group $SU(2)$ is compact and simply connected
343: \at{define this in the Lie group chapter!}
344: there is a one-to-one correspondence between the representations
345: of the group and the representations of the Lie algebra
346: $su(2)\cong so(3)$.
347: The Lie algebra consists of antihermitian matrices but multiplying
348: them by $i$ we obtain Hermitian matrices and we may
349: use the Pauli matrices to describe $su(2)$. Finding all representations
350: of the Lie algebra $su(2)$ therefore gives all representations
351: of $SU(2)$.
352:
353: We put $t_i = \shalf \sigma_i$ for $i=1,2,3$ and define
354: $L_\pm = (t_1\pm i t_2)$ and obtain a triangulated algebra with
355: trivial center;
356: \[
357: t_3 \lp L_\pm =\pm L_\pm \,, ~~~ L_+\lp L_- =2t_3 \,.
358: \]
359: In a unitary representation we require that $t_3$ is Hermitian and
360: $L_{\pm}^{*}=L_\mp$. If $v$ is an eigenvector of $t_3$ with eigenvalue
361: $\alpha$, then $L_-v$ is an eigenvector of $t_3$ with eigenvalue
362: $\alpha-1$. For a finite-dimensional representation we cannot lower
363: the eigenvalue forever and hence there exists
364: a vector $v$ with $L_-v=0$.
365:
366: Assume that we have $t_3 v=\alpha v$
367: for some complex number $\alpha$. Acting on $v$ with $L_+$ we get
368: vectors with eigenvalues $\alpha,\alpha+1,\alpha+2,\ldots$. Again this
369: series has to terminate. Thus, there is an eigenvector $w$ with
370: eigenvalue $\alpha+N$ that is annihilated by $L_+$. Since $t_3$
371: is Hermitian, vectors with different eigenvalues are orthogonal and
372: hence linearly independent. Thus the $N+1$ vectors with eigenvalues
373: $\alpha,\ldots,\alpha+N$ form an irreducible representation. The trace
374: of $t_3$ is zero, since $\tr t_3=-i\tr (t_1t_2-t_2t_1)$. But
375: then the sum of the eigenvalues should vanish:
376: \[
377: 0=\sum_{n=0}^{N}\alpha+n=(N+1)\alpha +\frac{1}{2} N(N+1)
378: = \frac{1}{2}(N+1)(2\alpha + N)\,.
379: \]
380: It follows that $\alpha=-N/2$. Therefore the eigenvalues are the
381: integers
382: $-\frac{N}{2},-\frac{N}{2}+1,\ldots,\frac{N}{2}-1,\frac{N}{2}$.
383: Conversely, for all integers $N$ we find a
384: representation by giving vectors $e_\alpha$ with $-N/2\leq \alpha \leq
385: N/2$ and defining the action of $t_3$ and $L_\pm$ by the above
386: rules. We then recover $t_1$ and $t_2$ by
387: $t_1=\frac{1}{2}(L_++L_-)$ and
388: $t_2=\frac{1}{2i}(L_+-L_-)$. We can thus label the
389: finite-dimensional
390: representations of $su(2)$ by half-integers $j =0,1/2,1,3/2,2,\ldots$.
391: We denote them by $D_j$; note that we have $j=N/2$. The
392: dimension of the representation $D_j$ is $2j+1$ and the eigenvalues of
393: $t_3$ are
394: $-j,-j+1,\ldots,j-1,j$. The
395: Casimir $J^2$ defined by
396: \[
397: J^2 = t_1t_1+ t_2t_2 + t_3t_3 = L_+L_-
398: - t_3 +t_3 t_3
399: \]
400: has the value $j(j+1)$ on the representation $D_j$ since acting on the
401: state $v$ with eigenvalue $-j$
402: \[
403: J^2 v = (L_+L_-
404: - t_3 +t_3t_3)v= (0+j+ j^2)v= j(j+1)v\,.
405: \]
406: The number $j$ is called the \idx{spin} of the representation. Clearly
407: $D_j$ is irreducible.
408:
409:
410: The representations that correspond to nonintegral $j$ cannot be
411: lifted to
412: representations of $SO(3)$. Although the Lie algebras $su(2)$ and
413: $so(3)$ are isomorphic, the groups $SO(3)$ and $SU(2)$ are not!
414: As mentioned before, $SU(2)$ is the universal covering group of
415: $SO(3)$. In fact, we have $SO(3)=SU(2)/\Zz_2$. That means that there is
416: an action of $\Zz_2$ on $SU(2)$, such that $SO(3)$ is the manifold
417: $SU(2)$ with the points that are related by the $\Zz_2$-action
418: identified. In Section \ref{s.rot} we gave details on how
419: $SO(3)$ and $SU(2)$ are related by a 2-1 map $SU(2)\to SO(3)$.
420: \at{rephrase!}
421: If a representation of $SU(2)$ is such that $\Zz_2$-related
422: points have the same image under the representation we have a
423: well-defined representation for $SO(3)$; this thus precisely
424: corresponds to the $\Zz_2$-invariant representations. It turns out
425: that only the representations with integer $l$ correspond to
426: $\Zz_2$-invariant representations. In physics, particles are
427: represented by fields that take values in an
428: $su(2)$-representation. The representations $D_j$ for $j=1/2,
429: 3/2,5/2,\ldots$ correspond to \idx{fermions} and for $j=0,1,2,\ldots$
430: to \idx{bosons}.
431:
432:
433:
434: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
435: \section{Some unitary highest weight representations}\label{s.highest}
436:
437: For the quantum theory one considers the unitary highest weight
438: representations. We investigate the unitary highest weight
439: representations for the triangulated Lie algebras of rank and degree 1
440: listed in Section \ref{s.triang.rd1}. We thus look for a realization
441: of operators $a$, $a^*$, $h$ and $1$ such that $1$ acts as the
442: identity, $h$ acts diagonally and is Hermitian, $a^*$ is the
443: adjoint of $a$ and the
444: following relations hold (see \gzit{e.liequant} and Definition
445: \ref{d.lie-rep}):
446: \[
447: [a,h]=\hbar a\,,~~~[a^*,h]=-\hbar a^*\,, ~~~ [a,a^*]=\hbar (uh+v)\,.
448: \]
449: Furthermore, we assume there is a vector $|0\>$ with
450: \[
451: a|0\>=0\,,~~~ h|0\> = \alpha |0\>\,.
452: \]
453: By acting with $a^*$ on $|0\>$ we obtain the other vectors in the
454: representation. We define
455: \[
456: | k \> = \frac{(\hbar a^*)^k}{k!}|0\>\,,
457: \]
458: so that
459: \[
460: a^* | k -1\> = \hbar k | k\>\,.
461: \]
462: It follows that
463: \[
464: h |k\> = \hbar (k + \alpha) |k\>\,.
465: \]
466: We have $a|k\> = c_k |k-1\>$ and we want to determine $c_k$. Since
467: $aa^*=[a,a^*]+a^*a$ we find
468: \[
469: \hbar kc_k|k-1\>= aa^* |k-1\>
470: =\hbar \Bigl( (uh+v) + (k-1)c_{k-1}\Bigr)|k-1\>\,,
471: \]
472: from which it follows
473: \[
474: kc_k - (k-1)c_{k-1} = \hbar (uk + u\alpha +v)\,,
475: \]
476: which is solved by
477: \beqar
478: kc_k &=& 0+(1c_1 - 0c_0)+(2c_2-1c_1)+ \ldots + (kc_{k} -
479: (k-1)c_{k-1})\nn\\
480: &=&\hbar k \Bigl(\hbar \alpha u + v +\shalf \hbar u(k+1)\Bigr)\,,\nn
481: \eeqar
482: so that
483: \[
484: a|k\> = \Bigl( \alpha \hbar u + v +\shalf \hbar u(k+1)\Bigr)|k-1\>\,.
485: \]
486: The vectors $| k \>$ are orthogonal, as in
487: the case of the harmonic oscillator. So we suppose
488: $\< j | k\> = N_k\delta_{jk}$, and calculate $\< j | a^* | k\>$
489: in two ways:
490: \[
491: \< j | a^* | k\> = (k+1) \< j|k+1\>,
492: \]
493: \[
494: \< j | a^* | k\>
495: = (v + \bar \alpha \hbar u + \shalf \hbar u(k+1)) \<j-1| k\>.
496: \]
497: Choosing $k=j-1$ we find
498: \[
499: j\hbar N_j =
500: \left(v+u\hbar \bar \alpha +\shalf(j+1)u\hbar \right) N_{j-1}\,.
501: \]
502: For a representation we require that $N_j\geq 0$ for all $j$.
503: We may normalize
504: $N_0=1$ and it follows that we must have $\alpha\in\Rz$. To have a
505: faithful representation we need $|1\>\neq 0$ and thus $N_1>0$.
506: We distinguish further two cases:
507:
508: {\sc Case 1:} $u\geq 0$. By assumption
509: $\psi_0$ is a nonzero vector and thus has a positive norm. But then
510: all $N_j$ are positive. Hence we find nonzero vectors for all $j\in
511: \Nz_0$. An example of this case is given by $so(2,1)$, which is a
512: noncompact Lie algebra. More generally, noncompact Lie algebras
513: (defined by having a
514: Cartan--Killing form that is not negative definite) do not admit a
515: finite-dimensional unitary representation.
516:
517: {\sc Case 2:} $u<0$. In this case $N_j$ can
518: become negative, unless it becomes zero for some integer
519: $j_m$. Thus $ (j_m+1) + 2(\bar \alpha + \frac{v}{\hbar u})$. In
520: this case we thus have a finite-dimensional unitary
521: representation for every integer $j_m = 0,1,2,\ldots$. The
522: dimension of the representation is $j_m+1$. If $j_m=0$ the
523: vector $|1\>$ is already zero, and hence $a^*$ operates as $0$
524: in this representation. Therefore, if $j_m\geq 1$ the
525: representations are faithful.
526:
527: For the triangulated Lie algebras of rank and degree 1, there is
528: a Casimir operator of the form $C=\hbar a^*a - q(h)$ for some
529: quadratic $q(h)$. From Section \ref{s.su2rep} we know that $so(3)$ has
530: the Casimir $J^2=2aa^*+h^2+ih$. And for the algebra $os(1)$ the
531: element $C=\hbar a^*a-h$
532: is a Casimir. For the harmonic oscillator we then have
533: $C=0$, since $h$ is precisely $\hbar a^*a$. For
534: $so(2,1)$ this does not work; there is no analogue of the
535: number operator with only integer eigenvalues.
536: That is, the Lie algebra $so(2,1)$ does not admit a discrete Casimir.
537: \at{check!}
538:
539: