Vector parametrization of the N-atom problem in quantum

mechanics with non-orthogonal coordinates

Christophe Iung,a Fabien Gatti,a Alexandra Viela and Xavier Chapuisatb
a L aboratoire Structure et Dynamique des et Solides (UMR 5636) CC 014, Syste` mes Mole culaires
des Sciences et T echniques du L anguedoc, 34095 Montpellier Cedex 05, France Universite
b L aboratoire de Chimie (CNRS, URA 0506) Centre Scientique dOrsay 490), T he orique (Bat
Paris-Sud, 91405 Orsay Cedex, France Universite
Received 30th April 1999, Accepted 1st June 1999
This article aims to present a general method that enables one to build kinetic energy matrices in getting rid,
for the angular coordinates (internal and Eulerian), of the heaviness of dierential calculus (for expressing
kinetic energy operators) and numerical integration (for calculating matrix elements). Therefore, instead of
3N3 coordinates, only N1 radial distances are to be treated as coordinates. In the present formulation,
the system is described by any set of n vectors i =1, . . . , nN and the kinetic energy operator is expressed in MR
i
,
term of (n 1) angular momenta i =1, . . . , n 1N and the total angular momentum J. The formalism ML
i
,
proposed is general and gives a remarkably compact expression of the kinetic energy in terms of the angular
momenta. This expression allows one to circumvent the seeming angular singularities.
I. Introduction
In previous articles,h` the continuous geometrical representa-
tion of a deformable N-atom molecular system and the varia-
tional determination of its dynamical states, by
diagonalization of the Hamiltonian matrix in an appropriate
basis, have been studied for a triatomic molecule` or a mol-
ecule parametrized by Jacobi vectors. This approach avoids
the dierential calculus steps usually required for expressing
quantum mechanical kinetic energy operators in terms of
curvilinear internal coordinates. Indeed, many sets of internal
coordinates can be considered`h`" and the larger the mol-
ecule, the more diversied the coordinate sets that all include
3N6 internal coordinates. Moreover, the rotation of the
body-xed frame (BF) (whose axes rotate in a conventional
manner when the atoms move), is measured by three Euler
angles in the space-xed frame (SF).` In ref. 1, a set of n
Jacobi vectors i =1, . . . , n) has been used to describe the (R
i
,
molecule (n =N1). Vector has been taken parallel to the R
n
Gz"' axis whereas is parallel to the (xz)"' plane (G is the R
n
center of mass of the system). The (n 2) other vectors has
been viewed as totally free in BF. It is worth noticing that R
n
and played a specic role, being linked either totally R
n
(R
n
)
or partially to the denition of the orientation of BF. (R
n
)
Consequently, the BF components of the angular momenta
associated with and do not satisfy the usual proper- R
n
R
n
ties of angular momenta (commutation relations and
hermiticity) and do not commute with the other angular
momenta associated with the vectors [i =1, . . . , (n 2)] R
i
and the total angular momentum J. This crucial point will be
referred to several times in this work.
In the present article, we generalize the results obtained in
ref. 1 by using a set of n vectors which are not Jacobi vectors
and thus could be parametrized by non-orthogonal coordi-
nates, such as e.g. valence vectors. Our aim is to present a
general method that enables, for all angular coordinates
(internal and Eulerian), the construction of kinetic energy
matrices without resorting to (i) dierential calculus for
expressing kinetic energy operators and (ii) numerical integra-
tion to calculate angular matrix element for any selected
vectors i =1, . . . , nN. Consequently, only n radial dis- MR
i
;
tances must be treated numerically as coordinates instead of
3n coordinates when the angles are explicitly treated.
This work is based on the concept introduced in 1992 in ref.
3, where a general expression of the kinetic energy has been
given in terms of the BF-components of the angular momenta
associated with the vectors i =1, . . . , n 1N used to MR
i
;
describe the system. But the non-hermiticity of the BF-
components of the angular momenta associated with and R
n
and the non-commutation of them with the other R
n
angular momenta has not been fully appreciated in ref. 3. For-
tunately, we demonstrate in this article that most terms gener-
ated by non-hermiticity and non-commutation cancel each
other. Therefore, the quantization of the kinetic energy for N
atoms adds only very few specic terms, compared with the
classical expression of the kinetic energy.
This article also constitutes a generalization of ref. 2 which
has been dedicated to a vector parametrization of three-atom
systems by valence coordinates. In the latter work the
quantum expression of the kinetic energy has been established
in terms of the angular momenta associated with two valence
vectors only. The quantization has led to one single specic
quantum term, i.e. which has no counterpart in the classical
expression of the kinetic energy.
In Section 2, we present the quantum mechanical back-
ground used to express the kinetic energy operator in terms of
the momentum vectors i =1, . . . , nN conjugated to the n MP
i
;
vectors i =1, . . . , nN. In Section 3, the expressions of the MR
i
;
projections in BF of vectors i =1, . . . , nN are given, along MP
i
;
with their adjoints, in terms of the BF components of the
angular momenta i =1, . . . , n 1N of the (n 1) rotating ML
i
;
vectors i =1, . . . , n 1N and the total angular momen- MR
i
;
tum In Section 4, the results of Sections 2 and 3 are used for J.
deriving a compact expression of the kinetic energy operator
in terms of i =1, . . . , n 1N and This expression is ML
i
; J.
compared with its classical counterpart ; the main dierence
arises from the fact that the frame axes depend on and R
n
In Section 5, we propose a particular basis set which R
n
.
Phys. Chem. Chem. Phys., 1999, 1, 33773385 3377
results in an analytical angular representation of the kinetic
energy operator. Finally, in Section 6, the formalism is applied
to the hexa-atomic system (AB)CD(EF) which has been
recently tackled in the literature.`" We compare our method
with that initiated by Handy`` and used by Rempe and
Watts`" in order to give a developed expression of the kinetic
energy of (AB)CD(EF) described by valence coordinates. Some
general conclusions and prospective views are also drawn in
Section 6.
II. Quantum mechanical background
Let . . . , be the set of vectors chosen for the descrip- R

, R
`
, R
n
tion of the molecular system. It is possible to connect these
vectors to a set of Jacobi vectors . . . , (i.e. vectors r

, r
`
, r
n
pointing from one atomic group center of mass to another)
that describe the system uniquely. The relation between the
two sets is:
a
R

R
`

R
n
b
=A
a
r

r
`

r
n
b
(1)
where A is a non-singular constant matrix.`
The quantum expression of the kinetic energy operator, T,
in terms of Jacobi coordinates is given by```"
2T = ;
i
n p
i
sp
i
k
i
=(p

s , p
`
s , . . . , p
n
s)
]k(p

, p
`
, . . . , p
n
)t (2)
where the k matrix is the diagonal matrix of the reduced
masses associated with the Jacobi vectors and is the conju- p
i
gate momentum vector associated with Using eqn. (3) that r
i
.
relates the conjugate vector associated with to the P
i
R
i
p
i
vectors:
a
p

p
`

p
n
b
=At
a
P

P
`

P
n
b
(3)
Eqn. (2) can be rewritten in terms of the conjugate momentum
vectors (i =1, . . . , n) : P
i
2T = ;
ij
n
P
i
sM
ij
P
j
=(P

s , P
`
s , . . . , P
n
s)M(P

, P
`
, . . . , P
n
)t (4)
where the symmetric mass-dependent constant matrix M is
given by:
M=AkAt (5)
In Section 6, we will illustrate how the A and M matrices can
be easily obtained for a given system. In Section 3, we will
express the kinetic energy operator in terms of the total
angular momentum J and the angular momenta (i =1, . . . , L
i
n 1) associated with (i =1, . . . , n 1) and given by: R
i
J= ;
i
n
L
i
= ;
i
n
R
i
]P
i
(6)
The conjugate momentum can be expressed in terms of the P
i
angular momentum L
i
:
P
i
=P
i
re
i

e
i
]L
i
R
i
(7)
where denotes the unit vector along e
i
R
i
.
III. Conjugate momentum vector in the body xed
frame
First of all, the Gz"' axis chosen corresponds either to a sym-
metry axis of the system or to an axis that plays a specic role
in the modelization of the system. Vector is, by denition, R
n
parallel to this axis. Consequently, the orientation of is R
n
crucial and directly related to the physical property of the
molecule. For instance, in the study of treated in ref. 25, NH
`
links the nitrogen atom to the center of mass of the three R
n`
hydrogen atoms. Such a choice is adapted to the symmetry of
the equilibrium geometry and minimizes the eect of the
Coriolis couplings. is characterized by its spherical coordi- R
n
nates in SF. Consequently, the two rst Euler (R
n
, h
n
' , /
n
')
rotations and used to dene the Gz"' axis corre- D
z
(a) D
y
(b)
spond to and The frame obtained after these D
y
(h
n
') D
z
(/
n
').
two rotations is called A last rotation around is E
`
. D
z
(c) GzE`
dened such that lies parallel to the (x"'Gz"', x"' [0) R
n
half plane. This last rotation has no intrinsic physical
meaning.
The aim of this paper is to express the kinetic energy oper-
ator in terms of the BF components of n angular momenta,
namely the total angular momentum J and the (n 1) angular
momenta (i =1, . . . , n 1) associated to the freely rotating L
i
vectors (i =1, . . . , n 2) and to which is partially R
i
R
n
,
linked to BF frame. Consequently, the last angular momen-
tum is redundant and substituted for it. It is L
n
J;
i
n L
i
worth mentioning that is equal to zero because lies L
nz
"' R
n
parallel to the Gz"' axis.
Because of their dierent properties or expressions, one
should distinguish three types of angular momenta.
(i) The total angular momentum The BF components of J. J
are self-adjoint and satisfy the abnormal commutation rela-
tions. Their expressions are given elsewhere (for instance, refs.
26 and 27 or eqn. (19)(20) of ref. 1).
(ii) The angular momenta (i =1, . . . , n 2) associated L
i
with the freely rotating vectors (i =1, . . . , n 2). These R
i
vectors are independent of the denition of either SF, E2 or
BF. Consequently, they are characterized by the usual formu-
lae given in ref. 1 for instance [eqn. (34)(36)]. Their SF, E2
and BF components obey the normal commutation rela-
tions.`"
(iii) The angular momentum associated to L
n
R
n
.
The y-BF component of is not hermitian because BF L
n
depends on the orientation of We have established in R
n
.
ref. 1 that :
L
'n'B
"'
=L
'n'x
"'
_iL
'n'y
"'
=cot h
n
C
J
z
"'
;
i
n`
L
iz
"'
D
_ L
hn
(8)
L
'n'z
"'
=J
z
"'
;
i
n`
L
iz
"'
(9)
(L
'n'B
"'
)s =L
'n'B
"'
_ cot h
n
;
(L
'n'z
"'
)s =L
'n'z
"'
(10)
Using these equations, the BF components of can be L
n
s
expressed as:
(L
n
)s =L
n

1
0
i cot h
n
0
2
(11)
From now on, we are going to express the operators and P
i
s
in terms of the angular momenta (i =1, . . . , n 1) and P
i
L
i
their projections [eqn. (7)]. This formula will be necessary P
i
r
to establish in Section 4 the kinetic energy in terms of the BF
components of the angular momenta (i =1, . . . , n 1) and L
i
J.
3378 Phys. Chem. Chem. Phys., 1999, 1, 33773385
It is worth mentioning that a Euclidian normalization is
used,`"`" i.e. the elementary volume is:
dq =R
n
` sin b dR
n
da dbR
n
` sin h
n
"' dR
n
]dc dh
n
"' <
i
n`
R
i
` sin h
i
"' dR
i
d/
i
"' dh
i
"' (12)
In that case, the operators and are given by (P
i
r)s P
i
r P
i
r =
and: i L
Ri
(P
i
r)s =P
i
r
2i
R
i
(13)
Using eqn. (7), the BF components of the self-adjoint oper-
ators (i =1, . . . , n 2) and the operator are given by: P
i
P
n
P
i
=(P
i
)s =
P
i
r sin h
i
"' cos /
i
"'

1
R
i
(sin h
i
"' sin /
i
"'L
iz
"'
cos h
i
"'L
iy
"'
)
P
i
r sin h
i
"' sin /
i
"'

1
R
i
(cos h
i
"'L
ix
"'
sin h
i
"' cos /
i
"'L
iz
"'
)
P
i
r cos h
i
"'

1
R
i
(sin h
i
"' cos /
i
"'L
iy
"'
sin h
i
"' sin /
i
"'L
ix
"'
)
q n
t t
t t
t t
t t
t t
t t
t t
t t
t t
s p
(14)
Because of the non-hermiticity of the y-BF component of
[eqn. (10)], the x-BF component of is not self- L
n
P
n
adjoint.
(P
n
)s =e
n
P
n
rs
L
n
s ]e
n
R
n
(15)
A straightforward calculation provides the expression of
in terms of (P
n
)s P
n
:
(P
n
)s =P
n

a
i
R
n
sin h
n
"'
0
0
b
(16)
just in making use of the following commutators:
[L
'n'x
"'
, cos h
n
"' ] =0; [L
'n'x
"'
, sin h
n
"' ] =0 (17)
[L
'n'y
"'
, cos h
n
"' ] =i sin h
n
"' ;
[L
'n'y
"'
, sin h
n
"' ] =i cos h
n
"' (18)
[L
'n'z
"'
, cos h
n
"' ] =0; [L
'n'z
"'
, sin h
n
"' ] =0 (19)
Finally, the vector is obtained by substituting by its P
n
L
n
expression in terms of the other angular momenta:
1
R
n
A
J
y
"'
;
i
n
L
iy
"'
B
1
R
n
A
J
x
"'
;
i
n
L
ix
"'
B
P
n
r
q n
t t
t t
P
n
=t t (20)
t t
t t
s p
This equation, along with eqn. (8)(10), leads to the expres-
sion of the BF-components of P
n
s :
i cot h
n
"'
R
n
P
n
s =P
n
0 (21)
2i
R
n
q n
t t
t t
t t
t t
t t
s p
Therefore, all the expressions of the BF-components of P
i
and (i =1, . . . , n) are available. We are now able to P
i
s
propose a compact expression of the kinetic energy operator
in terms of the BF components of and (i =1, . . . , n 1). J L
i
IV. Quantum expression of the kinetic energy in
terms of angular momenta
First of all, the vectors are substituted in the expression of P
i
s
the kinetic energy operator [eqn. (4)] by their expressions
[eqn. (14), (16), (21)] which leads to:
2T = ;
ij
n
M
ij
P
i
P
j
(22)
;
i
n C
M
in
iP
ix
"'
R
n
sin h
n
"'
M
in
Ai cot h
n
"' P
ix
"'
R
n

2iP
iz
"'
R
n
BD
(23)
The rst term [eqn. (22)] can be identied with the expres-
sion of the kinetic energy in classical mechanics (which will be
called the classical kinetic energy), except for the BF com-
ponents of and which do not commute with the BF P
n
P
n
components of vectors Consequently, the commutators P
i
.
must be taken into account when [P
i
, P
j
] =P
i
P
j
P
j
P
i
deriving the kinetic energy. The result is:
2T = ;
i
n
M
ii
(P
i
)`
iM
nn
R
n
sin h
n
"'
P
'n'x
"'

iM
nn
R
n
(cot h
n
"' P
nx
"'
2P
nz
"'
) (24)
2 ;
i
n
;
j;ij
n
M
ij
P
i
P
j
2 ;
i
n`
M
in
P
i
P
n
2M
nn
P
n
P
n
(25)
;
i
n`
M
in
A

iP
ix
"'
R
n
sin h
n
"'
[P
n
, P
i
]
B
(26)
;
i
n`
M
in
Ai cot h
n
"' P
ix
"'
R
n

2iP
iz
"'
R
n
[P
n
, P
i
]
B
(27)
M
nn
Ai cot h
n
"' P
'n'x
"'
R
n

2iP
'n'z
"'
R
n

iP
nx
"'
R
n
sin h
n
"'
[P
n
, P
n
]
B
(28)
Therefore, the fact that the operators and are P
'n'y
"'
P
ny
"'
non-hermitian and that some commutators are not [P
i
, P
j
]
equal to zero, generate the last two terms in eqn. (24)(28). We
shall not examine eqn. (24) because it has been explicitly
treated in ref. 1, dedicated to the description of a molecule by
Jacobi vectors. We have established that eqn. (24) can be
rewritten in the following form:
;
i
n
M
ii
(P
i
)`
iM
nn
R
n
sin h
n
"'
P
'n'x
"'

iM
nn
R
n
](cot h
n
"' P
nx
"'
2P
nz
"'
) = ;
i
n
M
ii
C
(P
i
r)`
2iP
i
r
R
i
D
;
i
n AM
nn
R
n
`

M
ii
R
i
`
B
L
i
s L
i
;
i
n`
;
j;ij
n AM
nn
L
i
L
j
R
n
`
B

M
nn
(J J2;
i
n J L
i
)
R
n
`
(29)
We have shown that the order of the operators in the pro-
ducts that contain is not immaterial : placing the BF L
n
components of on the right hand side of all products is L
n
Phys. Chem. Chem. Phys., 1999, 1, 33773385 3379
particularly useful for simplifying the expression of the kinetic
energy operator and its representation in the angular basis set.
Therefore, it is worth noticing that the substitution of the (P
i
)`
by their expression [eqn. (7)] in the classical formula ;
i
n
leads exactly to the quantum expression eqn. (29), M
ii
(P
i
)`
except for which must be substituted for This is L
i
s L
i
(L
i
)`.
no longer true for i =n 1 and n.
Next, the quantization of the terms generated by the o-
diagonal masses is undertaken. Consequently, terms (25) to
(28) have to be calculated. The results are summarized below:
(i) Term (25) corresponds to the classical expression of the
kinetic energy generated by the o-diagonal masses. Vector
must be placed on the right of all scalar products where P
n
it appears.
(ii) Term (26) comes from the non-hermitian character of
and the non-commutation of with By means of P
n
P
i
P
n
.
the commutators given in Appendix 1, it can be shown that
this term is equal to zero, which implies that :
P
i
sP
n
P
n
s P
i
=2P
i
P
n
(30)
Here again, the order of the operators in the scalar product is
strictly xed.
(iii) Term (27) comes from the non-hermitian character of P
n
and the non-commutation of with With the help of the P
i
P
n
.
commutators given in Appendix 1, it can similarly be shown
that this term is equal to zero, i.e. :
P
i
sP
n
P
n
sP
i
=2P
i
P
n
(31)
(iv) Term (28) comes from the non-hermitian character of P
n
and and the non-commutation of with Using P
n
, P
n
P
n
.
Appendix 1, it can be shown that :
Eqn. (28) =
2iM
nn
R
n
]
G
P
n
r cos h
n
"'
[1 sin`(h
n
"' )]L
'n'y
"'
sin h
n
"'
H
(32)
The fact that term (28) is not equal to zero means that non-
hermiticity and non-commutation of the two operators gener-
ates purely quantum terms.
It should be emphasized that the calculations leading to
results in eqn. (30)(32) are long and tedious and that the
order of the operators in the scalar product is strictly xed, i.e.
is not immaterial.
The following equation establishes the relationship between
the classical expression of the kinetic energy and its
quantum counterpart :
2T
'
=2T
'''

iM
nn
R
n
sin h
n
"'
P
'n'x
"'

iM
nn
R
n
(cot h
n
"' P
nx
"'
2P
nz
"'
) (33)

2iM
nn
R
n
G
P
n
r cos h
n
"'
[1 sin`(h
n
"' )]L
'n'y
"'
sin h
n
"' R
n
H
(34)
2T
'''
= ;
i
n
M
ii
(P
i
)` 2 ;
i
n
;
j;ij
n
M
ij
P
i
P
j
2 ;
i
n`
M
in
P
i
P
n
2M
nn
P
n
P
n
(35)
Consequently, the quantum expression contains only two
additional terms generated by the diagonal masses and (M
nn
and two other terms coming from the o-diagonal M
nn
)
mass. M
nn
Another long and tedious calculation that takes into
account the non-commutation of the operators, leads to the
following expression of the kinetic energy for an N-atom mol-
ecule in terms of BF components of angular momenta and the
BF-angles and noted and h
i
"' /
i
"' h
i
/
i
(/
n
"' =0, h
n
"' =0) :
T J = ;
i
n
M
ii
A(P
i
r)`
2

iP
i
r
R
i
B
;
iji:j
n
M
ij
Msin(h
i
)sin(h
j
)cos(/
i
/
j
)
cos(h
i
)cos(h
j
)NP
i
r P
j
r
;
ijiEj
n
M
ij
sin(h
i
)sin(h
j
)sin(/
i
/
j
)
AP
i
rL
jz
R
j
B
;
ijiEj
n
P
i
r sin(h
i
)
AM
ij
cos(h
j
)
R
j

M
in
R
n
B
]
AeiiL
j
`eiiL
j

2i
B
;
i
n M
in
R
n
P
i
r sin(h
i
)
AeiiL
i
`eiiL
i

2i
B
;
ijiEj
n M
ij
R
i
P
j
r cos(h
j
)sin(h
i
)
AeiiL
i
`eiiL
i

2i
B
;
iji:j
n GM
ij
R
i
R
j
[cos(h
i
)cos(h
j
)

`
eiisin(h
i
)eijsin(h
j
)]

M
in
R
i
R
n
cos(h
i
)
M
jn
R
j
R
n
cos(h
j
)
M
nn
2R
n
`
HAL
i
`L
j

2
B
;
iji:j
n GM
ij
R
i
R
j
[cos(h
i
)cos(h
j
)

`
eii sin(h
i
)eijsin(h
j
)]

M
in
R
i
R
n
cos(h
i
)
M
jn
R
j
R
n
cos(h
j
)
M
nn
2R
n
`
HAL
i
L
j
`
2
B
;
iji:j
n M
ij
4R
i
R
j
sin(h
i
)sin(h
j
)
](ei'i`j'L
i
`L
j
`ei'i`j'L
i
L
j
)
;
i
n GM
nn
2R
n
`

M
ii
2R
i
`

M
in
cos(h
i
)
R
i
R
n
H
L
i
sL
i
;
iji:j
n GM
ij
R
i
R
j
sin(h
i
)sin(h
j
)cos(/
i
/
j
)
M
nn
2R
n
`
H
L
iz
L
jz
;
iji:j
n
sin(h
j
)
A

M
ij
R
i
R
j
cos(h
i
)
M
jn
R
j
R
n
B
]
AeijL
i
`eijL
i

2
B
L
jz
;
iji:j
n
sin(h
i
)
A

M
ij
R
i
R
j
cos(h
j
)
M
in
R
i
R
n
B
]
AeiiL
iz
L
j
`eiiL
iz
L
j

2
B
;
i
n M
in
R
n
P
i
r sin(h
i
)
AeiiJ`eiiJ
2i
B
;
i
n M
in
sin(h
i
)
R
i
R
n
AeiiJ`eiiJ
2
B
L
iz

GM
in
cos(h
i
)
R
i
R
n

M
nn
R
n
`
HAJ`L
i
JL
i
`
2
B
;
i
n AM
nn
J
z
L
iz
R
n
`
B

M
nn
J`
2R
n
`
;
i
n M
in
R
i
R
n
]sin(h
i
)
AeiiL
i
`eiiL
i

2
B
(36)
3380 Phys. Chem. Chem. Phys., 1999, 1, 33773385
The advantage of this equation compared to some developed
expression previously proposed by other authors in the case of
polyatomics is to be as compact and general as possible. This
later point will be discussed in Section 6. The physical
meaning of each term is clear because of the presence of the
angular momenta. Therefore, if we compare this equation to
those obtained by substituting, in the classical expression
vectors for their expression [eqn. T =
`
;
ij
n P
i
M
ij
P
j
, P
i
(7)], there is only one purely quantum term, namely ;
i
n
under the condition (M
in
/R
i
R
n
)sin(h
i
"')(eiiL
i
`eiiL
i
)/2,
that the BF-components of are appropriately placed on L
n
the right of each product.
In a previous work dedicated to triatomic molecules,` the
quantization of the classical expression of the kinetic energy
provided an additional term, which (iM
`
/R

R
`
)sin(h

"') L
y
,
corresponds to the extra term obtained in the general formula
if n =2 and were denoted in this (/
`
"' =0). M
`
, R

, R
`
h

"'
article as M, r, R and a, respectively. Therefore, the quantum
and classical expressions are very similar only if the non-
commutating operators are placed in the dierent products in
a prescribed order.
V. Angular basis functions: integration over
angular coordinates, matrix representation
We now focus our attention on the action of the kinetic oper-
ator on an angular basis set that described the various rota-
tions of the molecule, parametrized by three Euler angles (a =
and the spherical angles of the vectors h
n
' , b =/
n
' , c =/
n
E` )
(i =1, . . . , n 1) viewed in BF R
i
(h
i
"' , /
i
"').
Moreover, in the absence of external eld, SF is isotropic,
i.e. the orientation of z' is arbitrary, so that any observable
must be a-independent.`" The overall rotation of the molecule
can thus be described by the following basis set :`h``
S(a), b, c o J, 0, XT =Y
J
X(b, c)(1)X (37)
where X is the projection of on Gz"'. J
The current element of the working angular function basis
for the BF spherical angles of vectors (i =1, . . . , n 1) is R
i
given by:
S(a), b, c, h
n
"' , /

"' , h

"', . . . ,
/
n`
"' , h
n`
"' o X, l
n
, l

, X

, . . . , l
n`
, X
n`
T
J
=Y
J
X(b, c)(1)XP
ln
XXXn`[cos(h
n
"' )]
]Y
l
X(h

"' , /

"') . . . Y
ln`
Xn`(h
n`
"' , /
n`
"' ) (38)
where is a normalized associated Legendre func- P
l
X[cos(h)]
tion times (1)X and /) is a spherical harmonics Y
l
X(h,
Y
l
X(h, /) =P
l
X[cos(h)](1/J2p)eiX.
The current element of this working angular basis set is,
hereafter, denoted by MSangles o . . . T
J
N.
The action of the kinetic energy operator on the angular
basis functions requires the use of the basic formulae given in
Appendix 2 for the angular momenta (i =1, . . . , n 2) as L
i
well as the specic formula established in ref. 1 [eqn. (73a)],
given below for L
'n'B
"'
, L
'n'z
"'
:
L
'n'B
"'
MSangles o . . . T
J
N
=c
B
(l
n
, X
n
)S(a), b, c o J, 0, XT
P
ln
X'-
i
n

`
Xi'B[cos(h
n
"' )]
]S/

"' , h

"' , . . . , /
n`
"' , h
n`
"' o l

, X

, . . . , l
n`
, X
n`
T
(39)
L
'n'z
"'
MSangles o . . . T
J
N
=i
A
X ;
i
n`
X
i
B
MSangles o . . . T
J
N (40)
Actually, no real difficulties are generated by the fact that
the operators and (i =1, . . . , n 2) do not satisfy L
n
L
i
similar relationships. In order to illustrate this important
point, the following calculation shows how one of the more
complicated terms of the kinetic energy, [M
in
/4R
i
R
n
(i =1, . . . , n 2) acts upon the sin(h
i
"')sin(h
n
"' )eiiL
i
`L
n
` ]
angular basis function:
sin(h
i
"')sin(h
n
"' )eiiL
i
`L
n
` MSangles o . . . T
J
N
=c
`
(l
n
, X
n
)sin(h
i
"')sin(h
n
"' )
]eiiL
i
`S(a), b, c o J, 0, XT
P
l'n'
X'-
i
n

`
Xi'`[cos(h
n
"' )]
]S/

"' , h

"' , . . . , /
n`
"' , h
n`
"' o l

, X

, . . . , l
n`
, X
n`
T
J
=c
`
(l
n
, X
n
)c
`
(l
i
, X
i
)sin(h
i
"')sin(h
n
"' )
]eiiS(a), b, c o J, 0, XT
P
l'n'
X'-
i
n

`
Xi'`[cos(h
n
"' )]
]S/

"' , h

"' , . . . , /
n`
"' , h
n`
"' o l

, X

, . . . , l
i
, X
i
1, . . . T
J
=c
`
(l
n
, X
n
)c
`
(l
i
, X
i
)S(a), b, c o J, 0, XT
]sin(h
n
"' )P
l'n'
X'-
i
n

`
Xi'`[cos(h
n
"' )]
]MB
`
(l
i
, X
i
1)
]S/

"' , h

"' , . . . , /
n`
"' , h
n`
"' o l

, X

, . . . , l
i
1, X
i
, . . . T
J
B

(l
i
, X
i
1)
]S/

"' , h

"' , . . . , /
n`
"' , h
n`
"' o l

, X

, . . . , l
i
1, X
i
, . . . T
J
N
=c
`
(l
n
, X
n
)c
`
(l
i
, X
i
)
]MB
`
(l
i
, X
i
1)B
`
(l
n
, X
n
1)
]S . . . o . . . , l
n
1, X
n
, . . . , l
i
1, X
i
, . . . T
J
B
`
(l
i
, X
i
1)B

(l
n
, X
n
1)
]S . . . o . . . , l
n
1, X
n
, . . . , l
i
1, X
i
, . . . T
J
B

(l
i
, X
i
1)B
`
`(l
n
, X
n
1)
]S . . . o . . . , l
n
1, X
n
, . . . , l
i
1, X
i
, . . . T
J
B

(l
i
, X
i
1)B

`(l
n
, X
n
1)
]S . . . o . . . , l
n
1, X
n
, . . . , l
i
1, X
i
, . . . T
J
N (41)
In the latter expressions, the notation introduced in Appendix
2 has been used.
We now calculate the matrix representing the kinetic T J,
energy operator in the angular basis set The MSangles o . . . T
J
N.
integration is over the angles only, [i.e. the matrix elements
are expressed in terms of J, X, M, (i =1, . . . , n 2], and l
i
, X
i
on the one hand, and (i =1, . . . , n) on the other l
n
R
i
P
i
r
hand. J and M=0 are xed, i.e. is a diagonal block of T J T
at constant J. In the basis using an approach MSangles o . . . T
J
N,
similar to that used for eqn. (41), it is rather easy but long to
establish the matrix elements . . . , S . . . X
i
, l
i
, X
j
,
. . . , which are non-zero. Inspec- l
j
. . . o T J o . . . X
i
, l
i
, X
j
, l
j
. . . T
tion of eqn. (36) reveals that the non-zero matrix elements
full the following conditions:
(i) *X=XX@ =0, _1
_1 (#i =1, 2, . . . , n 2) (ii) *X
i
=X
i
X
i
=0,
. . . , n 2) (iii) *X=*X
i
*X
j
(#i Dj =1,
_1 (#i =1, 2, . . . , n 1) (iv) *l
i
=l
i
l
i
=0,
Consequently, the non-zero matrix elements . . . , S . . . X
i
, l
i
,
. . . , with i \j are given by the X
j
, l
j
. . . o T J o . . . X
i
, l
i
, X
j
, l
j
. . . T
following formulae in which only the modied quantum
numbers are mentioned. Therefore, represents either or e
i
, and X) = c
`
(J, X) =`JJ(J1) X(X1) c

(J,
`JJ( J 1) X(X1).
Phys. Chem. Chem. Phys., 1999, 1, 33773385 3381
(i) Diagonal terms of T J
S . . . X
i
, l
i
, . . . , X
j
, l
j
. . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
= ;
i
n
M
ii
A(P
i
r)`
2

iP
i
r
R
i
B
;
i
n
`l
i
(l
i
1)
AM
nn
2R
n
`

M
ii
2R
i
`
B
;
iji:j
n M
nn
`X
i
X
j
2R
n
`
;
i
n M
nn
`XX
i
R
n
`

M
nn
`J( J 1)
2R
n
`
Let us note that
X
n
=X ;
i
n`
X
i
(42)
(ii) Non-diagonal terms for *X=0, *X
i
=0; *l
i
=_1
S . . . X
i
, l
i
(e
i
1) . . . o T J o . . . X
i
, l
i
, . . . T
=
G

M
in
P
n
r
2iR
i

M
in
P
i
r
2iR
n

M
in
2R
i
R
n
H
]Mc
`
(l
i
, X
i
)B
ei
(l
i
, X
i
1) c

(l
i
, X
i
)B
ei
`(l
i
, X
i
1)N

G
M
in
P
i
rP
n
r
M
in
R
i
R
n
`l
i
(l
i
1)
H
B
ei
"(l
i
, X
i
) (43)
(iii) Non-diagonal terms for *X=_1, *X
i
=_1; *l
i
=1
or 1
SX_1, . . . X
i
_1, l
i
(e
i
1) . . . o T J o . . . X
i
, l
i
, . . . T
=
G
<
M
in
P
i
r
2iR
n

M
in
X
i
2R
i
R
n
H
c
B
( J, X)B
ei
B(l
i
, X
i
)

M
in
2R
i
R
n
B
ei
"(l
i
, X
i
_1)c
B
(l
i
, X
i
)c
B
( J, X) (44)
(iv) Non-diagonal terms for *X=_1, *X
i
=_1; l
i
=0
SX_1, . . . X
i
_1, l
i
. . . o T J o . . . X
i
, l
i
, . . . T
=
M
nn
2R
n
`
c
B
(l
i
, X
i
)c
B
( J, X) (45)
(v) Non-diagonal terms for *X=0, *X
i
=*X
j
=_1;
*l
i
=*l
j
=0
S . . . X
i
_1, l
i
, . . . , X
j
<1, l
j
. . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
=c
B
(l
i
, X
i
)c
Y
(l
j
, X
j
)
M
nn
4R
n
`
(46)
(vi) Non-diagonal terms for *X=0, *X
i
=*X
j
=0; *l
i
=
or 1 _1; *l
j
=1
S . . . X
i
, l
i
(e
i
1), . . . , X
j
, l
j
(e
j
1) . . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
=M
ij
P
i
rP
j
rB
ei
"(l
i
, X
i
)B
ej
"(l
j
, X
j
)
M
ij
P
j
r
2iR
i
]c
`
(l
i
, X
i
)B
ei
(l
i
, X
i
1)B
ej
"(l
j
, X
j
)

M
ij
P
i
r
2iR
j
c
`
(l
j
, X
j
)B
ej
(l
j
, X
j
1)B
ei
"(l
i
, X
i
)

M
ij
P
j
r
2iR
i
c

(l
i
, X
i
)B
ej
"(l
j
, X
j
)B
ei
`(l
i
, X
i
1)

M
ij
P
i
r
2iR
j
c

(l
j
, X
j
)B
ej
`(l
j
, X
j
1)B
ei
"(l
i
, X
i
)

M
ij
4R
i
R
j
B
ei
(l
i
, X
i
1)B
ej
`(l
j
, X
j
1)c
`
(l
i
, X
i
)c

(l
j
, X
j
)

M
ij
4R
i
R
j
B
ei
`(l
i
, X
i
1)B
ej
(l
j
, X
j
1)c

(l
i
, X
i
)c
`
(l
j
, X
j
)

M
ij
4R
i
R
j
B
ei
(l
i
, X
i
1)B
ej
(l
j
, X
j
1)c
`
(l
i
, X
i
)c
`
(l
j
, X
j
)

M
ij
4R
i
R
j
B
ei
`(l
i
, X
i
1)B
ej
`(l
j
, X
j
1)c

(l
i
, X
i
)c

(l
j
, X
j
)
(47)
(vii) Non-diagonal terms for *X=0, *X
i
=*X
j
=_1;
*l
i
=_1; *l
j
=0
S . . . X
i
_1, l
i
(e
i
1), . . . , X
j
<1, l
j
. . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
=_
P
i
rM
in
2iR
n
B
ei
B(l
i
, X
i
)c
Y
(l
j
, X
j
)

C M
in
2R
i
R
n
c
B
(l
i
, X
i
)c
Y
(l
j
, X
j
)B
ei
"(l
i
, X
i
_1)

M
in
2R
i
R
n
X
i
c
Y
(l
j
, X
j
)B
ei
B(l
i
, X
i
)
D
(48)
(viii) Non-diagonal terms for *X=0, *X
i
=*X
j
=_1;
*l
i
=0; *l
j
=_1
S . . . X
i
_1, l
i
, . . . , X
j
<1, l
j
(e
j
1) . . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
=
G
<
P
j
rM
jn
2iR
n

M
jn
X
j
2R
j
R
n
H
c
B
(l
i
, X
i
)B
ej
Y(l
j
, X
j
)

M
jn
2R
j
R
n
c
B
(l
i
, X
i
)c
Y
(l
j
, X
j
)B
ej
"(l
j
, X
j
<1) (49)
(ix) Non-diagonal terms for *X=0, *X
i
=*X
j
=_1;
*l
i
=_1; *l
j
=_1
S . . . X
i
_1, l
i
(e
i
1), . . . , X
j
<1, l
j
(e
j
1) . . . o T J o . . . X
i
, l
i
, . . . , X
j
, l
j
. . . T
=
M
ij
2
G
P
i
rP
j
r _
P
i
rX
j
iR
j
<
P
j
rX
i
iR
i

`X
i
X
j
R
i
R
j
H
]B
ei
B(l
i
, X
i
)B
ej
Y(l
j
, X
j
)

M
ij
2R
i
G
<iP
j
r
X
j
R
j
H
c
B
(l
i
, X
i
)B
ei
"(l
i
, X
i
_1)B
ej
Y(l
j
, X
j
)

M
ij
2R
j
G
_iP
i
r
X
i
R
i
H
c
Y
(l
j
, X
j
)B
ej
"(l
j
, X
j
<1)B
ei
B(l
i
, X
i
)

M
ij
2R
i
R
j
c
B
(l
i
, X
i
)c
Y
(l
j
, X
j
)B
ei
"(l
i
, X
i
_1)B
ej
"(l
j
, X
j
<1)
(50)
VI. Example and discussion
The conguration of an N-atom molecule can be described by
N1 relative position vectors after elimination of the center-
of-mass motion. Many sets of coordinates commonly used for
describing molecules can be viewed as spherical coordinates
for these vectors. The spherical angles are local, i.e. they are
dened for frames which change from one vector to another.
The coordinates actually consist of (i) the N1 vector
lengths, (ii) N2 planar angles between pairs of vectors, (iii)
N3 dihedral angles between two vectors around a third
one, and (iv) three Euler angles orienting the BF frame with
respect to the SF frame. These 3N3 coordinates are the
local spherical coordinates for the N1 vectors. This article
has aimed at exploiting this type of parametrization.
3382 Phys. Chem. Chem. Phys., 1999, 1, 33773385
Fig. 1 (AB)CD(EF) system parametrized by 5 valence vectors.
The rst innovation of the method proposed in this paper is
to build a particularly compact expression of the kinetic
energy. Moreover, in this expression, each term possesses its
own physical meaning and is individually hermitian. This
approach is absolutely general and can be applied to any
system. In order to show how it is applied in a particular case,
we shall consider the system (AB)CD(EF) (Fig. 1) because the
determination of its kinetic energy operator (for J =0) has
been the subject of a recently published work.`" In that work,
Rempe and Watts have followed the method initiated by
Handy`` and used valence coordinates to describe the molec-
ular deformations. They have actually applied, with the help
of Mathematica,`" the chain rule twice and obtained the devel-
oped expression of the kinetic energy containing more than
400 terms, all of them being not individually hermitian. Let us
apply our approach to this system with the same coordinates,
i.e. the spherical coordinates of the valence vectors rep- R
i
resented in Fig. 1. Each vector is parametrized by spherical R
i
coordinates and respectively bond lengths, bond R
i
, h
i
/
i
,
angles and torsion angles. The correspondence between
Rempe et al.s valence coordinates and the spherical coordi-
nates of used in our approach is easily established since R
i
R
n
and are and R
n
CDm CE m .
Fig. 2 The 5 Jacobi vectors used in the calculation of the matrix M.
Fig. 3 Valence vectors and used to parametrize and (R

, R
`
R
`
) CH
`
Jacobi vectors and describing the motion of A and B in the (R
"
R
`
)
course of an reaction. SN
`
(x, y, z) Rempe axes ](y, x, z) axes in this paper
h

" , h
`
" , h
`
" , h
"
" , h
`
"
]h

, p h
`
, h
`
, p h
"
, h
`
In order to derive a compact expression of the kinetic energy
operator with the angular momentum coupling terms in eqn.
(36), we have to determine the M matrix [eqn. (5)] for
(AB)CD(EF). We follow the method proposed in Section 2.
First a set of 5 Jacobi vectors i =1, . . . , 5) has to be (r
i
,
dened (Fig. 2). The A matrix relating the valence vectors (R
i
,
i =1, . . . , 5) parametrizing the system to the Jacobi vectors
selected (Fig. 2) is:
R

m
"
m
^"
0
m
''
m
^"''
1
m
'
m
''
r

R
`
0
m
'
m
''
m
^"
m
^"''
1
m
'
m
''
r
`
R
`
=
m
^
m
^"
0
m
''
m
^"''
1
m
'
m
''
r
`
(51)
R
"
0
m
'
m
''
m
^"
m
^"''
1
m
'
m
''
r
"
R
`
0 0 0 0 1 r
`
q n q nq n
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
t t t tt t
s p s ps p
where and are respectively the masses m
^
, m
"
, m
'
, m
'
, m
'
m
'
of A, B, C, D, E and F while m
^"
=m
^
m
"
, m
''
=m
'
m
'
,
and Next, m
''
=m
'
m
'
m
^"''
=m
^
m
"
m
'
m
'
.
from eqn. (5), the following M matrix is obtained.
k
^'
0
1
m
'
0 k
''
0
M=
1
m
'
0 k
"'
0
1
m
'
0
1
m
'

1
m
'
1
m
'
q
t
t
t
t
t
t
t
t
t
t
s
0
1
m
'
1
m
'

1
m
'
0
1
m
'
(52)
k
''

1
m
'

1
m
'
k
''
n
t
t
t
t
t
t
t
t
t
t
p
where
k
ij
=
1
m
i

1
m
j
(i, j =A, B, C, D, E, F).
Consequently, the expansion including more than 400 terms
proposed by Rempe and Watts can be factorized by using eqn.
(36), with n =5 and the masses given in eqn. (52). The M
ij
compact character of the present method is thus clearly illus-
trated in this particular case.
The second innovation consists in obtaining analytical
expressions for the action of all angular operators in the basis
functions in an adequate representation. Simultaneously,
getting rid of the angular singularities is a straightforward
matter. The next step for completing the present formalism is
the demonstration of the fact that the treatment of the angular
singularities can be generalized to the radial singularities.``
This generalization is required for studying e.g. the inversion
motion of the ammonia molecule.`` To this large amplitude
motion is associated the vector linking the center of mass of
to the nitrogen atom. The vector length is zero when the H
`
molecule is planar and the subsequent seeming radial singu-
larity can also be avoided. Moreover, our approach can be
improved by taking into account the permutation symmetries
of the system. A symmetry adapted basis can be dened. All
these improvements have been used for studying the ammonia
spectroscopy.``
Phys. Chem. Chem. Phys., 1999, 1, 33773385 3383
If there are no identical particles, the method introduced
can be straightforwardly used. The exibility of our approach
is also worth mentioning: any set of vectors can be used. For
instance, for the study of a reaction SN
`
(ACH
`
B]ACH
`
B), the combination of Jacobi and valence vectors illus-
trated in Fig. 3 is advisable. To achieve the construction of
matrix TJ radial basis functions are selected. The integrals
over the radial coordinates must be numerically calculated.
From that point on, the rest of the work is numerical and the
numerical eort will clearly impose limits to the size of the
systems that can be actually treated. For systems free in all
their deformation degrees of freedom, ve particles may be
treated at present. If larger systems are considered, model con-
straints must be introduced, such as freezing a part of the
system. This subject, is to a large extent, still to be explored.
In all cases, it should be emphasized that it is protable to
have a quantum FBR (nite basis representation) in which the
kinetic energy matrix is sparse. Combined with a DVR (dis-
crete variable representation) for the potential, the FBR that
we propose constitutes an appropriate framework for future
dynamical studies of more-than-three particle molecules (see
ref. 35 and ref. 36 for instance).
Acknowledgements
Claude Leforestier is warmly thanked for many fruitful Prof.
discussions.
Appendix 1
In this appendix, the various commutators that are required
for the calculations in this article are given. They can be easily
obtained from the denition of the BF-components (i =1, L
i
. . . , n 2).
[L
ix
, sin /
i
] =i cos`(/
i
)cot h
i
;
[L
ix
, cos /
i
] =i cos /
i
sin /
i
cot h
i
(53)
[L
ix
, sin h
i
] =i sin /
i
cos h
i
;
[L
ix
, cos h
i
] =i sin /
i
sin h
i
(54)
[L
iy
, sin /
i
] =i sin /
i
cot h
i
cos /
i
;
[L
iy
, cos /
i
] =i sin`(/
i
)cot h
i
(55)
[L
iy
, sin h
i
] =i cos /
i
cos h
i
;
[L
iy
, cos h
i
] =i cos /
i
sin h
i
(56)
[L
iz
, sin /
i
] =i cos /
i
; [L
iz
, cos /
i
] =i sin /
i
(57)
[L
iz
, sin h
i
] =[L
iz
, cos h
i
] =0 (58)
Appendix 2
This appendix puts together the various formulae required for
application of the kinetic energy operator upon the basis func-
tions given in eqn. (41) (see ref. 33).
cos(h)P
l
X(cos h) =B
`
" (l, X)P
l`
X (cos h) B

" (l, X)P

l
X (cos h)
(59)
sin hP
l
X(cos h) =B
`
`(l, X)P
l`
X`(cos h) B

`(l, X)P
l
X`(cos h)
(60)
sin hP
l
X(cos h) =B
`
(l, X)P
l`
X(cos h) B

(l, X)P
l
X(cos h)
(61)
dP
l
X(cos h)
dh
=B
"
`(l, X)P
l
X`(cos h) B
"
(l, X)P
l
X(cos h)
(62)
where
B
`
" (l, X) =
S(l X1)(l X1)
(2l 1)(2l 3)
;
B

" (l, X) =
S (l X)(l X)
(2l 1)(2l 1)
(63)
B
`
`(l, X) =
S(l X1)(l X2)
(2l 1)(2l 3)
;
B

`(l, X) =
S(l X)(l X1)
(2l 1)(2l 1)
(64)
B
`
(l, X) =
S(l X1)(l X2)
(2l 1)(2l 3)
;
B

(l, X) =
S(l X)(l X1)
(2l 1)(2l 1)
(65)
B
"
`(l, X) =
`
J(l X1)(l X) ;
B
"
(l, X) =
`
J(l X)(l X1) (66)
The following relationships are obtained for the spherical
harmonics:
cos(h)Y
l
X(h, /) =B
`
" (l, X)Y
l`
X (h, /)
B

" (l, X)Y

l
X (h, /) (67)
exp(i/)sin(h)Y
l
X(h, /) =B
`
`(l, X)Y
l`
X`(h, /)
B

`(l, X)Y
l
X`(h, /) (68)
exp(i/)sin(h)Y
l
X(h, /) =B
`
(l, X)Y
l`
X(h, /)
B

(l, X)Y
l
X(h, /) (69)
To specify how the operators and L
i`
"' , L
i
"' , L
iz
"'
(L
i
`)"'
(i =1, . . . , n 2) act upon the spherical harmonics, the follow-
ing classical expressions for the kinetic momenta are used:
L
iB
"'Y
li
Xi(h
i
, /
i
) =c
i
B(l
i
, X
i
)Y
li
XiB(h
i
, /
i
)
=Jl
i
(l
i
1) _X
i
(X
i
1)Y
li
XiB(h
i
, /
i
) (70)
L
iz
"'
Y
li
Xi(h
i
, /
i
) =X
i
Y
li
Xi(h
i
, /
i
)
(L
i
`)"'Y
li
Xi(h
i
, /
i
) =`l
i
(l
i
1)Y
li
Xi(h
i
, /
i
) (71)
whereas, for the operators fullling the J
`
"' , J

"' , J
z
"' , (J`)"'
abnormal relationships, we have:
J
B
"'Y
J
X(b, c) =c
Y
Y
J
XY(b, c)
=JJ( J 1) <X(X1)Y
J
XY(b, c) (72)
J
z
"'
Y
J
X(b, c) =XY
J
X(b, c) ;
(J`)"'Y
J
X(b, c) =`J( J 1)Y
J
X(b, c) (73)
References
1 F. Gatti, C. Iung, M. Menou, Y. Justum and X. Chapuisat, J.
Chem. Phys., 1998, 108, 8804.
2 F. Gatti, C. Iung and X. Chapuisat, T HEOCHEM., 1998, 430,
201.
3 X. Chapuisat and C. Iung, Phys. Rev. A, 1992, 45, 6217.
4 B. Podolsky, Phys. Rev., 1928, 32, 812.
5 B. R. Johnson, J. Chem. Phys., 1983, 79, 1906.
6 R. T. Pack and G. Parker, J. Chem. Phys., 1987, 87, 3888.
7 R. T. Pack and G. Parker, J. Chem. Phys., 1989, 90, 3511.
8 X. Chapuisat and A. Nauts, Phys. Rev. A, 1991, 44, 1328.
9 Y. Justum, M. Menou, A. Nauts and X. Chapuisat, Chem. Phys.,
1997, 223, 211.
10 X. Chapuisat, A. Nauts and J.-P. Brunet, Mol. Phys., 1991, 72, 1.
11 M. J. Bramley, W. H. Green and N. C. Handy, Mol. Phys., 1991,
73, 1183.
3384 Phys. Chem. Chem. Phys., 1999, 1, 33773385
12 M. J. Bramley and N. C. Handy, J. Chem. Phys., 1993, 98, 1378.
13 A. G. Csazar and N. C. Handy, J. Chem. Phys., 1995, 102, 3962.
14 V. Aquilanti and S. Cavalli, J. Chem. Phys., 1986, 85, 1355.
15 V. Aquilanti and S. Cavalli, J. Chem. Soc., Faraday T rans., 1997,
93, 801.
16 G. Brocks, A. der Avoird, B. T. Sutclie and J. Tennyson, Mol.
Phys, 1983, 50, 1025.
17 N. Anderson and B. T. Sutclie, Int. J. Quantum Chem., 1996, 60,
37.
18 T. J. Lukka, J. Chem. Phys., 1995, 102, 3945.
19 A. Kuppermann, J. Phys. Chem. A, 1997, 101, 6368.
20 S. Rempe and R. Watts, J. Chem. Phys., 1999, 108, 10095.
21 E. B. Wilson, J. C. Decius and P. C. Cross, Molecular V ibrations,
McGraw-Hill, New York, 1955.
22 N. C. Handy, Mol. Phys., 1987, 61, 207.
23 D. W. Jepsen and J. O. Hirschfelder, Proc. Natl. Acad. Sci. USA,
1959, 45, 249.
24 A. Nauts and X. Chapuisat, Mol. Phys., 1985, 55, 1287.
25 F. Gatti, J. Chem. Phys., submitted.
26 R. N. Zare, Angular momentum, Wiley, New York, 1988.
27 J. H. Van Vleck, Rev. Mod. Phys, 1951, 23, 213.
28 A. Messiah, Mecanique quantique, Dunod, Paris, 1995.
29 X. Chapuisat, A. Belafhal, A. Nauts and C. Iung, Mol. Phys.,
1992, 77, 947.
30 F. Gatti, C. Iung, C. Leforestier, M. Menou, Y. Justum, A. Nauts
and X. Chapuisat, T HEOCHEM., 1998, 424, 181.
31 A. S. Davydov, Quantum Mechanics, Pergamon Press, Oxford,
1965.
32 L. C. Biendenharn and J. D. Louck, Angular momentum in
Quantum Mechanics, Addison-Wesley, Reading, MA, 1981.
33 G. B. Arfken and H. J. Weber, Mathematical methods for Physi-
cists, 4th edn., Academic Press, London, 1995.
34 S. Wolfram, Mathematica, a system for doing mathematics by com-
puter, 2nd edn. Addison-Wesley, Reading, MA, 1991.
35 C. Leforestier, L. B. Braly, K. Liu, M. Elrod and R. Saykally, J.
Chem. Phys., 1997, 106, 8527.
36 F. Gatti, C. Iung, C. Leforestier and X. Chapuisat, J. Chem.
Phys., in the press.
Paper 9/03466H
Phys. Chem. Chem. Phys., 1999, 1, 33773385 3385