Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
The Gaussian overlap approach (GOA) becomes inappropriate for describing the rotation of
weakly deformed systems. A modification is proposed which allows to maintain the GOA
for small deformations. The zero-point energy subtraction, derived from it, provides a simple
and reliable approximation for angular momentum projection. It becomes obvious,
however, that the projection complicates the equations which determine the motion along
the deformation path. These effects are studied in some simple models and the results are
condensed into a simple interpolation formula for the total zero-point energy.
1. Introduction
In microscopic theories of collective motion, one starts the full generator coordinate method leads to the well
with a set of deformed states [qi), where the ql are known angular momentum projection, which has al-
parameters describing the collective degrees of free- ready been applied to several nuclei [5, 6]. The energy
dom. The dynamics, given by the microscopic Hamil- difference between the projected and the unprojected
tonian/~, is traced down into the collective channel by PES can be understood as the rotational ZPE. Thus a
allowing only variations within the given space [q~). study of the ZPE also tackles problems of a.m. pro-
This is the well known generator coordinate method jection. In particular, the question of how to interpret
[1] which leads to an integral equation involving the the projected PES is discussed in this context.
Hamiltonian overlap (qll H ]q'i). Usually, one goes a The angular coordinates, beeing defined only in a finite
step further and tries to derive a collective Schr6dinger interval, are not very well suited for a GOA treatment.
equation from it. In particular, one prefers local In Section 2, it is shown that a slight modification of the
expansions, like the Gaussian overlap approach (GOA) GOA is possible which properly takes into account the
or an adiabatic time-dependent Hartree-Fock structure of the angular coordinates but still remains of
(ATDHF) [2] treatment, which are based on the over- second order in the collective derivatives. With this
laps at q~= q'~ and require at most second order deriva- modified GOA a simple formula for a.m. projection is
tives, 0~ [q~). In this paper we want to study the rele- obtained, whose relevance is tested in a harmonic
vance of the GOA for an axially symmetric system oscillator model. The a.m. projected PES, however,
having three degrees of freedom, one describing a de- cannot be interpreted directly as the PES for collective
formation and two for rotations of the deformed system motion. The ZPE for the deformation channel still
as a whole. Thereby we concentrate on the potential in ought to be subtracted. To do this properly, one has to
the collective Hamiltonian, often called the potential- exceed second order GOA and to take into account the
energy surface (PES). That means that we are mainly coupling of deformation and rotation, which is at least
concerned about the structure of the zero-point energy of fourth order. This is discussed in Section 3, where
(ZPE) [3, 4], contained therein. For rotational motion finally the experience, gained from these more elab-
orate considerations, is cooked down to a simple
* Supported by the Gesellschaftfilr Schwerionenforschungunder interpolation formula for the total ZPE, which again is
contract No. 06 Mz 709 of second order.
0340-2193/78/0285/0093/$01.40
94 P.-G. Reinhard: The Zero-Point Energy for Rotation
The last two terms therein subtract from V the kinetic S (1-x2)(-E R)
0
ZPE of the collective wave packet [fl, (#~, (ps). This ZPE 1 -(1-x2'
subtraction is important, first for theoretical reasons, as
it stabilises the collective Hamiltonian against the o
process classical limit and subsequent requantisation, = E R(c~g'/g) l== x./r (2.5)
and second for practical reasons, as the variation of the
where
ZPE can have quite an influence on the PES. It changes,
e.g., the barriers for fission by some MeV [4]. It can be 1
shown that in most cases this ZPE, as obtained from g(c0 = 5 dx e- ~(*-~'~).
0
second order GOA, is sufficient [7 3.
At small deformations, however, one should doubt the The modifying factor ~g'/g is shown as a function of
validity of the G O A for the rotational motion, since the 1/2-~ in Figure 1 (solid line). It properly approaches - 1
wave packets become less localised in the angular for large ZR, but suppresses the rotational ZPE for
coordinates. The width parameter 2 R becomes small vanishing deformation (2R --+0).
there and the overlaps do not decrease rapidly enough This behaviour can be understood by realising that the
within the range of(~o - ( 0 ' ) ~ 2n. Thus the GOA, tacitly subtraction of the rotational ZPE (2.5) is nothing else
assuming an infinite range for the coordinates, becomes than a.m. projection within the model (2.3). Simply
inappropriate. It ought to be modified such that the integrating over 12 and 12', we have tacitly assumed a
overlaps take into account the rotational structure of weight function f(12)= const., which just corresponds
the system. This can be achieved by replacing (qo- ~o')2 to the ground state with angular momentum J = 0. At
in (2.1) through (1 -(12- ~t'~,)2), where 12 and 12' are unit large deformations, the deformed state is well oriented
vectors, describing the orientation of the symmetry (A (p <2n) and the projection is equivalent to the ZPE
axis. Thus the modified G O A reads subtraction --ER, as it is given by the unmodified
P.-G. Reinhard: The Zero-Point Energy for Rotation 95
1. . . .
5, . . . . p )
9. ........... N;
"~ ..... N
"-.~\ --- N=16
9~\.~ N --+ m
\ (M~vJ'
-0.5
~\\....
9k '-...
k\. '"',,.
k \. """ "....
x \ ................... .................. /
\ /7s /-
-1 \. /
Z PE
' 01a 10 ' 172 ~'
Fig. 1. A comparison of the rotational Z P E in the deformed oscillator Fig. 2. The uncorrected PES (solid line) and the PES with the
model for different particle numbers N, drawn as a function of 1/;~R rotational Z P E (2.5) subtracted (dashed line) for the nucleus Ge 72,
described in the Ni'sson model. The deformation parameter fl is the
ratio of the average extensions in z- and in y-direction
GOA. At very small deformations, however, the state results converge rapidly towards the ZPE of the
vector has only small J 4 0 admixtures. Therefore, the modified GOA (which is the limit N ~ oo), showing that
change in energy has to approach zero, as it is the prescription (2.6) provides indeed a reliable ap-
reproduced properly by the ZPE (2.5). proach for a.m. projection.
Thus with the modified GOA (2.3) we have abtained a To have a realistic example, the approach (2.6) is
rather simple prescription for evaluating the a.m. applied to the nucleus Ge 72, where the wave functions
projected PES are taken from a Nilsson model and the PES is
A ~ A obtained with the Strutinsky prescription [8]. As
(LHL) ~, shown in Figure 2, the projected PES (dashed line) in
Y~roj = ~ - ~ ( g/g)l~=j:>/2 (2.6)
comparison to the unprojected one (solid line) is
comparable to the results from other, more elaborate,
which requires not more than second order derivatives a.m. projection calculations [9].
(, represented here by the total angular momentum ],).
In order to test the validity of this formula, we have
applied it to a schematic model, consisting in N 3. Interpretation of the Projected PES
distinguishable particles which are all put into a 1 s
state of a deformed oscillator, The question arises how to use the PES obtained from
~o,,n(r) = (~/2o)s/4 e- Zoe~'/3[r2+ (e- if- 1 ) ( , " l~>2] (2.7) the a.m. projection. In a projected Hartree-Fock or
Hartree-Bogoliubov theory it serves mainly to find the
where fi = fl/1/N. minimum and the wave function there. This procedure
This leads to the exact overlaps is justified, since the Hartree-Fock ground state is
allowed to contain the ZPE for collective vibrations
( I ) = [ 1 + 88sinh 2 (fi/2) (1 - (•. fl,)a)] - N/2 about it. For describing collective motion, however,
one still has to subtract the ZPE for the deformation
(Y) = (,) (-G) channel. One is tempted to do this in second order
sinh (fi/2)(e a + 1) (1 - ( ~ . fa') 2) GOA, i.e. Z P E = - E / 2 , and to interpret the result
(2.8)
(2e~+ 1)(1 +88sinh (fi/2) (1 - ( 0 . 0 ' ) 2 ) ) . as the PES for collective motion 9 But this is a bit too
hasty a conclusion. At large deformations we have
The detailed derivation is given in the Appendix. three degrees of freedom and three parts of ZPE to be
(From the Hamiltonian overlap only the kinetic energy subtracted, once El2 and twice ER/2. At small defor-
is of interest, because the potential energy does not mations, however, we do subtract only one part with
affect the rotations9 It is interesting to note that these the above procedure. Somewhere underway the sub-
overlaps reduce to the modified GOA (2.3) in the limit traction of another two ZPE's must have gone lost. This
N + o o . In Figure 1 we show the ZPE according to the rather handwaving argument is confirmed by the more
definition (29 evaluated with the exact overlaps (2.8), detailed analysis in the following. It will be become
for different particle numbers. Obviously, the exact clear that indeed the projected PES cannot be in-
96 P.-G. Reinhard: The Zero-Point Energy for Rotation
terpreted that simply as the PES for collective motion. The factor ~.R2g'/~ behaves similar as ~g'/g in (2.5). It
A proper treatment of the ZPE for the deformation starts with value zero at 2R 2 = 0 and approaches - 1
channel reveals a strong enhancement at small defor- for 2R2-+ oo. This rotational ZPE together with an
mations, which largely compensates the reduction of independent ZPE subtraction for the R degree of
the rotational ZPE there. freedom (taken from second order GOA) would yield V
=2/2M(l+2R2~'/~) which must be wrong, as the
exact result is V--0. One has to take into account the
3.1. The Model of a Gaussian Wave Packet coupling of R and f2, represented by the term (R - R ' ) 2.
in Three Dimensions (1 - f2. f2')/2 in (3.3), which has gone lost on the way to
second order G O A in (3.4). This drastically changes the
In order to understand, what happens in the process of overlaps for R motion after projection. They become
projection, we first consider a very simple model, which
allows to write down each step analytically. It consists (I) = (e -88 -R'V _ e -88 R') 2)
in a Gaussian wave packet in three dimensions, 9 (1 - - e - XR2) - 1/2 (1 - e-XW2)- 2/2
( I } = e - 88
For these overlaps a G O A is not very appropriate 9 It
22
(T}=(I}{3~ 81v1~(R-R')2~) (3.2) will even break down at 2R2--+0, because ),aOA/2
= (8"8R} approaches zero there. One has to use the full
thus leading to a ZPE R = 3 2/4M independent of R, i.e. a momentum expansion, Equation (2.4), and obtains for
ZPE of 2/4M for each degree of freedom. That means in the R motion ZPE R = - 2/4M(3/2 +),R 2 ~'/~), which
terms of PES, the unsubtracted PES is ~//"=(RITIR} adds up to - 3 2 / 4 M together with the rotational ZPE,
= 3 2 / 4 M and the true PES, after ZPE subtraction, thus yielding the desired effective PES V = 0.
becomes V=0. Altogether we have learned from this example: 1) The
Spherical coordinates, R--* R, I2, are introduced now. ZPE, left out in the a.m. projection, is not lost, but is
Thus one has to replace transferred into the radial channel. 2) One has to take
into account a coupling between the orientation f2 and
( R - R') 2 = ( R - R ' ) 2 the radial coordinate. This requires to go further than
+ [(R + R') z - (n - n')2](1 - f~. f2')/2 (3.3) second order in the derivatives. 3) The overlaps after
projection are rather likely not to fit into a G O A
in the overlaps (3.2). For second order G O A with scheme.
respect to R and f2 this reduces to
/
,/ the coordinates chosen. This, on the other hand,
suggests a general scheme for systematically deriving
t' modified GOA's. One should start with a cartesian
/ coordinate system for the collective parameters % i.e. a
70 system in which the kinetic energy can be expected to
018 1,0 ' 1'2 ~' take the form ~'. a2. There a GOA is most likely to be
Fig. 4. The uncorrected PES (solid line) and the PES with full ZPE
subtraction in the interpolated form (3.9) (dashed line) or from the valid. From this system one can undertake any desired
simple G O A (dotted line) transformation. The corresponding modified GOA is
obtained by inserting the reverse transformation into
line) which has an unreasonable bump at small defor- the GOA of the cartesian system. In the case of
mations, since the rotational ZPE is described wrong quadrupol motion, e.g., one should start with the five
there. deformation coordinates c~2, and then apply the well
known transformation to fi, 7 and the Euler angles.
4. Conclusion
The starting point was the observation that a GOA Appendix. Norm and Hamiltonian Overlap
expansion becomes inappropriate for describing ro- in the Deformed Oscillator Model
tations, if the system is only weakly deformed and the
angular orientation becomes rather uncertain, i.e. First we express the wave function (2.7) in a matrix
Aq)~2rc. In Section 2, a modified GOA has been notation
proposed which takes into account the periodicity of (pp,e(r)+0Z/2o) a/4e 2 (A.1)
the angular variables by incorporating the scalar
product of the orientations into the exponential. This
where A=e~/3(l+(e g-1)~?| fi=fi/l/N and
f~ | ~ is a dual product9 This simplifies the following
provides a very simple prescription for a.m. projection,
manipulations. Using the linear transformation r ' = ((A
requiring only second order derivatives, like the GOA.
Its reliability has been tested in a simple model. The
+d')/2)l/2r, we obtain for the norm overlap
most striking feature of the projected rotational ZPE is (P,~lfi',~'}
its fast fall offto zero at small deformations, driving the
projected PES to deformed minima.
One ought to be careful, however, with the interpre-
--- Det (7"6/20) 3/2 5 d 3 r ' e -;'~
tation of the projected PES. For determining the
Hartree-Fock minimum it might be useful. But if one is
interested in a PES for collective motion, one still has to
subtract the ZPE for the remaining deformation chan-
nel. The simple GOA added to the rotation ZPE gives = [88 - (~- fa') 2) d/a cosh(g3)
the wrong prediction. As shown in Section 3, the a.m.
9 sin(fl/2)sinh(fi'/2)
projection changes drastically the dynamics of the
deformation channel. In the spirit of a GOA expansion q- c o s h 2 ( 3 / 3 ) cosh(23/3)3 N/2
this can be accounted for by including the coupling where a=(fi-B')/2. For fl=fl', i.e. a =0, this obviously
between deformation and rotation with the fourth reduces to the rotational overlap (2.8). A Taylor
order term @~ J;2 ~ ) . Thi s yields, after projection, the expansion for 5 and 1-(f2-fa') 2 yields the GOA
overlaps for the fi-channel which usually cannot be expression
treated with second order GOA. A full momentum
expansion has to be applied. It yields a fi-ZPE which
increases at small deformations, thus compensating the 1 a2 1 ~'fl' . 2 q-N/2
decrease of rotational ZPE there. Both contributions +5N+ 1g-if-sin (~0)]
together give a rather smooth ZPE subtraction. In
order to avoid the complications of an extended GOA ~ e - ~-'~ - B')2- }~-si"2'~o) (A.3)
P.-G. Reinhard: The Zero-Point Energy for Rotation 99
where 2~ = 1/6 and 2R = ft. fi'/4.
The coupling between Again, in the limit N ~ oo only the lowest order in fi
f2 and/r is contained in 2R sin 2 A(p, since 2R ~B" fl' =fi2 =fi/1/N and 6 = 6~1IN
contributes. Thus the Hamil-
-(fi- fl')2/4.
In the limit N---, Go the GOA (A.3) is even tonian overlap becomes
exact, because the higher terms in the Taylor expansion
are supressed with g)Zn/N'.
It is interesting to note that
the fourth order term, ~ fi' sin 2(A q)), survives in the limit (H)=(I>{~f-N8~Tr{(A-d')2(A+A')Z} }
N ~ o% because the angular coordinate does not carry
the factor 1/N,
as fi does. Thus the (1/N)'-counting is
= (I){Yf 4M2~ 8M2~ sinz(Acp)flfl'}
shifted for all angular terms by one place.
For evaluating the Hamiltonian overlap from a micro-
scopic Hamiltonian,/t = T + ~,,we do not employ a two =(I){'U-Ep@c~2-ER~sin2(A~o)} (A.6)
body force for the potential energy, but are approx-
imating it by a single particle potential. It is a deformed
oscillator potential, whose deformation is taken as the where E~=ER=2o/2M and 2a,2 R are given in (A.3).
average between the deformation/3 of the bra and fi' of This is obviously the form as applied in (3.7). The
the ket. Thus we obtain the elementary overlap (i.e. the rotational overlap (2.8), for N < o% is obtained easily by
overlap for only one particle) taking the overlap before the limit N ~ oo, but at fi = fi',
i.e. 6 = 0. Finally, we want to point out again, that in the
(~,~IHI~,)~pa,t above model the limit N ~ oo proves the modified
~O GOA (2.3) to be exact for both overlaps, ( H ) and ( I ) .
=(~/2o) /
3 2
Sd re 2
3 ---rAr
o
\(T+ A \~]
22' ~r (A
~- + zr]]e-aZ~
~0 ^ ^
References
-(Tz/2~
- ( 23 )M
`o 20 r(e[_~,)er
8M
e-T.A+A')~