Professional Documents
Culture Documents
H. P. Yuen, Phys. Rev. A 13, 2226 (1976)
H. P. Yuen, Phys. Rev. A 13, 2226 (1976)
2226
TWO- P HO TON COHE RENT STATES 0F T HE RADIATION FIE LD 222'7
the context of quantum optics. These and other ad- frequency v, we have the usual canonical variables
ditional papers will be mentioned in relation to (q, p),
some of our results whenever appropriate. = = i( ,'fi—m)'/'(a —a) . (2.4)
This work is not at all a study of possible math- q —(ff/2&d)'/'(a +a), P —
ematical generalizations of coherent states. In In general, we can introduce two dimensionless
fact, the TCS's can be considered as "coherent self-adjoint operators (a„a,),
states" in a broad sense, as discussed later. How-
ever, they should be distinguished from the coher-
a -=a, +ia, ; a, =a„a,=a, . (2. 5)
'
ent states of Glauber, because in quantum optics We also write
the term coherent state is usually understood in
the Glauber sense which carries the implication
Q=-Q, +iQ„. Q„Q, real, (2.8)
of equal fluctuations in the quadratures a, and a, . and refer to (a„a,) or (n„n, ) as
the quadrature
In Sec. II we establish notations and briefly re- components (quantum or classical) of the radiation
capitulate the basic results in quantum optics used mode. These quadrature components are the nat-
in this paper. For a single degree of freedom, the ural variables in describing a TCS, even when the
definition and properties of two-photon coherent mode has a single frequency. Defining the eigen-
states are developed in Sec. III, together with a states n, ), of a„ o. , ), of a„
l l
p( ', (
f *
=x(v"*n(& 'n/, ' (2. 19)
It follows from (3.2) that
(3.2)
Similar to
the form
m, = (5')-(
In&, the states
i )-'/
' 0,
Im~&
.
can be expressed in
(3.9)
„. &~I&&&, =(——2- )&~I&&& (3.17)
I & I
& The solution of (3.17) is of the form
They are complete orthonormal. Furthermore, b ' a I' —va+'/2
&o. IP), = exp[&z*p/p ——, I&, +f(P*, P)].
acts as the lowering operator for Im ). The quasi- I
IM & &&&&& = M
&&('P, ', z ~ &, ), &&&
I
&'&.
H„(z)t
(3.22)
Other properties associated with o&& carry over I
where H„(z} is the nth Hermite polynomial with
identically to IP), . complex argumentz. Writing
The results of this subsection on IP&z and In&~&
depends only on the relation (3.4), and not on the
more specific (3.5}. The specific characteristics
of (3. 1) are developed below. Certain properties of and using (3.20) and (2. 3), we find
TCS's have been discussed by Stoler' in the con-
text of minimum-uncertainty states. &nlP)g = (n} p} "(v/2p)""H„[P(2 pv) ")
"exp[- z IP I'+(v*/2I )P']. (3.23)
B. Wave functions
For p = 1, v = 0, the asymptotic forms of H„(z) for
The general coherent-state wave function &a I P&~ various";;rg;- all lead to . (2. 3) as they must.
of a TCS is first determined as follows: From Equation (3.23) shows that the counting statistics
(3.1), (3. 10), and (2. 13) we obtain of a TCS is far from the Poisson (2. 3).
HORACE P. YUEN
(3.24}
Note the Gaussian form of (3.24) and (3.25). Var- &a) =-,(PIaIP), =/ *P vP-'= P -P, +=-2P„ (3.28)
ious reduction of (3.24) and (3.25) to the previous
&a"a& = I p I'+ v I'
formulas can be easily observed. The relative I
phases of (3.20), (3.21), (3.24), and (3.25) are &ga2& ~nv &~22 &m (3.30)
mutually consistent, as well as being consistent '
with the above wave functions for the coherent
&ha', &
= —,
I
i( v I', (3.31a)
states. '
&aa,') =-, Ii(+vI', (3.31b)
From
&na(aa2& =—
((a, —$, )(a2 —P, )) = ,'i(/(—'v —v*/2+1),
&nIp&, = P&nI», &&», Ip&, (3.32a)
for the quasiphoton number states I»2& we obtain &na2na, & =-,'2(/(*v —v*}2 —1}. (3.32b)
(n Im, ) = (I))(/2, v; m, n), Equation (3.32) is, of course, consistent with the
=q, (i(, ;
m (n, m odd, n odd,
(3.42)
This density function is illustrated in Fig. 1.
Equation (3.3S) is conveniently expressed in the
matrix form
registered for a coherent state at u, in a counting is investigated in some detail in this section, which
gives all of the U~.
experiment employing detectors responding only
to fields at (d, . In this connection one may also
A. Positive definite quadratic Hamiitonian
note that for a fixed mode the frequency is also
fixed by the dispersion relation. Changing the Consider the general quadratic Hamiltonian
frequency of a fixed mode by dielectric modulation =g(f, ata+ f,*a'+f, at'+f 3 a+ f, at), (4.2)
would not yield a minimum-uncertainty state at f re- Ho
quency ( ~, from a coherent state at frequency ~, . A where the c numbers f, may be time dependent.
coherent state at a different frequency (d, would be From Hermiticity f, has to be real. For a mode of
obtained instead. frequency co, we have f, = ro, and Sf,a~a represents
Alternatively, one may regard TCS's and ordi- the free radiation energy of that mode. The f, and
nary coherent states as (mathematically) equiva- f, terms then represent interaction energies. In
lent, since they are related through a unitary particular, the f, terms describe a two-photon
transformation or some kind of sealing. The ques- mechanism, whereas the f, terms describe the
tion of scaling was already dealt with in Sec. IIIC usual one-photon or linear driving mechanism.
concerning the noise in quadrature components. If a physical system with Hamiltonian given by
It is true that two-photon coherent states are co- (4.2) is stable, Ho should not have a spectrum
herent states in a broad sense: they are the co- ranging through the entire continuum from -~ to
herent states with respect to b(vs 0) instead of a. ~ in order for a ground state to be defined, i. e.„ .
However, this abstract unitary equivalence means the spectrum of H+ should be lower bounded. By a
little in describing actual experimental results. shift of scale one may require that the Hamilton-
Thus (n, a), = (n o.) but (n o), has little resem-
~ ~ ~
ian be positive definite. 3 If this condition is not
blance to (n n). In general, for an arbitrary uni-
~
satisfied, (4.2) cannot be properly used to de-
2234 HORACE P. YUEN 13
scribe a physical system of interest in our present Uz"'(a*, a; t, t, ) = exp[A(t) +B(t) a'+ C(t) a*'
context. It has to be extended to include a more
complete description of other relevant i'actors of + D(t) I
a I'+&(t) a+ F(t)a~],
importance.
Under the condition (4.8}
where
(4.3)
dA = —
the Hamiltonian (4.2) can be changed, similar to
" 2 (2 f 2 C+ f F +f,*F),
2
2
(4.9a)
the Bogoliubov transformation, to the following
form: dB =- i f,*(D+1)', (4.9b)
ffo =f. b'5+f0(I |If' —vf. I'+ vI'& (4 4&
b —
= pa+ vat+ pf,* —vf„
I
linear canonical transformation from a to b can al- Using (3.20), we find by comparison that
ways produce f, terms in b from a free Hamilton-
ian h f, a~a. In fact, it is easily checked that an f,
U, (t, t.) p = p+ &(t); l (t), (t))
I &
"', (4.12)
I
"
term results if and only if the v term mixing bt is 4(t) e(t)+ i[p &(t) —p&*(t)]
= (4.13}
present in the transformation. The transformed for a, real-valued (P, f) independent function e(t)
Hamiltonian contains a constant term v I', so that I and
the eigenstates of b have this added excitation ener-
gy compared to the true ground state 0}. Such a A(t) =--'h u(t)+ v*(t)l'(t)/2t (t)
I
while
t;(0) =0, (4,19) u)(~) =
f 'u(r) uu, (4.28)
is an additional constraint on the pair t«(t), v(t) ob- Therefore, under an arbitrary Uq(t, t, ) a. TCS
tainable from an H. Thus not every pair of func- wH. l ~emain a TCS for all time, but with P, p, , and
tions t«, (t), v(t) obeying (3.2) can occur in a Uq(t, to). v time va. rying. By solving p(t), v(t), and g(t) in
Only those obeying (4.33) are permitted in the form terms of p, , and vo and t«(t, to), v(t, t, ), and 5(t, t, ),
(4. 12). However, this does not impose any con- (4.39)-(4.42) and (4.27) show explicitly that any
straint on the mean- square fluctuation behavior ob- iP;t«, v) can be obtained from any given iP, ; ««„v, )
tainable in (3.61), since only it«i, ivy are involved by a Uq(t, to).
here. Moreover, (4.27) shows that any particular It also follows from (4.41) that if f, is turned off
pair of values (p, , v) obeying (3.2) can be obtained at time t, then
at a specific t with a proper choice of f„even un-
t«(t) = t«(t )e«e&«' f &
v(t) = v(t )e«e« ~ «t& (4 43)
der (4.3} with f, fixed.
Any desirable pair of functions I«, (t), v(t) obeying Thus apart from the optical oscillation the values
(4.33) and (3.2) can be generated by an f~ through p, and v are invariant in the absence of two-photon
(4.15). This corresponds to the solution of the interaction in H. Furthermore, such invariance
Riccati equation (4.23) for various For .example, f is obtained when and only when
we can assume (f, = &0, t, = 0)
f2=0, (4.44)
p(t) =e'~'" cosh(yt), v(t) =e«~'" sinh(yt), (4.34)
i.e., v(t}=0.
In general, p(t} and v(t} are deter-
and determine the phase P(t) so that (4.33) is sat- mined only by f, (t) independently of f, (t), as seen
isfied. Equation(4. 34) has the interesting charac- from (4.15). This is in accordance with the fact
teristic that that the new quantum noise behavior is entirely a
'"' («'(t)) =-'e'"' consequence of f, 40. Such correlation between f,
(~a'(t)) =-'e (4.35)
and v is also observed in Sec. IVA.
without any shift of phase to a'. Substitution of On the other hand, the added value f(t) depends
(4.34) into (4.33) shows that &t&(t) obeys the differen- on bothf, andf, but is always zero forf, =0,
tial equation
Uq(t, t, ) iP) = iP)„when f, =0. (4.45)
dQ 1 Therefore aEE of the Ui's are generated by H& with
dt cosh'(yt)+sinh '
(yt)
f, =0. The addition of f, merely introduces a con-
so that stant shift to b, as in (4. 5). The states (4. 39) are
eigenstates of
&t&(t) = («&/y) tan [tanh(yt)]. (4.36) b (t) = u(t) a(t, }+v(t) a'(t, ), (4.46)
The corresponding f, (t) is given by
where a(t, ) is the Schrodinger-picture fixed photon
f, = 2 tanh(2yt) —iy/2&». (4.37) annihilation operator. We have
Any arbitrary state i
&|&, ) will evolve into b(t) = Uq(t, to) a(to) Uqt(t& to)+ f(t) = Ui aUii. (4.47)
Equation (4.47} also shows
t«(t) a„(t)+ v(t) a„(.t) =a(t&&)+ g(t) (4.48)
(4.38) for the Heisenberg operator
under Uq. If ) = P, ; p, „v, ), (4.38) can be read-
i &C&, i
where , v, g, and 8 are still the solution to to the standard quantum theory of parametric in-
p,
(4.15)-(4.20). The state Utq(t, f,) ~P), is obtained teraction. "'"
By methods widely used in nonlinear
from evolution of the system backwards in time optics, ' classical equations of motion for the field
under the same f&(t) in Hq, and the operator b(t) is and the active atomic variables can be obtained in
similarly the ae(t) moving ba, ckwards, apart from which the virtual intermediate states are accounted
a constant.
Even though no solution of Uq(f, to) in quadrature
for through P. Examples of such treatment for two-
photon systems can be found in many places. "
can be obtained in general, its structure and be- 'These classical equations of motion with a quan-
havior have already been fully determined. These tum-mechanically computed susceptibility can be
behaviors are important in the applications of directly quantized. The effective quantum Hamil-
TCS's. tonian follows from the resulting quantum equa-
tions of motion.
V. STATE OF TWO-PHOTON LASING RADIATION In addition to absorption, two-photon emission
which provides a two-photon lasing mechanism is
We suggest that the output radiation of an ideal
also described by (5.1). In considering the condi-
monochromatic two-photon laser is in a TCS. In
tions for stimulated two-photon emission in either
contrast, the output of an ideal one-photon laser is the standing-wave or traveling-wave configuration
in an ordinary coherent state. Thus TCS's are the
with loss included, it is easily found that stimulated
characteristic states of two-photon lasing pro-
emission cannot be set up from spontaneous emis-
cesses (two photons from the same mode}.
A. Two-photon laser
sion alone. A sufficiently strong field at the lasing
frequency ~ has to be present initially. There- ""
fore we consider the two-photon lasing configura-
Two photons of the same frequency (d from the tions depicted in Fig. 2. It is not the purpose of
same radiation mode can be absorbed in a single this article to present the semiclassical theory of
atomic transition between two levels 2 and 1 via two-photon laser operation, or to discuss the op-
an intermediate state, i.e. , a transition second or- erating consideration in a realistic material sys-
der in p. A. Under the usual dipole approximation tem. But the following facts may be observed:
the parities of levels 2 and 1 have to be identical. In the traveling-wave configuration, Fig. 2(a),
First-order dipole transitions between the two lev- the gain is obtained from a large external pump
els are forbidden in this circumstance. On the field which may have many frequencies but which
other hand, higher-order multipole contributions is still in a single temporal mode. Neglecting var-
to second-order p A transitions, which may con-
~
ious loss mechanisms, this lasing process can be
nect states of different parity, can be significant. 4' described by the interaction Hamiltonian (5.1) with
First-order transitions between states of opposite an initial radiation state p, having a nonzero
parity are also possible via the A' term in the amount of power, trp, a a WO. If a laser is used for
radiation interaction Hamiltonian. ~~4 This is pos- the external pump, p, = ~n)(o. ~, but intense ther-
sible because in the multipole expansion of the mal light can also be used. The radiation state p,
vector potential A(r}, higher-order multipole terms as well as other dynamical variables, are spatial-
introduce atomic coordinates into the A' Hamilton- ly dependent in this case. In the standing-wave
ian. Two-photon absorption spectroscopy is an
important experimental technique by now, and " case, Fig. 2(b), only a single standing-wave mode
is under consideration. The gain is initially ob-
many discussions of two-photon absorption pro- tained from the usual one-photon lasing mechanism,
cesses can be found in the literature.
The radiation-matter (to be referred to as atoms) contained in the same cavity. "
the host atoms for both lasing transitions being
The Hamiltonian is
interaction responsible for the two-photon transi- now
tion can be expressed in the form
H ~ = 5( p Ma~'+ p*Mt a') (5.1) Ep Ep
transition through an intermediate state has been (a ) E2- E 25~ (b) E2 El = 2%~
suppressed through a nonlinear susceptibility, or
I
—
Ep EI
equivalently through the coupling coefficient p,
(5.1) is only an effective Hamiltonian. In this case FIG. 2. Two possible configurations for stimulated
it can be derived by following a procedure similar two-photon emission.
2238 HORACE P. YUEN
H = a(&uata+p2M, a++p,*M2~a2+p, M, a~+p,*M [a), through the Fokker-Planck equations for p "(n*, a).
(5 2) "
In particular, all of the characteristic functions
are still in Gaussian form, and for y44r,
with a vacuum initial state; the subscripts 2 and 1 1 y
for p and M refer to the two-photon and one-photon
(na~' (t})=— + e '"'4"
4 y+4z y+4r
lasing material. Population inversion is required
Yn
in both configurations. Other possible configura- (1 e-(y~r) t) (5.4)
2(y+ 4r)
tions for two-photon lasers can be developed, but
these two seem to be the most important. with a' =ac'"'. It can be readily shown that (&a&'(f)&
Spontaneous two-photon emission has been ex-
perimentally observed, "
but not stimulated two-
&0 and that the uncertainty principle is obeyed for
all t. Note that even when n =0, the state is no
photon emission. Two-photon stimulated emission
was first suggested for giant pulse generation. " longer pure in the presence of y. Asymptotically
we have
When such a two-photon laser operates far above
threshold with amplitude and phase stability, the (&aP (t) ) yn-/8r, (&a212(t) ) - ~; t - ~, 2r» y.
fluctuation in M should be relatively small, so that (5 5)
M is approximately a c number. By ignoring the
The condition 2r»y can be satisfied together with
quantum nature of M, (5.2) or (5.3) becomes an Ho
of the form (4.2). If the dissipative mechanisms
the stability condition (4.3) for the usual values of
are also neglected, (4.12) shows that a TSC ~P&, Y p
n represents the thermal noise contribution from terrnined only by explicitly working out the detailed
the reservoir that couples to the radiation field. quantum theory of two-photon lasers. The quantum
This quantum description of loss has been exten-
sively discussed in the literature. ""
For an ini-
theory of one-photon lasers has been extensively
developed, "~'
but the two-photon laser requires
tial coherent state, (5.3) can be solved exactly. more careful approximations because of the small
TWO-PHOTON COHERENT STATES OF THE RADIATION FIE LD 2239
noise in one quadrature component and the more itial states of the atoms are identical, but they are
delicate equations of motion involved. Hopefully, different in the two-photon case. The question of
the TCS picture can be checked experimentally, as quantum statistics is similar in both cases. The
discussed in Sec. VI, by practically realizing a pump field of Fig. 3 should be treated as a quan-
two-photon laser. In any case, P), serves as the
~
tum degree of freedom. A fully quantum treatment
basic description to be averaged further by a of degenerate parametric oscillators can be equal-
classical density function in the presence of other ly or more complex than a fully quantum treatment
random interference. of two-photon lasers, depending on whether the
Recently, the quantum theory of multiphoton atomic variables are included in the parametric
lasers, including a quantum treatment of the atoms case and whether the important intermediate state
counts in time T is merely a scaling of (6.1), fluence of nonideal detectors and noise on these
photocount statistics should be worked out in de-
(-1)" 8 "Q(sn, Tx} tail for actual comparison with experiments.
(6.4)
From (3.23) the function Q(X} for a TCS can be VII. APPLICATIONS OF TWO-PHOTON
evaluated in closed form using the summation for-
mula for Hermite polynomials, " COHERENT STATES
(3.29), and
A. Ideal linear amplifier
&a""&=("a&'+2 P I'(4I v I'+» v I'
I I
clearly possible for small v and large p, with non- To be definite let v &0. The quantum noise (nP,'&
vanishing P. Recall that g, &1 is a characteristic decreases from (nP, = —,' to (nP,'& = —,'(p —v)', while
&
FIG. 3. Phase-matched degenerate parametric amp- the signal-to-noise ratio cannot be improved by
lification with pump fiel. d at frequency 2~. amplification. The introduction of f, into U does
TWO- PHOTON COHERENT STATES OF THE RADIATION FIE LD 2241
not increase the signal energy, i.e., the part of a decrease of (&a', ) in P) can be obtained only at
~
which is proportional to P', . The advantage of am- the expense of spending a portion of available en-
plification is practical, similar to the classical ergy S in the form of added quantum noise energy
case. Raising the signal level makes possible ~
v ~', from (3.35). Thus (&a', ) -0
requires S -~.
.
easier measurement of P, Since the two-photon This consideration is not important in the context
amplification process (4.45) operates above thres- of local oscillators at a receiver where large
hold, it can be superior to one-photon amplifiers enough amount of power is usually available to
operating below threshold in its possibilities of make (&a', ) sufficiently small. A four-orders-of-
yielding higher gain and lower noise. If amplifica- magnitude reduction from the quantum noise —, of
'
tion for both quadrature components is desired, n, (~ v ~' =10') for a GHz bandwidth optical signal
the input beam can be f irst split and two diff erent at ~/2m=10" Hz requires only a v~' correspond-
ing to -1 pW. This will usually bring it down to the
~
'.
state local mode (&a2») = —, Further discussion P, =P, =O, (7.7)
can be found in Ref. 1. Note that this advantage of
' cannot be obtained if a one-photon-laser we have
(&a2») « —,
local oscillator is used to produce a coherent- (S/N), iq) =4(S'+S). (7.8)
state receiver mode as and a Uo(f, t, ) subsequently
Compared to the signal-to-noise ratio
applied, because of the corresponding signal at-
tenuation. An absolutely small (4a») is needed in (S/N), i8) = 4$ (7.9)
this scheme.
in a coherent state, we see that the available "sig-
If an arbitrary amount of power is available, it
nal energy" is effectively increased from S to S'
is possible to have a state with (&a,')
ever, for a fixed total radiation energy
-0. How-
+S. This is actually a reduction of quantum noise. "
Among other things, this increase in (S/N), leads
trpa a —S, (7.5) to a higher information capacity even when the
2242 HORACE P. YUEN
*Work supported by the National Aeronautics and Space Lett. 10, 84 (1963).
Administration under Grant No. NGL 22-009-013 and 5Itis actually a power spectral density,i.e. , power per
by the Office of Naval Research under Contract No. unit frequency.
NR042-342. 6H. P. Yuen, Phys. Lett. 51A, 1 (1975).
H. P. Yuen, in Proceedings of the 1975 Conference on N. Bloembergen, Nonlinear Optics (Benjamin, New
Information Sciences and Systems (Johns Hopkins U. P. , York, 1965).
Baltimore, 1975), p. 171. E. H. Kennard, Z. Phys. 44, 326 (1927).
2Coherent states are excluded when we refer to a mini- ~D. Stoler, Phys. Rev. D 1, 3217 (1970); 4, 1925 (1971).
mum-uncertainty state. For a mode of frequency ~, E. Y. C. Lu, Nuovo Cimento Lett. 2, 1241 (1971); 3,
coherent states are minimum-uncertainty states with 585 (1972).
~'g q') = g, p ') =-'h~. Minimum-uncertainty wave D. Stoler, Phys. Rev. Lett. 23, 1397 (1974).
packets are treated in almost every textbook in quan- B. R. Mollow, Phys. Rev. 162, 1256 (1967).
tum mechanics; see also Ref. 3. The distinction bet- ~3J. Perina, V. Perinova, and R. Horak, Czech. J. Phys.
ween minimum-uncertainty radiation states and cohe- B 23, 975 (1973); J. Perina, V. Perinova, and L. Mista
rent states is clarified in Sec. III E. ibid. 24, 482 (1974).
3W. H. Louisell, quantum Statistical Properties of Rad- D. W. Robinson, Commun. Math. Phys. 1, 159 (1965).
iation (Wiley, New York, 1973). '5J. Manuceau and A. Verbeure, Commun. Math. Phys.
R. J. Glauber, Phys. Rev. 131, 2766 (1963); Phys. Rev. 9, 293 (1968).
~~F. A. Berezin, The Method of Second Quantization 39N. N. Bogoliubov, J.
Phys. USSR 11, 23 (1947).
(Academic, New York, 1966). 40H. T. Davis, Introduction to Non/inear Differential
and
~~R. Haag, in Lectures in Theoretica/ Physics, edited by Integra/ Equations {Dover, New York, 1962).
%. E. Brittin and A. 0. Barut (University of Colox'ado 4l%. T. Reid, Riccati Differentia/ Equations (Academic,
Press, Denver, 1960), Vol, . XX, p. 353. New York, 1972).
+R. J. Glauber, in Quantum Optics and E/ectronics, 42B. Guccione and J. Van Kranendonk, Phys. Rev. Lett.
edited by C. M. De%itt et a/. {Gordon and Breach, New 14, 583 0.965).
York, 1965), p. 63. M. Iannuzzi and E. Pol, aeco, Phys. Rev. Lett. 13, 371
~9J. R. Klauder and E. C. G. Sudarshan, I"undamenta/s (1964).
of Quantum Optics (Benjamin, New York, 1968). 44R. Guecione-Gush, H. P. Gush, and J. van Kranendonk,
2
V, Bargmann, Commun. Pure Appl. Math. 14, 187 Can. J. Phys. 45, 2513 (1967).
(1961). +D. Prichard, J. Apt. and T. %. Dueas, Phys. Rev. Lett.
~'C. L. Mehta and E. C. G. Sudarshan, Phys. Rev. 138, 32, 641 (1974); M. D. Levenson and N. Bloembergen,
B274 {1965). Phys. Rev. Lett. 32, 645 (1974); P. F. Liao and J. E.
+J. R. Klauder, J. McKenna, and D. G. Currie, J. Math. Bjorkholm, Phys. Bev. Lett. 34, 1 (1975).
Phys. 6, 733 (1965). 46W. H. Louisell, A. Yariv, and A. E. Siegman, Phys.
23J. R. Klauder, Phys. Rev. Lett. 16, 534 (1966). Rev. 124, 1646 (1961),
24M. M. Miller and E. A. Mishkin, Phys. Bev. 164, 1610 47B. R. Mollow and R. J. Glauber, Phys. Rev. 160, 1076
(1967). 0.967); 160, 1097 0.967).
+See, e.g. , E. M. Bel.enov and I. A. Poluektov, Zh. Eksp.
~5E. C. G. Sudarshan, Phys. Rev. Lett. 10, 277 (1963).
~~B. J. Glaubex, in Physics of Quantum Electronics, Teox. Fiz. 56, 1407 (1969) [Sov. Phys. JETP 29, 754 —
edited by P. L. Kelly et al . (MeGraw-Hill, New York, {1969)];L. E. Estes, L. M. Narducci, and B. Shammas,
1966), p. 788. Nuovo Cimento Lett. 1, 775 {1971); M. Takatsuji,
2~R. J. Glaubex, in Laser Handbook, edited by F. T. Phys. Rev. A 4, 808 (1971).
Arecchi and E. O. Sehulz-DuBois (Elsevier, New York, 4~P. P, Sorokin and Braslau, IBM J. Res. Dev, 8, 117
¹
postul. ate of quantum field theory; see, e.g. , R. Haag of generating ( p)~ in terms of the constraint (7. 5).
and B. Schror, J. Math. Phys. 3, 248 (1962).