1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \chapter{Harmonic oscillators and coherent states}\label{c.harmonic}
3:
4: Part IV applies the concepts introduced so far
5: to the study of the dominant kinds
6: of elementary motion in a bound system, vibrations (described by
7: oscillators, Poisson representations of the Heisenberg group),
8: rotations (described by a spinning top, Poisson representations of
9: the rotation group), and their interaction. On the quantum level,
10: quantum oscillators are always bosonic systems, while spinning systems
11: may be bosonic or fermionic depending on whether or not the spin is
12: integral. The analysis of experimental spectra, concentrating on
13: the mathematical contents of the subject, concludes our discussion.
14:
15: \bigskip
16: This chapter is a detailed study of harmonic oscillators (bosons,
17: elementary vibrations), both from the classical and the quantum point
18: of view. We introduce raising and lowering operators in the symplectic
19: Poisson algebra, and show that
20: the classical case is the limit $\hbar\to 0$ of the quantum harmonic
21: oscillator.
22:
23: The representation theory of the single-mode Heisenberg
24: algebra is particularly simple since by the Stone--von Neumann theorem,
25: all unitary representations are equivalent. We find that the quantum
26: spectrum of a harmonic oscillator is discrete and consists of the
27: classical frequency (multiplied by $\hbar)$ and its nonnegative
28: integral multiples (overtones, excited states).
29:
30: To make work in the representation where the harmonic oscillator
31: Hamiltonian is diagonal, we introduce Dirac's bra-ket notation, and
32: deduce the basic properties of the bosonic Fock spaces, first for
33: a single harmonic oscillator and then for a system of finitely many
34: harmonic modes.
35:
36: We then introduce coherent states, an overcomplete basis representation
37: in which not only the Heisenberg algebra, but the action of the
38: Heisenberg group is explicitly visible.
39: Coherent states are quantum states that behave as classically as
40: possible, thereby making a bridge between the quantum system and
41: classical systems.
42: The coherent state
43: representation is particularly relevant for the study of quantum
44: optics, but we only indicate its connection to the modes of the
45: electromagnetic field.
46:
47:
48: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
49: \section{The classical harmonic oscillator}\label{s.charm}
50:
51: The classical one-dimensional harmonic oscillator without damping,
52: introduced in Section \ref{s.cao}, is defined by the Hamiltonian
53: \lbeq{ch5.ham}
54: H = \frac{p^2}{2m}+ V(q)\,,
55: \eeq
56: where $V(q)$ is quadratic and bounded from below, so that there are
57: constants $q_0$, $V_0$ and $k>0$ with
58: \[
59: V(q) = V_0 + \frac{k}{2}(q - q_0)^2\,.
60: \]
61: The number $k$ is called the {\bfi{stiffness}}; the greater
62: the constant $k$, the more difficult is it to move away from
63: equilibrium. The Hamilton equations are:
64: \[
65: \dot q = \frac{p}{m}\,, ~~~ \dot p = - V'(q) = - k (q- q_0)\,.
66: \]
67: A complex exponential ansatz shows that the solution of the Hamilton
68: equations is:
69: \[
70: q(t) = q_0 + 2 \re (e^{i\omega t}x)\,, ~~~
71: p(t) = \re (i\omega m e^{i\omega t}x) \,,
72: \]
73: where $x$ is a complex number depending on the initial
74: conditions, and
75: \[
76: \omega = \sqrt{\frac{k}{m}},
77: \]
78: is the {\bfi{frequency}} of the harmonic oscillator.
79: It is convenient to express the variables in terms of a so-called
80: complex {\bfi{normal mode}}, the function $a(t)$ defined by
81: \[
82: a(t) := \sqrt{\frac{k}{2\omega}}(q(t)-q_0)
83: + i\frac{p(t)}{\sqrt{2m\omega}}\,.
84: \]
85: One can recover $q$ and $p$ through
86: \lbeq{ch5.pq}%
87: q(t) = q_0 +
88: \frac{1}{2}\sqrt{\frac{2\omega}{k}}(a(t)+a^*(t))\,, ~~~ p(t) =
89: \frac{1}{2i}\sqrt{2m\omega}(a(t)-a^*(t))\,,%
90: \eeq
91: hence the description by a normal mode is equivalent to the
92: original description. Differentiating $a(t)$ and using $\dot q = p/m$,
93: we obtain
94: \[
95: \dot a(t) = \sqrt{\frac{k}{2\omega}}\frac{p(t)}{m}
96: - \frac{i}{\sqrt{2m\omega}}k(q(t)-q_0)
97: = -i\omega a(t)\,.
98: \]
99: We conclude that $a(t)$ has to obey
100: \[
101: a(t)= a(0) e^{i\omega t}\,.
102: \]
103: We calculate the Lie product of $a$ and $a^*$ and find
104: \beqar
105: a \lp a^* &=& \frac{\partial a }{\partial p}\frac{\partial a^* }{\partial q} - \frac{\partial a^* }{\partial p} \frac{\partial a }{\partial q}\nonumber\\
106: &=& 2i\sqrt{\frac{k}{2\omega}}\frac{1}{\sqrt{2m\omega}} \nonumber\\
107: &=& i\,,\nonumber
108: \eeqar
109: that is, we obtain the relation
110: \lbeq{ch5.ccr}
111: a\lp a^* = i\,.
112: \eeq
113: The relation \gzit{ch5.ccr} is called the
114: {\bfi{canonical commutation relation} (CCR)} for the harmonic
115: oscillator.
116: More generally, one finds for the Lie product of general functions
117: $f,g$ of $a$ and $a^*$ the formula
118: \lbeq{ch5.pois}
119: f\lp g = i \frac{\partial f}{\partial a}\frac{\partial g}{\partial a^*}
120: -i \frac{\partial f}{\partial a^*}\frac{\partial g}{\partial a}.
121: \eeq
122: This will be seen later as a special case of a general principle for
123: constructing so-called Lie--Poisson algebras from a Lie algebra.
124:
125:
126: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
127: \section{Quantizing the harmonic oscillator}\label{s.qharm}
128:
129: For a classical harmonic oscillator, the Lie product in the CCR
130: \gzit{ch5.ccr} is defined via the Poisson bracket.
131: To quantize the harmonic oscillator, all we do is replace the Lie
132: product in the CCR by its quantum analogue. Thus we postulate the
133: existence of an operator $a$ and its conjugate $a^*$ with the relation
134: \[
135: \frac{i}{\hbar} [a,a^*] = i\,,
136: \]
137: equivalently
138: \lbeq{ch5.qccr}
139: [a,a^*]=\hbar\,.
140: \eeq
141: Note that equation \gzit{ch5.qccr} has the right behavior under
142: $\hbar \to 0$, since in the limit that $\hbar$ goes to zero, we have
143: to end up in the classical regime, where the operators $a$ and $a^*$
144: become functions on phase space and hence commute.
145:
146: Equation \gzit{ch5.qccr} defines a $*$-algebra, i.e., an associative
147: algebra with unity and an involution $*$, generated by $a$ with the
148: relation $aa^* - a^*a=\hbar$.
149: Later we look for representations in a Hilbert space, where the
150: involution then corresponds to Hermitian conjugation. But already at
151: this level, we call expressions in $a$ and $a^*$ operators.
152:
153:
154: The quantum mechanical Hamiltonian for the harmonic oscillator
155: is the operator given by direct substitution of the $p$ and $q$
156: from \gzit{ch5.pq}:
157: \beqar
158: H &=& \omega \Bigl(\frac{a-a^*}{2i}\Bigr)^2
159: + \omega \Bigl(\frac{a+a^*}{2}\Bigr)^2+ V_0 \nonumber\\
160: &=& \shalf \omega (aa^* + a^*a) + V_0 \nonumber\\
161: &=& \omega a^* a + \shalf \omega \hbar + V_0 \,.
162: \eeqar
163: Since only differences in energy are important, one often chooses
164: $V_0= -\shalf \hbar \omega$ to get the simple formula $H=\omega a^* a$.
165:
166: In the classical theory we have commuting variables $a$ and $a^*$ with
167: a Lie product $a\lp a^* = i$. That is, we have a commutative $*$-Poisson
168: algebra. In the quantum theory we have an
169: associative algebra generated by $a$ and $a^*$ with the relation
170: $aa^*-a^*a=\hbar$. Since in the quantum theory two seemingly different
171: polynomial expressions (such as $aa^*$ and $a^*a+\hbar$) can be the
172: same, there is a need for a preferred ordering of $a$ and $a^*$ in
173: monomials. The {\bfi{normal ordering}} is that ordering of $a$
174: and $a^*$
175: in monomials where all $a^*$'s are moved to the left of the $a$'s.
176: It is easy to see that every noncommutative polynomial in $a$ and
177: $a^*$ can be normally ordered by repeated use of the relation
178: $aa^*=a^*a+\hbar$; in the process of normal ordering, lower degree
179: monomials are generated with higher powers of $\hbar$. We give the
180: following proposition that guarantees that taking $\hbar \to 0$ we
181: recover the classical theory:
182:
183: \begin{prop}
184: Let $f$ and $g$ be noncommutative polynomials in $a$, $a^*$ and
185: $\hbar$. Viewing $f$ and $g$ as polynomials in commuting variables
186: $a$ and $a^*$, one can calculate $f\lp g$ using \gzit{ch5.pois}.
187: As noncommutative polynomials one can calculate the commutator
188: $[f,g]= fg-gf$. The two results are related by:
189: \lbeq{e.classlim}
190: \frac{i}{\hbar} [f,g] = f \lp g + O(\hbar)\,.
191: \eeq
192: One expresses this relation by saying that the quantum Lie product is
193: a {\bfi{deformation}} of the classical Lie product.
194: \end{prop}
195:
196: \begin{proof}
197: The order of the $a$ and $a^*$ does not matter since changing the
198: order we generate powers of $\hbar$. We use induction on the degree
199: of the polynomials. For degree zero and one, \gzit{e.classlim} holds.
200: Suppose it holds
201: for degree of $f$ smaller than $n$ and degree of $g$ one. If we write
202: $f= a^* S + Ta $ for some normally ordered polynomials $S$ and $T$
203: with degrees smaller than $n$, we see that for $g=a^*$:
204: \beqar
205: i[a^* S + Ta,a^*] &=& ia^* [S,a^*]+i\hbar T + i[T,a^*]a\nonumber\\
206: &=&\hbar a^*S\lp a^* + i\hbar T +\hbar T\lp a^* a +O(\hbar^2)\nonumber\\
207: &=& \hbar (a^* S)\lp a^* + \hbar (Ta)\lp a^* +O(\hbar^2)\nonumber\,,
208: \eeqar
209: and the result holds for $f$ arbitrary and $g=a^*$. For $g=a$ it goes
210: similar. Suppose the claim holds for all $g$ with degree $k$, with
211: $0\leq k\leq n$. Then for degree $n+1$ let us write
212: $g = a P+a^* Q + R$, where $P$, $Q$ and $R$ are polynomials of degree
213: strictly less than $n+1$. Then we have:
214: \beqar
215: \frac{i}{\hbar}[f,g] &=& \frac{i}{\hbar}[f, aP+a^*Q+R]\nonumber\\
216: &=& \frac{i}{\hbar}[f,a]P+\frac{i}{\hbar}a[f,P]+\frac{i}{\hbar}[f,a^*]Q
217: +\frac{i}{\hbar}a^*[f,Q]+\frac{i}{\hbar}[f,R]\nonumber\\
218: &=& f\lp aP+af\lp P+f\lp a^*Q+a^*f\lp Q+f\lp R + O(\hbar)\nonumber\\
219: &=& f\lp (aP) + f\lp (a^*Q) + f\lp R +O(\hbar)\nonumber\\
220: &=& f\lp (aP+a^* Q+R)+O(\hbar)\,.
221: \eeqar
222: And the proof is complete.
223: \end{proof}
224:
225: \bigskip
226: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
227: \bfi{Extension to the anharmonic case.}
228: The anharmonic oscillator can in principle be treated in a similar
229: fashion. Since the classical Lie product (the Poisson bracket) is
230: the same, we may proceed exactly as before, except that the formulas
231: involving the Hamiltonian are no longer valid. In particular, since
232: the frequency $\omega$ was determined by the Hamiltonian, it is now
233: an arbitrary constant. Thus there are multiple, inequivalent ways
234: of defining the quantities $a(t)$. Indeed, there is even more freedom
235: since the only important property to be preserved is the canonical
236: commutation relation.
237:
238: Generalizing the affine form of $a(t)$ in the harmonic case,
239: we choose it as an arbitrary affine combination of $q(t)$ and $p(t)$,
240: \lbeq{e.affine}
241: a = \lambda+ \mu q + i\nu p
242: \eeq
243: for suitable complex numbers $\mu$, $\nu$ and $\lambda$. As can be
244: easily verified, the canonical commutation relations \gzit{ch5.ccr}
245: are reproduced, so that the classical Lie product takes again the
246: form \gzit{ch5.pois}, exactly when the restriction
247: \[
248: 2\re \mu\ol \nu =1
249: \]
250: holds.
251: Having made a choice, we obtain a classical Hamiltonian $H=H(a,a^*)$ in
252: terms of $a$ and $a^*$. Using the Heisenberg dynamics and
253: \gzit{ch5.pois}, we obtain
254: \lbeq{e.normaldyn}
255: \dot a(t) = H\lp a(t) = -i \frac{\partial H}{\partial a^*}\,.
256: \eeq
257: We remark that if \gzit{ch5.ccr} holds for
258: $t=0$ then it holds for all $t$. Indeed, the derivative
259: of the left-hand side of \gzit{ch5.ccr} vanishes identically.
260:
261: Using a different choice of the parameters defining $a$ we get a
262: different variable $a'$, which is affinely related to the original $a$,
263: \lbeq{ch5.bog}
264: a' = \alpha + \beta a + \gamma a^*\,.
265: \eeq
266: The requirement that $a'$ satisfies the same commutation relations as
267: $a$ leads to the restriction
268: \lbeq{ch5.bog2}
269: |\beta|^2-|\gamma|^2 = 1.
270: \eeq
271: A transformation of the form \gzit{ch5.bog} satisfying \gzit{ch5.bog2}
272: is called a {\bfi{Bogoliubov transformation}}. Bogoliubov
273: transformations have important applications; for example, they were at
274: the heart of Hawkings' proof that black holes radiate.
275: \at{add references}
276: The generalization
277: of Bogoliubov transformations to systems of oscillating electron pairs
278: in metals is an important ingredient for the theory of Cooper pairs,
279: which explains superconductivity effects in metals at low temperature.
280: \at{add references}
281:
282: Different choices of the coefficients in the definition of $a$ lead
283: of course to different forms of $H(a,a^*)$; this means that different
284: Hamiltonians $H(a,a^*)$ can describe the same oscillator.
285: The particular choice above for the harmonic oscillator is the one
286: leading to $H(a,a^*)=E_0+\omega a^*a$, for which the dynamics
287: \gzit{e.normaldyn} takes the simple form $\dot a = -i\omega a$. In
288: theoretical physics there are different operators $a_k$ and
289: $a^{*}_{k}$ labeled by some parameter $k$.
290: One tries to find by means of Bogoliubov
291: transformations the simplest form of the Hamiltonian. The preferred
292: form is the form where $H$ is diagonalized: $H=\sum_k a^{*}_{k}a_k +
293: \ldots$, where the dots contain terms of higher order in the operators
294: $a_k$ and $a^{*}_{k}$.
295:
296: \bigskip
297: The quantization of an anharmonic oscillator is done as in the
298: harmonic case.
299: For each classical Hamiltonian polynomial in $a$ and $a^*$,
300: there is a unique normally ordered quantum version.
301: However, when modeling the same system both in a classical and in
302: a quantum setting, the coefficients of the quantum system in a normal
303: ordering of the operators must be taken to depend on $\hbar$,
304: and the form of this dependence is not determined by the
305: quantum-classical correspondence. Therefore, the best
306: fit of coefficients of $H$ to experimental data will generally produce
307: different optimal values in the classical and the quantum case.
308: In a quantum field theory, the coefficients will also be dependent on
309: the scale at which frequencies remain unresolved, giving so-called
310: {\bfi{running coupling constants}} which play an important role in
311: renormalization techniques.
312:
313:
314: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
315: \section{Representations of the Heisenberg algebra}
316: \label{section-reps-heis}
317:
318: We saw at the end of Section \ref{s.charm}
319: that the Heisenberg algebra $t(3,\Cz)$ can be considered
320: as being generated by $1,a$, and $a^*$ where $a$ and $a^*$ satisfy
321: the CCR $a\lp a^*=i$.
322:
323: In the classical case, we know a realization of these
324: commutation relations in terms of a Poisson bracket. In the quantum
325: case, we must find a representation in terms of operators in a Hilbert
326: space. The representations of physical interest are the
327: unitary representations, which represent the one as identity and
328: behave properly under the $*$-operation.
329: In this section we construct a unitary representation of the Heisenberg
330: algebra.
331:
332: In the quantized version of a classical theory the functions on
333: phase space become elements of some associative algebra $\Ez$.
334: For a representation we want to realize the algebra $\Ez$ as a
335: subalgebra of an algebra of linear operators.
336:
337: The approach of Schr\"odinger (1926) to this
338: problem was to take as Hilbert space the space of square integrable
339: complex-valued functions $\psi$ on $\Rz^3$; then the Schr\"odinger
340: equation for the dynamics of a pure state takes the form of a wave
341: equation, which was familiar to physicists at that time and hence
342: came to dominate quantum mechanics. The approach taken by Schr\"odinger
343: proved to be very successful and is also presented in many quantum
344: physics textbooks.
345:
346: For a single particle, Schr\"odinger's representation
347: is quite intuitive: the real-valued function $|\psi(x)|^2$ is the
348: probability density for the presence of a particle at a point $x$
349: in space.
350: For multiparticle systems, the intuitive advantages of Schr\"odinger's
351: representation is no longer given, as the wave functions are no longer
352: in physical space $\Rz^3$ but, for $n$ particles, in an abstract
353: $3n$-dimensional configuration space. For systems involving an
354: unconserved number of
355: particles, in particular for interactions with light, and for systems
356: in the thermodynamic limit, things are even more complicated since
357: the configuration space becomes infinite-dimensional, and the wave
358: function representation becomes unwieldy. Nevertheless, there are
359: interesting papers using the resulting functional Schr\"odinger
360: equation to illuminate the relations between classical solitons and
361: quantum bound states (see, e.g., \sca{Jackiw} \cite{jackiw77}).
362:
363: One year earlier than Schr\"odinger, Heisenberg invented his
364: infinite-dimensional
365: matrix algebra. We present Heisenberg's approach since it generalizes
366: easily to the most complex quantum systems, including the universe as
367: a whole.
368:
369: \bigskip
370: We now look at an arbitrary unitary representation
371: $J:\Lz\to \Lin\Hz$ in a Euclidean space $\Hz$
372: satisfying
373: \[
374: J(a^*)=J(a)^*, ~~~J(1)=1.
375: \]
376: We shall write the
377: operators corresponding to $a$ and $a^*$ in the representation again
378: by $a$ and $a^*$ (rather than using $J(a)$, etc.), in order to
379: avoid clumsy notation. This will not cause problems since the
380: representation turns out to be faithful. Then the operator
381: \[
382: n:=\frac{1}{\hbar}a^*a,
383: \]
384: for reasons that will soon be apparent, is called the {\bfi{number
385: operator}}, satisfies the commutation relations
386: \[
387: [a,n]= a\,, ~~~ [a^*,n]=-a^*\,,
388: \]
389: as is easily checked. This implies that the vector space generated by
390: $1$, $a$, $a^*$ and $n$ is closed under the commutator, and hence
391: forms a Lie $*$-algebra $\Lz$ with the quantum Lie product,
392: called the {\bfi{oscillator algebra}} $os(1)$. In this
393: section (as always when classifying unitary representations),
394: it will be more convenient to work directly with commutators.
395:
396:
397: \bigskip
398: We now illustrate an important technique in representation theory,
399: which in many cases of interest provides all irreducible representations
400: of a certain kind. See Section \ref{s.highest} for some other
401: applications.
402:
403: We define the {\bfi{Verma module}}
404: \at{check adequateness of the name}
405: corresponding to a complex number $\lambda$ by
406: \[
407: V_\lambda = \{ \psi \in \ol\Hz \mid n\psi = \lambda \psi\} \,,
408: \]
409: where the Hilbert space $\ol \Hz$ is the closure of $\Hz$.
410: Since $V_\lambda$ is a vector space closed under multiplication by
411: operators from $\Lz_0$, the space $V_\lambda$ is an $\Lz_0$-module.
412:
413: If $V_\lambda \neq 0$, i.e., if it contains a nonzero vector,
414: then $\lambda$ is an eigenvalue of $n$, and
415: any nonzero $\psi\in V_\lambda$ is a corresponding eigenvector.
416: Thus the nonzero Verma modules are just the eigenspaces of the
417: eigenvalues of $n$.
418: Since we consider here only unitary representations where * is the
419: adjoint, this implies that
420: \[
421: \lambda=\frac{\psi^*
422: n\psi}{\psi^*\psi}=\frac{\|a\psi\|^2}{\hbar \|\psi\|^2}\geq 0
423: \]
424: is real and nonnegative. Noting that in general $n$ is Hermitian, we
425: now make the slightly stronger assumption that $n$ is self-adjoint
426: as a densely defined operator of the Hilbert space $\ol\Hz$. Then
427: the spectral theorem implies that the infimum
428: \[
429: \widehat\lambda = \inf_{\psi\ne 0} \frac{\psi^* n\psi}{\psi^*\psi}
430: \]
431: is a real and nonnegative number, attained for some $\widehat\psi\ne
432: 0$, and $\widehat\psi$ is an eigenvector of $n$ corresponding to the
433: eigenvalue $\widehat\lambda$. Thus $V_{\widehat\lambda} \neq 0$.
434: Now consider
435: an arbitrary $\lambda$ with $V_\lambda \neq 0$ and a nonzero
436: $\psi\in V_\lambda$. Then \at{why?}
437: $na\psi = (\lambda - 1)a\psi$, hence
438: $a\psi \in V_{\lambda-1}$. If $a\psi =0$ then
439: \[
440: \lambda\psi^*\psi=\psi^*n\psi = \frac{1}{\hbar } \psi^* a^* a \psi = 0,
441: \]
442: hence $\lambda=0$; and if $a\psi \ne 0$ then $V_{\lambda-1}\ne 0$, and
443: $\lambda-1$ is an eigenvalue of $n$. In the latter case, we can repeat
444: the step once, or more often. But since all eigenvalues are nonnegative,
445: this can happen only a finite number of times, and ultimately we must
446: end up with the other alternative. Hence zero is an eigenvalue
447: (in particular $\widehat\lambda=0$) and $\lambda-n=0$ for some
448: nonnegative
449: integer $n$. Thus the only possible eigenvalues are
450: nonnegative integers. That all these actually are eigenvalues follows
451: by a similar argument. Indeed, with $\psi$ as before, we have
452: \[
453: na^*\psi = ([n,a^*]+a^*n)\psi = (n+1)a^*\psi = (\lambda+1)a^*\psi,
454: \]
455: hence $a^*\psi\in V_{\lambda+1}$. Since
456: \[
457: \|a^*\psi\|^2=\psi^* a a^* \psi = \psi^* (\hbar+a^*a)\psi =
458: \hbar\|\psi\|^2+\|a\psi\|^2\ge \hbar\|\psi\|^2>0,
459: \]
460: $V_{\lambda+1}\ne 0$ and $a^*\psi$ is an eigenvector for the
461: eigenvalue $\lambda+1$. By induction, we reach all positive integers
462: from $V_{\widehat\lambda}=V_0$.
463: Thus we have proved the following theorem:
464:
465: \begin{thm}
466: In a representation in which $n$ is self-adjoint,
467: a Verma module $V_\lambda$ of the oscillator algebra $\Lz$ is nonzero
468: if and only if $\lambda$ is a nonnegative integer.
469:
470: In particular, the spectrum of the Hamiltonian $H=\omega a^*a$ of a
471: quantum harmonic oscillator consists of the nonnegative integral
472: multiples of $\omega\hbar$.
473: \end{thm}
474:
475: \at{The self-adjointness requirement is just the requirement that
476: we actually represent the group! mention this somewhere in an earlier
477: chapter!}
478:
479: The results obtained justify the following terminology. The operator
480: $a$ is called a {\bfi{lowering operator}}, since its application
481: to an eigenstate of the number operator $n$ lowers the associated
482: eigenvalue by one. The operator $a^*$ is called a
483: {\bfi{raising operator}}, since
484: its application to an eigenstate of the number operator $n$ raises
485: the associated eigenvalue by one. Together, the operators $a$
486: and $a^*$ are called {\bfi{ladder operators}}. A unit vector in the
487: Verma module $V_0$ is called a {\bfi{ground state}} (in the second
488: quantized language of quantum field theory a {\bfi{vacuum vector}}).
489: For a {\bfi{ground state}}, i.e., a nonzero vector $\psi\in V_0$,
490: we have $n\psi=0$,
491: hence $\|a\psi\|^2=\psi^*a^*a\psi=\hbar \psi^*n\psi=0$ and therefore
492: $a\psi=0$. Thus the ground state is annihilated by the lowering
493: operator. Therefore $a$ is also called an {\bfi{annihilation
494: operator}}; if this term is used then $a^*$ is called a
495: {\bfi{creation operator}}.
496:
497:
498: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
499: \section{Bras and Kets}\label{s.braket}
500:
501: In his groundbreaking work on quantum mechanics, Dirac introduced a
502: notation for vectors and operators that is widely used by physicists
503: but is quite different from what mathematicians are used to. Dirac's
504: bra-ket calculus is not very well defined in the way actually used
505: by physicists, since the basis vectors considered in the calculus
506: do not necessarily lie in the Hilbert space in which everything
507: should happen from a strictly axiomatic point of view.
508:
509: We define here a precise version of Dirac's bra-ket calculus,
510: which can also satisfy mathematicians. Instead of working in a
511: Hilbert space we consider a fixed dense subspace which we denote by
512: $\Hz$. Thus $\Hz$ is a vector space with a Hermitian inner product
513: $\<\cdot|\cdot\>$, antilinear in the first argument and linear in
514: the second, such that $\<\psi|\psi\>$ is always real and nonnegative,
515: and the relation
516: \lbeq{e.sym}
517: \<\phi|\psi\>^*=\<\psi|\phi\>
518: \eeq
519: holds, where $\alpha^*$ denotes the complex conjugate of a number
520: $\alpha\in\Cz$. The inner product defines a Euclidean norm
521: $\|\psi\|:=\sqrt{\<\psi|\psi\>}$, and the Hilbert space is the
522: closure $\ol\Hz$ of $\Hz$ in the topology induced by this norm.
523: We refer to the elements of $\Hz$ as {\bfi{smooth vectors}} since
524: they correspond in the important special case $\Hz=C^\infty(\Rz)$
525: to arbitrarily often differentiable functions.
526:
527: Every smooth vector $\psi$ defines a continuous linear functional,
528: denoted by $\psi^*$, which maps $\phi\in\Hz$ to the complex number
529: \lbeq{e.dual}
530: \psi^*(\phi):=\<\psi|\phi\>.
531: \eeq
532: Dirac's idea was to turn this formula into a more suggestive
533: form by splitting the bracket $\<\psi,\phi\>$ into a
534: {\bfi{bra}} $\<\psi|$, standing for $\psi^*$, and a {\bfi{ket}}
535: $|\phi\>$, standing for $\phi$, and deleting the now superfluous
536: parentheses. Then the formula becomes
537: \lbeq{e.bra-ket}
538: \<\psi|\,|\phi\>:=\<\psi|\phi\>,
539: \eeq
540: which just asks us to replace two adjacent vertical bars by a single
541: one.
542:
543: If $\Hz$ is itself a Hilbert space (and in particular, if the
544: dimension of $\Hz$ is finite) then it is not difficult to see
545: that all continuous linear functionals arise in this way.
546: However, in many interesting infinite-dimensional vector spaces $\Hz$,
547: the situation is different.
548: For example, if $\Hz=C^\infty(\Rz)$ and $z\in\Rz$ then the
549: mapping $\delta_z$ which maps $\psi\in\Hz$ to
550: \[
551: \delta_z(\psi):=\psi(z)
552: \]
553: is a continuous linear functional which cannot be obtained as
554: $\psi^*$ for some smooth vector $\psi$.
555:
556: We can accommodate this in the bra-ket calculus by allowing
557: as bras {\em all} continuous linear functionals rather than only those
558: which have the form $\psi^*$ with $\psi\in\Hz$. We simply need to
559: label the continuous linear functional as bras $\<\psi|$ with symbols
560: $\psi$ from a set $\Hz^*$ such that the functionals of the form
561: $\psi^*$ with $\psi\in\Hz$ get the label $\psi$. The set $\Hz^*$ can
562: be made canonically into a vector space containing $\Hz$ as a subspace
563: by requiring the mapping $*:\psi\to \psi^*:=\<\psi|$ to be antilinear.
564: Then $\Hz^*=\Hz$ in case $\Hz$ is a Hilbert space, but in general
565: $\Hz$ may be a proper subspace of $\Hz^*$. Since the inner product
566: extends continuously from $\Hz$ to the Hilbert space completion
567: $\ol\Hz$, every element of $\ol\Hz$ defines a continuous linear
568: function. Thus, in general, the Hilbert space $\ol\Hz$ sits somewhere
569: in between $\Hz$ and $\Hz^*$,
570: \lbeq{e.triple}
571: \Hz \subseteq \ol\Hz \subseteq \Hz^*.
572: \eeq
573: Frequently, some extra ''nuclear'' structure on $\Hz$ is assumed
574: which turns \gzit{e.triple} into a so-called \bfi{Gelfand triple}
575: or \bfi{rigged Hilbert space}
576: (see, e.g., \sca{Maurin} \cite{Mau}, \sca{Bohm \& Gadella} \cite{BohG},
577: and for applications to resonances \sca{Kukulin} et al.
578: \cite{KukKH}); however, on the
579: level of our discussion, we don't need this extra structure.
580:
581: If $\Hz=C^\infty(\Rz)$, physicists call the vectors $\psi\in\Hz^*$
582: \bfi{wave functions}\index{wave function} -- well being aware that
583: they are not always functions in the standard sense --,
584: and write them with a dummy argument $x$ as
585: $\psi(x)$. For example, they consider $\delta_z$
586: to be a \bfi{shifted delta function}\index{delta function}, and write
587: it as $\delta(x-z)$.
588:
589: A wave function which is in the Hilbert space $\ol\Hz\subseteq \Hz^*$
590: is called {\bfi{normalizable}}, the remaining wave functions
591: are called {\bfi{non-normalizable}}. In mathematical terms, the
592: normalizable wave functions are equivalence classes of square integrable
593: functions, with two functions being regarded as equivalent when they
594: differ only on a set of measure zero. The shifted delta functions are
595: examples of non-normalizable wave functions.
596:
597: For a general Euclidean space $\Hz$, we refer to the elements of
598: $\Hz^*$ as {\bfi{rough vectors}} since they correspond in the
599: special case $\Hz=C^\infty(\Rz)$ to functions that are less smooth,
600: possibly not even continuous, and possibly (as in case of the
601: $\delta_z$) not functions at all.
602:
603:
604:
605: Having extended the bra-ket notation to allow rough vectors as labels
606: in bras, the symmetry property \gzit{e.sym} is lost. To restore that,
607: we simply extend the inner product to enforce the validity of
608: \gzit{e.sym} by defining $\<\psi|\phi\>:=\<\phi|\psi\>^*$ if
609: $\phi\in\Hz^*$ and $\psi\in\Hz$. This can be done consistently,
610: and implies that now kets can be labeled by rough vectors, too.
611: But now the formula \gzit{e.bra-ket} makes trouble. What is
612: $\<\psi|\phi\>$ when both $\phi$ and $\psi$ are rough vectors?
613: In general, there is no solution; this product cannot be always
614: defined. However, one can consistently define it in certain cases,
615: namely when $\phi$ is in some subspace $\tilde\Hz$ of $\Hz^*$ and
616: the linear functional $\psi^*$ defined at first only on $\Hz$
617: can be extended to $\tilde\Hz$ by some limiting procedure.
618: We won't list here the various possibilities; our usage of bras
619: and kets will be restricted to cases where at least one of the
620: two labels in an inner product is smooth.
621:
622: The main use of Dirac's notation is for the specification of
623: vectors and matrices in a particular representation of the
624: algebra of quantities. We first review the notation in the case
625: where a countable orthonormal basis of smooth states is available.
626: In this case there is a countable set $K$ of {\bfi{labels}}
627: such that the basis consists of the kets $|k\>$ with $k\in K$,
628: and orthogonality implies that
629: \[
630: \<j|k\> = \delta_{jk}\,,
631: \]
632: and the {\bfi{resolution of unity}}
633: \[
634: \sum_k |k\>\<k| = 1\,.
635: \]
636: In the finite-dimensional case, there is a close correspondence to
637: the notation of linear algebra if we take $\<k|$ to be the $k$th unit
638: row vector with a 1 in position $k$ and zeros elsewhere, and
639: $|k\>$ to be its transpose, the $k$th unit column vector.
640: \[
641: x = \sum_k x_k|k\>
642: \mbox{~~~ represents the vector $x=(x_k)$},
643: \]
644: and
645: \[
646: \<k|x = x_k
647: \]
648: gives the components of $x$.
649: \[
650: A = \sum_{jk} |j\>A_{jk}\<k|
651: \mbox{~~~ represents the matrix $A=(A_{jk})$},
652: \]
653: \[
654: \<j|A = \sum_{k} A_{jk}\<k|
655: \mbox{~~~ represents $A_{j:}$, the $j$th row of $A$},
656: \]
657: \[
658: A|k\> = \sum_{j} |j\>A_{jk}
659: \mbox{~~~ represents $A_{:k}$, the $k$th column of $A$},
660: \]
661: and
662: \[
663: \<j|A|k\>=A_{jk}
664: \]
665: gives the matrix entries of $A$. Compared to the standard linear
666: algebra notation there is no gain.
667:
668: The situation is different when $K$ is a structured set, for example
669: a set of pairs $(k,s)$ where $k$ is a momentum label and $s$ a spin
670: label, or other such sets arising naturally in the dynamical symmetry
671: approach of Section \ref{s.chains}. Then the index notation becomes
672: somewhat cumbersome to comprehend, and the more lengthy bra-ket notation
673: is superior.
674:
675: \at{continuous ''basis'' case missing}
676:
677:
678: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
679: \section{Boson Fock space}\label{s.bfock}
680:
681: As we have seen in Section \ref{section-reps-heis}, every nice unitary
682: representation of the oscillator algebra contains a ground state
683: $\widehat\psi$ of norm 1, and hence the representation contains the
684: vectors $(a^*)^k \widehat\psi$ ($k=0,1,2,\dots$). Their span defines a
685: Euclidean vector space \idx{$\Fz_+$} whose closure $\ol\Fz_+$
686: is a Hilbert space, called the {\bfi{single mode bosonic Fock
687: space}}, or simply Fock space. Clearly, $\Fz_+$ is closed
688: under the action of $\Lz$, hence we have a unitary representation of
689: $\Lz$ on $\Fz_+$.
690: It is not difficult to see that different choices of the ground
691: state either define the same Fock space (if the ground states
692: differ only by a phase) or orthogonal Fock spaces.
693: Indeed, if $\Fz\subset \Fz_+$
694: is an invariant submodule, it needs to have a vector $\tilde\psi_0$,
695: which necessarily coincides with the ground state of $\Fz_+$ up to a
696: complex number. Thus an arbitrary unitary representation is a direct
697: sum of Fock spaces. Thus the representations on a Fock space
698: are {\em irreducible} representations. We shall show in
699: a moment that the unitary representation on a Fock space is
700: essentially unique. This is the content of the celebrated {\bf
701: \idx{Stone--Von Neumann theorem}}, which actually is about the
702: representation of the Heisenberg group.
703:
704: Bosonic Fock spaces with more degrees of freedom are obtained by taking
705: tensor products of the Fock space with one degree of freedom, and
706: describe systems of quantum oscillators. As we shall see in
707: Chapter \ref{c.spin}, there is also a fermionic counterpart of Fock
708: spaces, which are related to so-called
709: Clifford algebras. The single mode case describes a so-called
710: {\bfi{qubit}} and is simply the vector space $\Cz^2$; the general
711: case is a tensor product of these, and describes systems of qubits.
712:
713: We now study the structure of $\Fz_+$ for a given ground state
714: $\widehat\psi$ of norm 1 in more detail. The properties found will
715: lead to a construction of a Hilbert space which actually contains a
716: representation of the Heisenberg algebra (which, so far, we simply
717: had assumed).
718:
719: \begin{prop}\label{prop-5.5}
720: The vectors
721: \lbeq{e.kket}
722: |k\> := \frac{1}{k!} (a^{*})^{k} \widehat \psi~~~(k=0,1,2,\dots)
723: \eeq
724: satisfy the relations
725: \[
726: a^*|k-1\> = k |k\> \nn,~~~ a|k\> = \hbar |k-1\> \nn,~~~
727: n|k\> = k |k\>,
728: \]
729: \[
730: \<k|k'\> = \frac{\hbar^k}{k!}\delta_{kk'}.
731: \]
732: \end{prop}
733: \begin{proof}
734: The first relation is just definition. For the second observe that
735: $a\widehat\psi = 0$ and $[a,(a^{*})^k]=k(a^{*})^{k-1}$. For the third,
736: just combine $n=\frac{1}{\hbar}a^*a$ and the first and second
737: relation. For the fourth relation we have $\<k| =
738: \frac{1}{k!}(\widehat\psi)^* a^k$ and $\<k|k'\>=0$ if $k\neq
739: k'$ since eigenvectors of a Hermitian operator corresponding to
740: different eigenvalues are orthogonal.
741: So only the normalization needs to be checked:
742: \[
743: \< k| k\> = \frac{1}{k^2 }\<k-1|a a^*|k-1\> = \frac{1}{k^2}\<k-1 |
744: \hbar +\hbar n |k-1\> = \frac{\hbar}{k}\<k-1|k-1\>\, .
745: \]
746: Using induction the fourth equality follows.
747: \end{proof}
748:
749: In the Fock space $\Fz_+$, the vectors are by definition the linear
750: combinations
751: \[
752: \psi = \sum_{k=0}^\infty \psi_k |k\>\,,~~~\psi_k\in \Cz\,.
753: \]
754: Proposition \ref{prop-5.5} gives us the relations \lbeq{fock-op}
755: (a\psi)_k = \hbar \psi_{k+1}\,,~~~(a^*\psi)_k = k\psi_{k-1}\,,~~~
756: (n\psi)_k = k\psi_k\,, \eeq and $ \varphi^* \psi = (\sum \varphi_k
757: |k\> )^* \sum \psi_l |l\> = \sum \frac{\hbar^k}{k!}\bar\varphi_k
758: \psi_k$, hence \lbeq{fock-ip}
759: \varphi^* \psi = \sum \frac{\hbar^k}{k!}\bar\varphi_k \psi_k\,.
760: \eeq
761: Equations \gzit{fock-op} and \gzit{fock-ip} are an equivalent
762: description of the equations of Proposition \ref{prop-5.5}.
763:
764: We now define $\Hz$ as the closure of $\Fz_+$. Then the operators
765: $a,a^*$ and $n$ are defined on a dense subset. Previously we have
766: seen that if the canonical commutation relations admit an
767: irreducible representation, then it has to be of the form as
768: described by Proposition \ref{prop-5.5}. But now we can say more:
769:
770: The set $\Hz$ of vectors $\psi\in\Cz^\infty$ with finite norm
771: \[
772: \|\psi\|:=\sqrt{\sum_{k=0}^\infty \frac{\hbar^k}{k!}|\psi_k|^2}
773: \]
774: is a Hilbert space with inner product \gzit{fock-ip}, on which
775: the definitions \ref{fock-op} give densely defined operators
776: $a,a^*,n$. The components of $a^* \psi$ grow significantly
777: faster than those of $\psi$, so that $a^* \psi\in \Hz$ only for
778: $\psi$ in a proper subspace of $\Hz$. This subspace is dense,
779: since it contains the dense subset of $\psi$ with only finitely many
780: nonzero entries. Note that the operators $1$, $a$, $a^*$ and $n$,
781: and hence all elements of $\Lz$ are represented
782: by infinite tridiagonal matrices. This is the representation of the
783: quantum harmonic oscillator discovered by Heisenberg in his
784: groundbreaking paper \cite{Hei}.
785:
786: It is now easy to check that the operators
787: $1$, $a$, $a^*$ and $n$ satisfy the canonical commutation relations,
788: that $a^*$ is the Hermitian conjugate of $a$, and that $n=a^*a$.
789: Thus we have a representation of $\Lz$. The representation is
790: irreducible since acting repeatedly with $a^*\in\Lz$ on the vector
791: $\widehat\psi$ with entries $\widehat\psi_k=\delta_{k0}$ (the ground
792: state) gives a basis of $\Hz$. Combining this with the uniqueness
793: statement obtained before, we arrive at the
794: following theorem of Stone and Von Neumann (but essentially already
795: obtained in \cite{Hei}): \at{repeat why unique}
796:
797: \begin{thm}
798: The canonical commutation relations admit an (up to equivalence)
799: unique irreducible unitary representation on a Hilbert space such that
800: the action of $a$, $a^*$ and $n=a^*a$ is defined on a dense subspace
801: and $n$ is self-adjoint.
802: \end{thm}
803:
804: The theorem holds with a similar proof for arbitrary finite-dimensional
805: Heisenberg algebras coming from a nondegenerate alternating form.
806: It fails spectacularly in infinite dimensions. In this case there are
807: uncountably many inequivalent representations; see, e.g.,
808: \sca{Barton} \cite{Bar} for an (in spite of the title of the book)
809: elementary discussion of these. Their existence
810: is one of the main stumbling blocks for extending quantum mechanics
811: to quantum field theory.
812:
813:
814: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
815: \section{Bargmann--Fock representation}\label{s.bargmann}
816:
817: We present an important but easy representation of the Heisenberg
818: algebra $h(n)$, which will be useful to us when we study coherent
819: states in Section \ref{sec-coh-state}.
820: \at{where is $os(n)$ defined? Not needed here, but it should be
821: defined somewhere, for completeness.}
822: Consider the vector space of
823: complex polynomials in $n$ variables $\Cz[z_1,\ldots,z_n]$. We then
824: identify $a_k$ and $a^{*}_{k}$ with the operators defined
825: by\footnote{Remember the transformations
826: $a_k=\frac{p_k-iq_k}{\sqrt{2}}$ and
827: $a_{k}^{*}=\frac{p_k+iq_k}{\sqrt{2}}$.}
828: \[
829: (a_k p)(z_1,\ldots,z_n):= z_kp(z_1,\ldots,z_n)\,,
830: \]
831: and
832: \[
833: (a^{*}_{k}p)(z_1,\ldots,z_n):=
834: \frac{\partial}{\partial z_k}p(z_1,\ldots,z_n)\,.
835: \]
836: It is easy to check that this indeed defines a representation. We can
837: even make a unitary representation out of this. For that purpose we
838: consider the vector space $\Hz$ of all entire functions on $\Cz^n$ with
839: finite norm with respect to the inner product
840: \[
841: \< f|g\> = \int_{\Cz^n} \overline{f(z)}g(z)e^{-\bar z\cdot
842: z}\,.
843: \]
844: The space $\Hz$ with the above inner product is a Euclidean space; its
845: closure is a Hilbert space, the multi-dimensional version of the
846: Bargmann--Fock space described in Section \ref{s.bfock}.
847: \at{check name. It should be the Bargmann-Fock representation only.
848: The space is called Fock space almost universally}
849:
850: The operators $a_k$ and $a^{*}_{k}$ are adjoints of each
851: other. An orthogonal
852: basis is given by the monomials:
853: \[
854: \< z_{1}^{k_1}\ldots z_{n}^{k_n}|z_{1}^{l_1}\ldots
855: z_{n}^{l_n}\> = \prod_{i=1}^{n}k_i!\delta_{k_i,l_i}\,.
856: \]
857: From the discussion in Section \ref{sec-quad-rep} it follows that the
858: quadratic expressions
859: modulo the linear expressions in the elements $z_i$ and
860: $\frac{\partial}{\partial z_k}$ form the Lie algebra $sp(2n,\Cz)$.
861: Taking
862: all quadratic expressions (so not modding out by the linear
863: polynomials) in the elements $z_i$ and
864: $\frac{\partial}{\partial z_k}$ one obtains a central extension of
865: $isp(2n)$.
866:
867: The above representation is irreducible (one sees rather quickly that
868: starting with $1$, acting with $p_k$ gives all entire functions) and
869: is called the Bargmann--Fock representation. By the Stone--Von Neumann
870: theorem, which says that there is only one irreducible representation
871: of the Heisenberg algebra, the Bargmann--Fock representation is up to
872: isomorphism the only irreducible representation of the Heisenberg
873: algebra.
874:
875: We have seen in Section \ref{sec-quad-rep}
876: that the quadratic expressions (modulo linear terms) in
877: the $q_i$ and the $p_k$ rotate the generators $q_i$ and $p_k$ into
878: each other under the action of the Lie product. In other words,
879: the action of the quadratic expressions
880: builds a representation of $sp(2n,\Cz)$ inside the Bargmann--Fock
881: representation. That this happens is not so strange. Let us consider
882: the automorphism group of the Heisenberg algebra, consisting of all
883: the invertible maps $h(n)\to h(n)$ preserving the Lie product. But
884: from equation \gzit{sympl.comm} we see that the automorphism group
885: contains the group $Sp(2n,\Cz)$. Now let us denote the above given
886: Bargmann--Fock representation by $U: h(n)\to \Lin(H)$, then using
887: $Sp(2n,\Cz)$ we get a new representation of the Heisenberg algebra
888: as follows. For each $g\in Sp(2n,\Cz)$ we consider the representation
889: \[
890: U_g: h(n)\to \Lin(H)\,,~~~ U_g(x) = U(gx)\,.
891: \]
892: Since $Sp(2n,\Cz)\subset \Aut(h(n))$ the $U_g$ are indeed
893: representations. But the unitary irreducible representation of $h(n)$
894: is unique, up to isomorphism, and hence there must be a unitary
895: operator $R(g)$ such that
896: \[
897: U_g = R(g)UR(g)^{-1}\,.
898: \]
899: It is clear that the $R(g)$ are
900: determined up to a sign. Thus $R(g)R(h)=\pm R(gh)$ and we say that the
901: $R(g)$ form a
902: {\bfi{projective representation}}\index{representation!projective}
903: of the group $Sp(2n,\Cz)$.
904: This representation is called the {\bfi{metaplectic
905: representation}}\index{representation!metaplectic}.
906: The operators $R(g)$ themselves form a group,
907: closely related to the {\bfi{metaplectic group}} $Mp(2n,\Rz)$,
908: the universal covering group of the Lie
909: algebra $sp(2n,\Rz)$. The metaplectic group is a two-fold cover of
910: $Sp(2n,\Rz)$, hence has a center of order 2, while our group has
911: the multiplicative group of the reals as center. Factoring out the
912: positive reals leaves the metaplectic group.
913:
914:
915:
916: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
917: \section{Coherent states for the harmonic oscillator}
918: \label{sec-coh-state}
919:
920: Coherent states were introduced in 1963 by \sca{Glauber} \cite{glauber},
921: who recognized their importance in quantum optics; he received in 2005
922: the Nobel prize for his work in this direction. But the notion
923: of a coherent state (without the name) was already
924: introduced by Erwin Schr\"odinger \cite{schroedinger} in 1926 when
925: he was looking for solutions to the Schr\"odinger equation that
926: satisfy the {\bfi{Heisenberg uncertainty relation}}
927: \lbeq{e.qmunc}
928: \Delta p\Delta q \geq \frac{\hbar}{2},
929: \eeq
930: \at{derive this in Chapter 5} where $\Delta x$ denotes the
931: variance of a quantity $x$. Schr\"odinger was looking for states
932: that were as classical as possible, having equality $\Delta p\Delta q =
933: \frac{\hbar}{2}$. The coherent states, and only these satisfy
934: equality; they therefore build a bridge between classical physics
935: and quantum physics that deepened as the notion of coherent states
936: was extended to more general situations.
937:
938: \at{remark that the Schr\"odinger equation is easy to solve either
939: in a spectral representation, or in a coherent state representation.}
940:
941: To introduce Glauber's coherent states, we remind the reader
942: that for the harmonic oscillator we constructed the Fock
943: space $\Hz$ of $\psi=(\psi_k)_{k\geq 0}$ satisfying
944: \lbeq{7.norm}
945: \sum_{k=0}^{\infty} \frac{\hbar^k}{k!} \ol \psi_k \psi_k <
946: \infty\,.
947: \eeq
948: One may regard $\psi$ either as a vector with infinitely many
949: components, or as an infinite sequence.
950: Equivalently, in Dirac's bra-ket notation, the $\psi_k$'s are the
951: complex coefficients in the expansion of $\psi$ with respect to an
952: eigenbasis $|k\>$ of the number operator,
953: $\psi = \sum \psi_k|k\>$.
954: The inner product is given by
955: \[
956: \varphi^* \psi = \sum_{k=0}^{\infty} \frac{\hbar^k}{k!} \ol
957: \varphi_k \psi_k\,.
958: \]
959: The operators $a$, $a^*$ and $n$ act as $(a\psi)_k = \hbar
960: \psi_{k+1}$, $(a^* \psi)_k = k \psi_{k-1}$ and $(n\psi)_k =
961: k\psi_k$. We now define a {\bfi{coherent state}} for the harmonic
962: oscillator to be a vector of the form
963: \[
964: |\lambda,z\> := (\ol \lambda, \ol \lambda \ol z, \ol \lambda
965: \ol z^2,\ldots)\,; ~~~(\lambda,z\in\Cz)
966: \]
967: in other words, a state $\psi$ with coefficients
968: $\psi_k = \ol \lambda \ol z^k$. By \gzit{e.kket},
969: we can write
970: \[
971: |\lambda,z\> = \sum_{k=0}^{\infty} \ol \lambda \frac{\ol
972: z^k}{k!}(a^{*})^{k}
973: \hat\psi\,,
974: \]
975: where $\hat\psi$ is the ground state. Even more, we have
976: \lbeq{exp.in-n}
977: |\lambda,z\> = \sum_{k\geq 0} \bar \lambda \bar z^k |k\>\,,
978: \eeq
979: and we see that $\psi\in\Hz$ since
980: \[
981: \sum_{k\geq 0} \hbar^k |\lambda|^2 \frac{|z|^{2k}}{k!} =|\lambda|^2
982: e^{\hbar |z|^2}<\infty\,.
983: \]
984: The inner
985: product between two coherent states is given by
986: \[
987: \< \lambda',z'|\lambda,z\> = \sum_{k=0}^{\infty} \frac{\hbar^k
988: z'^{k}\ol z^{k}}{k!}\lambda'\ol \lambda = \lambda'\ol \lambda
989: e^{\hbar z'\ol z}\,.
990: \]
991: It is easy to see that
992: \[
993: \< \lambda,z | n\> =\frac{\lambda \hbar^nz^n}{n!}\,.
994: \]
995: Suppose $\psi$ is an element of $\Hz$, then
996: \lbeq{7.coh.fun} %
997: \<\lambda,z|\psi\> = \lambda \sum_{k=0}^{\infty}
998: \frac{(\hbar z)^k}{k!} \psi_k \equiv \lambda \psi(z)\,,%
999: \eeq%
1000: which defines the function $\psi(z)$ corresponding to
1001: $\psi$. Conversely, given an analytic function $g$
1002: \[
1003: g(z) = \sum_{k\geq 0} \frac{(\hbar z)^k}{k!}g_k\,,
1004: \]
1005: with $\sum_{k\geq 0}\frac{\hbar^2|g_k|^2}{k!}<\infty$ we assign to
1006: $g$ the
1007: element $\psi_g= (g_k)_{k\geq 0}$ in $\Hz$. We claim that
1008: $\psi \mapsto \psi(z)$ is a map from $\Hz$ to the set of analytic
1009: functions. In order to prove the claim we have to prove that the
1010: power series \gzit{7.coh.fun} converges everywhere. We calculate
1011: the radius of convergence $R$
1012: \[
1013: R = \limsup_{k\to \infty}
1014: \frac{1}{\sqrt[k]{\frac{\hbar^k\psi_k}{k!}}} = \limsup_{k\to
1015: \infty}\sqrt[k]{\frac{k!}{\hbar^k \psi_k}} \to \infty
1016: \]
1017: since $\psi_k$ satisfies \gzit{7.norm}. Hence the function
1018: $\psi(z)$ is analytic everywhere. The state $\psi$ is uniquely
1019: described by the function $\psi(z)$ in the sense that
1020: $\psi(z) = 0 \Leftrightarrow \psi =0$, since
1021: \[
1022: \frac{1}{\lambda \hbar^k}\frac{d^k}{dz^k} \psi(z)\Big|_{z=0} =
1023: \psi_k\,.
1024: \]
1025: The inner product between $\varphi$ and $\psi$ now becomes
1026: \[
1027: \varphi^* \psi = \sum_{k=0}^{\infty}
1028: \frac{1}{k!|\lambda|^2\hbar^k} \Big[\frac{d^k}{dz^k}
1029: \ol\varphi(z) \frac{d^k}{dz^k} \psi(z)\Big]_{z=0}\,.
1030: \]
1031: So we can use the powerful theorems of complex analysis to deal
1032: with the states in the Hilbert space $\Hz$. For the relations between
1033: complex analysis and coherent states, including important
1034: generalizations to coherent states associated with other Lie groups,
1035: see \sca{Perelomov} \cite{perelomov}, \sca{Upmeier} \cite{upmeier},
1036: \sca{Faraut \& Koranyi} \cite{faraut}.
1037:
1038: Every element in $\Hz$ is a linear combination
1039: of coherent states, but the combination is in general not unique.
1040: For the harmonic oscillator a set of finitely many coherent states
1041: $|\lambda,z\>$ with different $z$ is linearly independent, since
1042: suppose
1043: \[
1044: |v\>:=\sum_{i=1}^{n} |\lambda_i,z_i \> = 0\,, ~~~ \lambda_i\neq
1045: 0\,,
1046: \]
1047: then it follows that
1048: \[
1049: \<\mu,w |v\> = \sum_{i=1}^{\infty} \mu\lambda_{i}e^{\hbar
1050: wz_i}=0\,,
1051: \]
1052: for all $w$. But a finite set of exponential functions is
1053: linearly independent. Hence it follows that a finite set of
1054: coherent states $|\lambda,z\>$ with different $z$ is linearly
1055: independent. The set of linear combinations of finitely many
1056: coherent states is dense in $\Hz$. The coherent states form a
1057: kind of a ``basis", but an overcomplete set. Such a set is called a
1058: {\bfi{frame}}. Frames are widely used in wavelet analysis.
1059:
1060: We can get a so-called {\bfi{tight frame}}\index{frame!tight}
1061: by taking the coherent states
1062: with unit norm. A tight frame in a Hilbert space $\Hz$ is a set of
1063: vectors $\left\{v_\sigma \right\}_{\sigma \in \Sigma}$ in $\Hz$,
1064: where $\sigma$ is some index set, such that any vector $f$ admits a
1065: unique
1066: expansion $V=\sum_\sigma c_\sigma v_\sigma$. To demonstrate that the
1067: coherent states build a tight frame we simplify the notation and put
1068: $\hbar=1$. Next we define
1069: \[
1070: | z \> = | e^{-\shalf |z|^2}, z\> \,.
1071: \]
1072: The vectors $|z\>$ have a unit norm; $\< z|z\>=1$.
1073: We calculate for an element $|f\> = \sum_k f_k|k\>$ the following
1074: \[
1075: |z\> \< z | f\> = \sum_{k,n} e^{-|z|^2} \bar z^n z^k
1076: \frac{f_k}{k!}|n\>\,.
1077: \]
1078: The coefficient of each component $|n\>$ equals
1079: \[
1080: \varphi_n=e^{-|z|^2}f(z)\bar z^n\,,
1081: \]
1082: where $f(z)$ is defined as
1083: \[
1084: f(z)=\sum_n \frac{z^n}{n!}f_n.
1085: \]
1086: From the above discussion we know that $f(z)$ is analytic
1087: everywhere. Thus the vector $\sum_n\varphi|n\>$ is of finite norm;
1088: \[
1089: \sum_n \frac{|\varphi_n|^2}{n!}=\sum_n
1090: e^{-2|z|^2}|f(z)|^2\frac{|z|^{2n}}{n!}=e^{-|z|^2}|f(z)|^2<\infty\,.
1091: \]
1092: Hence $ |z\> \< z | f\>$ represents an element in $\Hz$
1093: and we can integrate each component to get
1094: \lbeq{res.id1}
1095: \frac{1}{\pi}\int_{\Cz} |z\> \< z | f\> d^2 z =\frac{1}{\pi}\int_{\Cz}
1096: |z\> f(z) e^{-\shalf |z|^2}dz = |f\>\,,
1097: \eeq
1098: where the integration measure is $dz= d({\rm Re} z)d({\rm Im} z)$
1099: and where we used
1100: \lbeq{nice.prop}
1101: \int_\Cz \bar z^n z^m e^{-|z|^2} dz =\pi n!\delta_{n,m}\,.
1102: \eeq
1103: In physics literature one writes the result \gzit{res.id1} as
1104: \[
1105: \frac{1}{\pi}\int |z\> \< z | dz =1\,.
1106: \]
1107: In mathematics, such an expression is called a \bfi{resolution of the
1108: identity}\index{resolution of unity}\index{resolution of the identity}.
1109: The fact that the coherent states admit a resolution of
1110: the identity makes them useful. We now wish to show that
1111: the expansion of $|f\>$ in coherent states is unique, thereby proving
1112: that the coherent states make up a tight frame. We use \gzit{res.id1}
1113: to compute the inner product of $|f\>$ with a coherent state $\< w|$
1114: \[
1115: \< w| f\> = \frac{1}{\pi}e^{-\shalf |w|^2}\int_\Cz e^{w\bar z - |z|^2}
1116: f(z) dz\,.
1117: \]
1118: But $f$ is an analytic function, so we first try $f=z^n$. Using
1119: \gzit{nice.prop} we obtain the identity
1120: \[
1121: \frac{1}{\pi}\int_\Cz e^{-|z|^2}e^{w\bar z}z^m dz = w^m\,.
1122: \]
1123: Hence we derive the more general identity for analytic functions
1124: \[
1125: \frac{1}{\pi}\int_\Cz e^{-|z|^2}e^{w\bar z}f(z) dz = f(w)\,,
1126: \]
1127: from which we obtain
1128: \[
1129: f(w)=e^{\shalf |w|^2}\< w| f\> \,.
1130: \]
1131: Hence the expansion of $f$ in coherent states is unique, since if
1132: $f=0$, then all the $\< z | f\>=0$ and the expansion vanishes
1133: identically. Note that the above discussion only works for analytic
1134: functions $f$. If we admit a non-analytic $f$ we get for example
1135: \[
1136: \int_\Cz |z\> \bar z^n e^{-\shalf |z|^2} dz =0\,,
1137: \]
1138: for all $n>0$. There is a relation between coherent states and the
1139: Hilbert space $\Hz_\Cz$ of analytic functions $f:\Cz\to \Cz$ such that
1140: \[
1141: \int_\Cz |f(z)|^2 e^{-|z|^2}dz<\infty\,,
1142: \]
1143: for which the $(z^n)_{n\geq 0}$ form a basis. We refer the interested
1144: reader to \sca{Glauber}\cite{glauber},
1145: \sca{Segal}\cite{segal}, \sca{Bargmann}\cite{bargmann1,bargmann2}.
1146: We just remark
1147: that if $f\in\Hz_\Cz$ and expand $f$ as
1148: \[
1149: f(z)=\sum_{n\geq 0} \frac{f_nz^n}{n!}
1150: \]
1151: then
1152: \beqar
1153: \int_\Cz |f(z)|^2 e^{-|z|^2} dz &=& \int_\Cz \sum_{n,m\geq 0}\frac{\bar
1154: z^nz^m \bar f_n f_m}{n!m!}e^{-|z|^2}dz\nn\\&=& \pi \sum_{n\geq
1155: 0}\frac{|f_n|^2}{n!}\,.\nn
1156: \eeqar
1157: Hence $f$ defines an element in the Fock space $\Hz$. (We have assumed
1158: one can change the order of integration and summation, but that can be
1159: made rigorous, see \sca{Bargmann} \cite{bargmann1} for a readable
1160: explanation.)
1161: The above
1162: discussion on the uniqueness of the expansion of an analytic function
1163: in terms of coherent states was taken from \sca{Glauber} \cite{glauber},
1164: which is a
1165: very readable account on coherent states and the physics and
1166: mathematics behind them.
1167:
1168: We now reinsert the constant $\hbar$ to see some of the behavior of the
1169: coherent states. Remember the formula for $q$
1170: \[
1171: q={2\omega m}^{-\shalf}(a+a^*)\,.
1172: \]
1173: We see easily that $a|\lambda,z\> = \hbar \ol z |\lambda,z\>$.
1174: With a bit more work we see that
1175: \[
1176: \<\lambda,z|a^* |\lambda,z\> = \sum_{k,l} |\lambda|^2
1177: \frac{z^i\ol z^j}{k!l!} \hat\psi^* a^k (a^{*})^{l+1}\hat\psi = z
1178: |\lambda|^2 \sum_{k=0}^{\infty}\frac{|z|^{2k}}{k!}\,,
1179: \]
1180: from which it follows that
1181: \[
1182: \<\lambda,z|a^*|\lambda,z\> = z\<\lambda,z|\lambda,z\>\,.
1183: \]
1184: We thus see that we can associate the real part of $z$ with the
1185: position;
1186: \[
1187: \< \lambda, z| q|\lambda,z \>
1188: = {2\omega m}^{-\shalf}(z+z^*) \<\lambda, z|\lambda,z \> \,.
1189: \]
1190: In a similar fashion the imaginary part of $z$ is related to the
1191: momentum $p$. Let us pause for a while to see what the above
1192: means. The harmonic oscillator has a very symmetric shape. One can
1193: show that the wave functions which are eigenvectors of the number
1194: operator $n$ respect the symmetry $q\to -q$ in
1195: the sense that if $q\to -q$ then they change with a factor $(-1)^n$ (see
1196: any introductory book on quantum mechanics,
1197: e.g., \sca{Griffiths} \cite{griffiths}).
1198: This means that for all wave functions that are
1199: eigenfunctions of the number operator the
1200: average position is precisely in the middle, at $q_0$. Therefore the
1201: momentum has zero expectation value. The coherent states represent
1202: shifted states; their position is not in the middle. Note that we have
1203: used the Heisenberg picture were the states are time-independent.
1204: To see the time-dependent behavior of the coherent states, we consider
1205: the product
1206: \[
1207: e^{-iHt}|\lambda,z\>\,.
1208: \]
1209: If we now shift the lowest energy to zero (that is, we choose
1210: $V_0=\shalf \hbar \omega $), and use \gzit{exp.in-n} and
1211: \[
1212: e^{-i\frac{H}{\hbar} t}|n\>=e^{-i\omega n t}|n\>\,,
1213: \]
1214: we see
1215: \[
1216: e^{-i\frac{H}{\hbar}t}|\lambda,z\>= |\lambda,ze^{-i \omega t}\>\,.
1217: \]
1218: The coherent states thus swing from left to right in the potential
1219: with a frequency $\omega$ and with amplitude $|z|/2m\omega$.
1220:
1221: In order to see the action of the Heisenberg group on the coherent
1222: state, we calculate
1223: \[
1224: (e^{\alpha n}\psi )_k = e^{\alpha k} \psi_k~~\Rightarrow e^{\alpha
1225: n} |\lambda,z\> = |\lambda,e^\alpha z\>\,.
1226: \]
1227: From $a|\lambda,z\> = \hbar \ol z |\lambda,z\>$ it follows that
1228: \[
1229: e^{\alpha a}|\lambda,z\> =e^{\hbar \alpha \ol z}|\lambda,z\> =
1230: |e^{\hbar \alpha \ol z}\lambda,z\>\,.
1231: \]
1232: Further, we have
1233: \[
1234: e^{\ol \alpha a^*} |\lambda,z\> = \sum_{k,l} \frac{\ol \alpha^k
1235: (a^*)^{k}}{k!} \ol \lambda \frac{\ol z^l (a^*)^{l}}{l!}\hat \psi
1236: = \sum_k \ol\lambda \frac{\ol{\alpha+z}^k}{k!}(a^*)^{k}\hat\psi =
1237: |\lambda,z+\alpha\>\,,
1238: \]
1239: and also we have
1240: \[
1241: e^\alpha |\lambda,z\> = |e^\alpha\lambda,z\>\,.
1242: \]
1243: We summarize this and write
1244: \beqar\label{7.group}
1245: e^{\alpha n } |\lambda,z\> &=& |\lambda,e^\alpha z\>\,, ~~~
1246: e^{\alpha a}|\lambda,z\> = |e^{\alpha\hbar z}\lambda,z\> \\
1247: e^{\ol \alpha a^*}|\lambda,z\> &=&
1248: |\lambda,z+\alpha\>\,,~~~e^\alpha |\lambda,z\> =
1249: |e^\alpha\lambda,z\>\,.%
1250: \eeqar%
1251: We can apply arbitrary group elements by taking products.
1252:
1253: \bigskip
1254: The {\bfi{Glauber coherent states}}\index{coherent state!Glauber}
1255: introduced in the present section
1256: for the Heisenberg group, can be generalized. Indeed, the concept of
1257: coherent states extends to a large class of
1258: Lie groups acting on so-called co-adjoint orbits of the group.
1259: In each case, the co-adjoint orbit provides a manifold of labels
1260: for the coherent states on which the group acts, and the
1261: coherent states span in an overcomplete fashion a Hilbert space
1262: on which the group acts as an irreducible highest weight
1263: representation. We shall discuss highest weight representations
1264: in Chapter \ref{c.highest}, but cannot give details for the general
1265: case mentioned here. Instead, we refer the reader to the book by
1266: \sca{Perelomov} \cite{perelomov} and to the extensive survey by
1267: \sca{Zhang} et al. \cite{ZhaFG}.
1268:
1269:
1270:
1271: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1272: \section{Monochromatic beams and coherent states}
1273:
1274: As indicated in Section \ref{s.gamma}
1275: for the case of a beam of monochromatic light, the modes
1276: of the electromagnetic field play the role of the annihilation and
1277: creation operator of the quantum field. Classically the observables
1278: are functions on phase space, hence specified to a certain observable
1279: we have an operator on the configuration space. Namely a physical
1280: configuration is specified by giving the values of the observables,
1281: and to any observable we assign the operator that reads off the value
1282: of that observable. Thus if a configuration of a laser beam, which we
1283: suggestively denote $|\E\>$, is
1284: specified by an electric field $\E(x,y,z,t)$, then the operator
1285: $\mathcal{E}(x,y,z,t)$ reads off the values of the components of the
1286: electric field at the space-time point $(x,y,z,t)$:
1287: \[
1288: \mathcal{E}(x,y,z,t)|\E\> = \E(x,y,z,t)|E\>\,.
1289: \]
1290:
1291: In the transition from classical mechanics to quantum mechanics the
1292: role of the operator $\mathcal{E}(x,y,z,t)$ is played by the operator's
1293: positive frequency part of the electromagnetic field.
1294: The above equation then
1295: tells us that $|\E\>$ is an eigenvalue of the annihilation operator.
1296:
1297:
1298: In a classical system there are many photons and the number of photons
1299: need not be constant, due to absorption and due to the constant photon
1300: production of the laser. Hence, from a micromechanical point of view,
1301: the quantum number $n$ is no longer a good quantum number to assign to
1302: a system resembling a laser. However we know that the electric field is
1303: nearly perfectly constant, and if the beam goes in one direction we
1304: can take the classical expression
1305: \[
1306: \E (x,y,z,t) = a^*\mathbf{u}(\x)^*\e^{i\omega_k t }\,,
1307: \]
1308: for the electric field. Since the classical state of the laser has a
1309: well-defined value of the electric field, the quantum state
1310: $|\E\>$ that
1311: mimics the classical state the most is the one where
1312: \[
1313: a^*\mathbf{u}(\x)^*\e^{i\omega_k t }|\E\>
1314: =\E (x,y,z,t) |\E\> \,.
1315: \]
1316: But then $|\E\>$ is an eigenvector of $a^*$. This similarity
1317: between coherent states and classical states is what motivated Roy
1318: Glauber to investigate coherent states and apply his analysis to the
1319: (quantum and semiclassical) theory of light.
1320:
1321: All this extends with suitable modifications to the other wave
1322: equations described in Sections \ref{s.beta} and \ref{s.alpha}.
1323: In each case, there are families of coherent states describing
1324: nearly classical ray-like behavior, and there are more exotic
1325: quantum states which behave quite unlike any classical system.
1326:
1327: \at{describe these coherent states?}
1328:
1329:
1330:
1331:
1332: