Mapping Wigner distribution functions into semiclassical distribution functions
G. W. Bund1,*and M. C. Tijero2,† 1
Instituto de Fı´sica Teo´rica, Universidade Estadual Paulista, Rua Pamplona, 145, 01405-900 Sa˜o Paulo, SP, Brazil 2Pontifı´cia Universidade Cato´lica de Sa˜o Paulo, Rua Marqueˆs de Paranagua´, 111, 01303-000 Sa˜o Paulo, SP, Brazil
共Received 1 April 1999; published 19 April 2000兲
A mapping that relates the Wigner phase-space distribution function of a given stationary quantum mechani-cal wave function, a solution of the Schro¨dinger equation, to a specific solution of the Liouville equation, both subject to the same potential, is studied. By making this mapping, bound states are described by semiclassical distribution functions still depending on Planck’s constant, whereas for elastic scattering of a particle by a potential they do not depend on it, the classical limit being reached in this case. Following this method, the mapped distributions of a particle bound in the Po¨schl-Teller potential and also in a modified oscillator potential are obtained.
PACS number共s兲: 03.65.Sq I. INTRODUCTION
The phase-space formulation of quantum mechanics be-gan with a paper of Wigner 关1兴. Because the phase-space formulation offers a framework in which quantum problems can be treated using the classical language as much as it is allowed, it has been applied to many areas of physics. Among these areas we mention statistical physics关2兴, quan-tum optics关3兴, collision theory 关4,5兴, nuclear physics 关6兴, and nonlinear physics 关7,8兴.
The phase-space distribution functions are the main tool of the phase-space formulation of quantum mechanics and among them the Wigner distribution function 共WDF兲 has been used with success, for instance, in the description of atom-molecule collision processes 关9,10兴, in nuclear physics 关6,11兴, and also for studying the classical limit of quantum mechanics关12,13兴. It is with this last topic that this work is concerned. In Sec. II we summarize Ref. 关12兴, where it is shown how to perform the mapping that relates the WDF of a given stationary quantum mechanical wave function, a so-lution of the Schro¨dinger equation, to a soso-lution of the Liou-ville equation, both subject to the same potential. Two ex-amples have already been considered in Ref. 关12兴: the infinite square well and the potential step. In Sec. III of this work we apply this same mapping to a particle bound in the Po¨schl-Teller potential关14兴 and in Sec. IV to a particle in the ground state of a modified oscillator potential. The potentials used in Sec. III and Sec. IV are smooth and more appropriate for classical or semiclassical approximations than those of Ref. 关12兴.
II. MAPPING THE WDF TO A SEMICLASSICAL DISTRIBUTION FUNCTION
The classical limit of quantum mechanics may be accom-plished in two steps. First the mapping of the WDF to a semiclassical phase-space distribution function共SDF兲 is ob-tained, and in a second step the limitប⫽0 is taken since the
SDF may still depend on ប. We consider only the first of those two steps 共Ref. 关12兴; the abbreviation SDF here corre-sponds to CDF in that reference兲.
The WDF(q, p,t) satisfies the equation关5兴
t ⫹ p m q⫹
冕
K共q,p⫺p⬘
兲共q,p⬘
,t兲dp⬘
⫽0, 共1兲where (q, p) is a point in phase space. The kernel K is given by K共q,p⫺p
⬘
兲⫽ i ប冕
dv 2បe (i/ប)(p⫺p⬘)v冋
V冉
q⫺v 2冊
⫺V冉
q⫹v 2冊册
, 共2兲V(q) being the potential. Equation 共2兲 becomes in the
clas-sical limit关5兴 Kc共q,p⫺p
⬘
兲⫽⫺ i ប冕
dv 2បe (i/ប)(p⫺p⬘)vvV q ⫽⫺V q p␦共p⫺p⬘
兲, 共3兲and Eq.共1兲 goes over into the Liouville equation
c t ⫹ p m c q ⫺ V q c p ⫽0. 共4兲
We are looking for a particular solution c of Eq. 共4兲 which may be considered the correct semiclassical approxi-mation of a given, a solution of Eq.共1兲. In order to obtain it we write the integral equation
共q,p,t兲⫽c共q,p,t兲⫺
冕
⫺⬁ ⬁ dt⬘
冕
dq⬘
d p⬘
⫻Gc共q,p;q⬘
, p⬘
,t⫺t⬘
兲冕
d p⬙
关K共q⬘
, p⬘
⫺p⬙
兲 ⫺Kc共q⬘
, p⬘
⫺p⬙
兲兴共q⬘
, p⬙
,t⬘
兲, 共5兲 *Electronic address: [email protected]relating Eqs. 共1兲 and 共4兲, where Gc(q, p;q
⬘
, p⬘
,t⫺t⬘
) is the retarded Green’s function, which satisfies the inhomoge-neous Liouville equationGc t ⫹ p m Gc q ⫺ V q Gc p ⫽␦共q⫺q
⬘
兲␦共p⫺p⬘
兲␦共t⫺t⬘
兲. 共6兲 A WDF that satisfies Eq. 共1兲 when inserted into Eq. 共5兲 gives a functioncthat satisfies Eq.共4兲 as can be verified by inserting Eq.共5兲 directly into Eq. 共4兲. So Eq. 共5兲 generates a SDF from a known WDF. This is the starting point for ob-tainingc. Using Eq.共1兲 and Eq. 共3兲, Eq. 共5兲 may be writtenc共q,p,t兲⫽共q,p,t兲⫹
冕
⫺⬁ ⬁ dt⬘
冕
dq⬘
d p⬘
⫻Gc共q,p;q⬘
, p⬘
,t⫺t⬘
兲冉
⫺ t⬘
⫺ p⬘
m q⬘
⫹V q⬘
p⬘
冊
共q⬘
, p⬘
,t⬘
兲. 共7兲We stress that Eq.共7兲 is equivalent to Eq. 共5兲 if satisfies Eq. 共1兲.
The WDF generated by a given wave function⌿(q,t), a solution of the Schro¨dinger equation corresponding to the potential V(q), is given by关1兴
共q,p,t兲⫽
冕
2dvបei pv/ប⌿冉
q⫺v2,t
冊
⌿*冉
q⫹v
2,t
冊
. 共8兲 In Eq. 共8兲, when ⌿ is stationary, does not depend on the time. In the particular case in which V(q) is quadratic in q, Eq. 共7兲 reduces to ⫽c, as in this case Eq. 共1兲 becomes identical to the Liouville equation.The retarded Green’s function satisfying Eq.共6兲 is
Gc共q,p;q
⬘
, p⬘
,,兲⫽e⫺␦„Q共q,p,⫺兲⫺q⬘
…⫻␦„P共q,p,⫺兲⫺p
⬘
…⫹共兲, 共9兲where the convergence factor exp(⫺) (→0), was intro-duced. Here „Q(q,p,t),P(q,p,t)… represents the classical trajectory in phase space of a particle subject to the potential
V(q), which at time t⫽0 occupies the phase-space point
(q, p), and ⫹ is the step function.
Using the classical equations of motion P/m
⫽Q(q, p,t
⬘
⫺t)/t⬘
and⫺V/Q⫽P(q, p,t⬘
⫺t)/t⬘
we get冋冉
t⬘
⫹ p⬘
m q⬘
⫺ V q⬘
p⬘
冊
⫻共q⬘
, p⬘
,t⬘
兲册
q⬘⫽Q(q,p,t⬘⫺t),p⬘⫽P(q,p,t⬘⫺t) ⫽ t⬘
„Q共q,p,t⬘
⫺t兲,P共q,p,t⬘
⫺t兲,t⬘
… 共10兲and Eq.共7兲 becomes 关12兴
c共q,p,t兲⫽共q,p,t兲⫺
冕
⫺⬁ t dt⬘
e⫺(t⫺t⬘) ⫻ t⬘
„Q共q,p,t⬘
⫺t兲,P共q,p,t⬘
⫺t,t⬘
兲…. 共11兲 When may be set equal to zero from the beginning, we get from Eq.共11兲c共q,p,t兲⫽ lim t⬘→⫺⬁
„Q共q,p,t
⬘
兲,P共q,p,t⬘
兲,t⬘
…. 共12兲 Thus, if the limit in Eq.共12兲 does exist the value ofcat the phase-space point (q, p) is the value of at the initial (t⫽ ⫺⬁) phase-space point of the classical trajectory that at timet⫽0 reaches the point (q,p). Assuming that does not de-pend on the time, Eq. 共11兲 gives, after integrating by parts, the mapping equation
c共q,p兲⫽ lim →0⫹
冕
⫺⬁ 0 de„Q共q,p,兲,P共q,p,兲…. 共13兲 For systems with one degree of freedom at phase-space points where the classical energy E(q, p) is negative, Eq. 共12兲 cannot be applied and then Eq. 共13兲 has to be used. Making the Fourier decomposition„Q共q,p,兲,P共q,p,兲…⫽
兺
n⫽⫺⬁⬁
Rn共q,p兲ein(q,p), 共14兲 where the period associated with the trajectory is T(q, p) ⫽2/(q, p), and inserting Eq. 共14兲 into Eq. 共13兲, we get
c(q, p)⫽R0( p,q), which is equal to c共q,p兲⫽ 1 T共p,q兲
冕
0 T(q, p) d„Q共q,p,兲,P共q,p,兲…, 共15兲 so the only nonvanishing term in the ⫽0⫹ limit corre-sponds to n⫽0 in Eq. 共14兲.For points (q, p) on the same closed classical path, Eq. 共15兲 gives the same value of c, because this expression is invariant under time translations of the integrand, and the trajectories „Q(q,p,t),P(q,p,t)… associated with these points are related to each other by making time translations. This invariance shows that c depends only on the energy
E(q, p) associated with the trajectory and it can be extended
to open classical paths by making use of Eq.共13兲.
When the WDF corresponds to a bound state in a poten-tial that vanishes at infinity, Eq.共12兲 may be applied directly for points with positive energy E(q, p), giving the result
c(q, p)⫽0. It can be shown also that the phase-space aver-aged classical energy for the SDF coincides with the quan-tum energy of the bound state. The mapped SDF in this case may be viewed as a stationary distribution function of clas-sical particles trapped by the potential.
In order to see how this theory works two examples were considered.
A. The infinite square well The ground state of the infinite square well
V共q兲⫽
再
0, 兩q兩⭐d ⬁, 兩q兩⬎d 共16兲 is given by ⌿共q兲⫽冑
1 dcos q 2d, 兩q兩⭐d. 共17兲The WDF corresponding to this bound state is关4兴
共Q,P兲⫽21បd
再
2 Pប冋
sin冉
2 P ប 共d⫺Q兲⫹ dQ冊
⫹sin冉
2 Pប 共d⫺Q兲⫺ dQ冊册
⫹冉
2 Pប ⫹d冊
⫺1sin冋冉
2 P ប ⫹ d冊
共d⫺Q兲册
⫹冉
2 Pប ⫺ d冊
⫺1 sin冋冉
2 P ប ⫺ d冊
共d⫺Q兲册冎
共18兲 for 0⬍Q⬍d and 共Q,P兲⫽共⫺Q,P兲 共19兲 for ⫺d⬍Q⬍0.Making a change of variables in Eq. 共15兲, we obtain for the SDF c共q,p兲⫽ 1 2d
冕
⫺d d 共q⬘
, p兲dq⬘
⫽ 4ប cos2共pd/ប兲 关共/2兲2⫺共dp/ប兲2兴2. 共20兲 We see that as the SDF c depends only on the classical energy, which in this case is p2/2m, it cannot depend on the coordinate q. In the limit ប→0 Eq. 共20兲 gives c(q, p)→(1/2d)⫺1␦( p), which describes a uniform distribution of
particles at rest inside a box.
B. The potential step The wave function for the potential step
V共q兲⫽
再
0, q⬍0
V0⬎0, q⭓0 共21兲
for incident energies E⫽k2/2m⬍V0 is given by
共q兲⫽
再
2A cos共kq/ប⫺␣/2兲, q⬍02A cos共␣/2兲e⫺q/ប, q⬎0, 共22兲
where ⫽(2mV0⫺k2)1/2 and ei␣⫽(ik⫹)/(ik⫺) gives the phase ␣. The WDF is given by 关4兴
共Q,P兲⫽4兩A兩 2
冋
C(⫺)共P,k兲cos冉
2共P⫺k兲 Q ប冊
⫹C(⫹)共P,k兲cos冉
2共P⫹k兲Q ប冊
⫹S(⫺)共P,k兲sin冉
2共P⫺k兲Q ប冊
⫹S(⫹)共P,k兲sin冉
2共P⫹k兲Q ប冊册
, 共23兲 for Q⬍0, where C⫿共P,k兲⫽k关共⫿2P⫹k兲2⫹2兴⫺1共⫾2P兲⫺1, S⫿共P,k兲⫽k共⫾2Pk⫺k2⫹2兲/关共⫿2P⫹k兲2⫹2兴 ⫻4P共k⫿P兲. 共24兲 For Q⬎0 the WDF is 共Q,P兲⫽4兩A兩 2 e⫺2Q冋
C0共P,k兲cos冉
2 PQ ប冊
⫹S0共P,k兲sin冉
2 PQប冊册
, 共25兲 where C0共P,k兲⫽4k2关共2P⫹k兲2⫹2兴⫺1关共2P⫺k兲2⫹2兴⫺1, S0共P,k兲⫽ k 2共k2⫹2⫺4P2兲 关共2P⫹k兲2⫹2兴关共2P⫺k兲2⫹2兴P. 共26兲 The SDF that corresponds to the WDF given in Eqs.共23兲 and共25兲 is 关12兴c共q,p兲⫽
再
兩A兩2关␦共p⫺k兲⫹␦共p⫹k兲兴, q⬍0
0, q⬎0. 共27兲
Equation 共27兲 describes the reflection of a classical par-ticle subject to the potential given by Eq. 共21兲 for energies below the height of the barrier. Equation共27兲 can be written
c共q,p兲⫽
再
k兩A兩2 m ␦冉
p2 2m⫺ k2 2m冊
, q⬍0 0, q⬎0, 共28兲 which explicitly shows the energy dependence of the SDF.We remark that, in order to treat the potential step with a normalized wave function, the time dependent wave packet approach for scattering can be used 关15兴. The corresponding time dependent WDF and SDF are calculated afterwards. In the limit when the incident packet reduces to a single plane wave one expects the result of Eq. 共27兲.
III. THE PO¨ SCHL-TELLER POTENTIAL
Here we present a third example, in which the mapping of the WDF corresponding to the ground state of the potential
V共q兲⫽⫺ U0
cosh2共aq兲, U0⬎0, 共29兲 into the SDF is performed. The parameters U0 and a are chosen such that the binding energy of the ground state be-comes equal to U0/2. This choice fixes the value U0 ⫽ប2a2/m.
The normalized wave function, the solution of the Schro¨-dinger equation corresponding to the potential given by Eq. 共29兲, is
共q兲⫽
冉
a2冊
1/2关cosh共aq兲兴⫺1. 共30兲
The WDF, which we obtained following Eq. 共8兲, is
共q,p兲⫽ sin共2pqប
⫺1兲
បsinh共2aq兲sinh关p共បa兲⫺1兴, 共31兲
which satisfies the condition兰(q, p)dq d p⫽1.
Making q⫽q
⬘
/(2a) and p⫽បap⬘
the expression 共31兲 becomes symmetric in the dimensionless variables q⬘
and p⬘
共see Fig. 2兲. In Fig. 2 we draw the curves of constant density in the first quadrant; for the other quadrants the symmetry
q→⫺q, p→⫺p should be applied. We observe that regions
in whichis positive alternate with regions where it is nega-tive. The points where (q, p) vanishes are given by the parabolas 2 pq⫽nប,n⫽⫾1,⫾2, . . . . The magnitude of in the region of negative values is small and occurs on points of phase space in which a classical particle would have posi-tive energy, thus not being trapped by the potential.
The classical trajectories „Q(q,p,t),P(q,p,t)…, the solu-tions of Newton’s equasolu-tions, are obtained by performing the integral
冕
dt⫽冕
dQ Q˙ ⫽冕
dQ冉
2E m ⫹ 2U0 mcosh2共aQ兲冊
⫺1/2 . 共32兲 The constants of integration are determined by imposing the condition (Q, P)⫽(q,p) at t⫽0.For E⬍0 the result is
sinh共aQ兲⫽a p cosh共aq兲sin共t兲
m ⫹sinh共aq兲cos共t兲,
共33兲
P cosh共aQ兲⫽p cosh共aq兲cos共t兲⫺m
a sinh共aq兲sin共t兲.
共34兲 The frequency is related to the energy E(q, p) of the tra-jectory by⫽a(2兩E兩/m)1/2. Equations共33兲 and 共34兲 may be written
sinh共aQ兲⫽
冉
U0 兩E兩 ⫺1冊
1/2
sin⌽共q,p,t兲, 共35兲
FIG. 1. The Po¨schl-Teller potential as a function of the dimen-sionless variable q⬘⫽aq.
FIG. 2. Curves of constant density(q,p) 共in units of ប⫺1), the WDF corresponding to the ground state for the potential in Eq.共29兲. The variables q⬘and p⬘are dimensionless, q and p being the co-ordinate and momentum, respectively.
P cosh共aQ兲⫽关2m共U0⫺兩E兩兲兴1/2cos⌽共q,p,t兲, 共36兲 where
⌽共q,p,t兲⫽t⫹共q,p兲.
. Here
共q,p兲⫽arctan关m sinh共aq兲/ap cosh共aq兲兴.
The SDF c is obtained by replacing (q, p) by „Q(q,p,t),P(q,p,t)… in Eq. 共31兲 for the WDF and integrat-ing it in t over one period. For this purpose, expressions共35兲 and共36兲 are solved explicitly in Q and P, the results being
Q⫽a⫺1ln关共1⫹F2兲1/2⫹F兴, P⫽共2mE兲1/2G共1⫹F2兲⫺1/2, 共37兲 where
F⫽共U0/兩E兩⫺1兲1/2sin⌽ and
G⫽共U0/兩E兩⫺1兲1/2cos⌽.
We mention here that(q, p) fixes the origin of the time and
c should not depend on it.
The final expression for c for E(q, p)⬍0 is
2បc共q,p兲⫽
冕
02
dsin兵2共aប兲
⫺1共2m兩E兩兲1/2g共1⫹ f2兲⫺1/2ln关共1⫹ f2兲1/2⫹ f 兴其
2 f共1⫹ f2兲1/2sinh兵共2m兩E兩兲1/2g关aប共1⫹ f2兲1/2兴⫺1其 , 共38兲
where f⫽(U0/兩E兩⫺1)1/2sin and g⫽(U0/兩E兩⫺1)1/2cos. Equation共38兲 shows thatcdepends on p and q only through
E(q, p).
For positive energies E(q, p) a similar calculation gives, as expected,c(q, p)⫽0, showing that the SDF corresponds indeed to a distribution of particles trapped inside the poten-tial.
In Fig. 3 we plot the curves along which c is constant 共which coincide with the classical trajectories兲. The trajec-tory markedc⫽0 in Fig. 3 corresponds to E(q,p)⫽0. All
trajectories for which E(q, p)⬎0 are situated at the right of this trajectory in this figure. As in the previous examples, we find also that only non-negative values for the density sur-vive. Fig. 4 is a superposition of Fig. 2 and Fig. 3. In Fig. 4, when comparing the quantal curve for a given with the corresponding classical curve 共same value of the density兲, one notices that the classical curve tends to move toward the left, away from the region corresponding to negative values of . Thus, the region in which the WDF assumes negative values is located in the domain wherecvanishes.
FIG. 3. Same as Fig. 2 forc(q, p). In this case the curves are
also the classical trajectories. The curve for whichc(q, p)⫽0
cor-responds to the energy E(q, p)⫽0. For points on the right of this curve E(q, p)⬎0 andc⫽0.
FIG. 4. Figures 2 and 3 are superposed. Dashed lines correspond toc and full lines to. Negative density curves are omitted.
IV. THE MODIFIED HARMONIC OSCILLATOR POTENTIAL
Now we present results corresponding to the modified harmonic oscillator 共MHO兲 potential
V共q兲⫽ប
2
2m关␣
2q2⫺2␣qtanh共q兲兴. 共39兲
The wave function that corresponds to the ground state of this potential is
⫽Ne(⫺␣q2/2)cosh共q兲, 共40兲 with normalization factor N2⫽2(␣/)1/2关exp(2/␣)⫹1兴⫺1 and energy eigenvalue
⑀共␣,兲⫽ ប 2
2m共␣⫺
2兲. 共41兲
The corresponding WDF may be calculated in closed form,
共q,p兲⫽Me(⫺␣q2⫺p2/␣ប2)
冋
e⫺2/␣cosh共2q兲⫹cos冉
2p␣ប
冊册
, 共42兲 where M⫽兵ប关1⫹exp(⫺2/␣)兴其⫺1 is the normalization factor. We shall consider ␣⫽2, in which case we get⑀(0,0)⫽0 and the potential, given by Fig. 5, has a hump in the middle of the oscillator.
Equation共15兲 corresponding to the SDFcwas evaluated numerically making use of the expression
c共q,p兲⫽T⫺1
冖
共Q,P„Q,E共q,p兲…兲mdQ P„Q,E共q,p兲…,
共43兲 where the integration is performed over a closed path with energy E(q, p). The period T was simply calculated through
T⫽
冖
mdQP„Q,E共q,p兲…, 共44兲
where P„Q,E(q,p)… is the momentum given as function of the coordinate Q,
P共Q,E兲⫽兵2m关E⫺V共Q兲兴其1/2. 共45兲
FIG. 5. The modified harmonic oscillator potential V⬘(q)
⫽2m(ប)⫺2V(q),q and V⬘(q) being dimensionless quantities.
FIG. 6. Curves of constant density(q,p) 共in units of ប⫺1) for the WDF corresponding to the ground state of the modified har-monic oscillator potential. The variables q⬘and p⬘are dimension-less.
FIG. 7. Same as Fig. 6 forc. In this case the curves are also
the classical trajectories. The dashed lines correspond to negative energies.
Equation共43兲 is obtained from Eq. 共15兲 by replacing the time
by the coordinate Q as variable of integration. A second convenient change of variables Q⫽Q() is made such that the singularities associated with the zeros of P(Q,E) at the classical turning points are canceled by corresponding zeros in the numerator in Eqs.共43兲 and 共44兲. Alternatively, we use the subtraction method for the elimination of these singulari-ties in order to evaluate these integrals.
In Figs. 6 and 7 we plot, respectively, lines of constant andc and in Fig. 8 we plotc versus E, corresponding to the MHO potential for the choice␣⫽2. From Fig. 6 we see that the domain of phase space where is negative has its coordinates restricted to the region of the hump of the poten-tial and it is periodic in the momentum variable. The value of
decreases rapidly in magnitude as the momentum in-creases. As can be seen from Figs. 7 and 8, there may exist more than one trajectory for a given value of c. In Fig. 9 we superpose parts of Figs. 6 and 7, plotting curves with fixed density ⫽c. The SDF we obtain by applying Eq. 共43兲 has similarities with the original WDF but again, as in the cases of Secs. II A, II B, and III, negative values are absent. One finds that for larger values of q the two curves with⫽c⫽const tend to get closer, a result that is expected because of the dominance of the harmonic oscillator part of the potential.
We remark here that the point (q, p)⫽(0,0), at the top of the hump, is a point of unstable equilibrium for classical
motion. Thus the period associated with the trajectory that passes through the point (q, p)⫽(0,0) becomes infinite since the particle has to arrive at the point q⫽0 with zero velocity. For this trajectory, which corresponds to E⫽0, the mo-mentum P(Q,0) given by Eq.共45兲 is proportional to Q in the neighborhood of the origin so that there is a pole at Q⫽0 in the integrand of Eq.共43兲 leading to a logarithmic singularity in the integral. This singularity is cancelled by the analogous singularity of T given by Eq. 共44兲. In fact, for phase-space points (q, p) in the neighborhood of the trajectory E(q, p) ⫽0 the SDF is given by បc⫽ 1 A⫹ln兩E共q,p兲兩 B⫹ln兩E共q,p兲兩, 共46兲
where A and B are slowly varying functions of E(q, p). Thus one gets the value c⫽1/ for points on the trajectory
E(q, p)⫽0. This value coincides with the maximum value of , which occurs precisely at the point (q, p)⫽(0,0). The functions A and B have to be evaluated numerically after subtraction of the pole term in the integrals in Eqs.共43兲 and 共44兲. As can be seen in Fig. 8, Eq. 共46兲 leads to a very sharp maximum ofcfor the trajectory passing through the origin of phase space and for which E(q, p)⫽0 and បc⫽1/. This path is compressed between the two curves for បc ⫽0.2 in Figs. 7 and 9, the upper curve corresponding to
⬘
(q, p)⫽2m(ប)⫺2E(q, p)⫽0.0010 and the lower one to
⬘
(q, p)⫽⫺0.0010.关1兴 E. Wigner, Phys. Rev. 40, 749 共1932兲.
关2兴 J.E. Moyal, Proc. Cambridge Philos. Soc. 45, 99 共1949兲. 关3兴 R.J. Glauber, Phys. Rev. 131, 2766 共1963兲.
关4兴 H.W. Lee and M.O. Scully, Found. Phys. 13, 61 共1983兲.
关5兴 P. Carruthers and F. Zachariasen, Rev. Mod. Phys. 55, 245 共1983兲.
关6兴 G.P. Maddison and D.M. Brink, Nucl. Phys. A 378, 566 共1982兲.
FIG. 8. The densityc共in units of ប⫺1) of the SDF as function
of the dimensionless classical energy⑀⬘(q, p)⫽E(q,p)2m(ប)⫺2.
FIG. 9. Figures 6 and 7 are superposed. Dashed lines correspond to and full lines to c.
关7兴 M.V. Berry, Philos. Trans. R. Soc. London, Ser. A 287, 237 共1977兲.
关8兴 K. Takahashi, Suppl. Prog. Theor. Phys. 98, 109 共1989兲. 关9兴 H.W. Lee, Phys. Rep. 259, 147 共1995兲.
关10兴 G.W. Bund, S.S. Mizrahi, and M.C. Tijero, Phys. Rev. A 53, 1191共1996兲.
关11兴 D. Galetti, A. Eiras, F.F. Souza Cruz, J.R. Marinelli, and M.M.
Watanabe de Moraes, Phys. Rev. C 48, 3131共1993兲. 关12兴 G.W. Bund, J. Phys. A 28, 3709 共1995兲.
关13兴 S.D. Bosanac, Phys. Scr. 57, 171 共1998兲.
关14兴 L. D. Landau and E. M. Lifshitz, Quantum Mechanics 共Perga-mon, London, 1958兲, p. 69.
关15兴 E. Merzbacher, Quantum Mechanics 共John Wiley, New York, 1998兲, p. 286.