JHU–TIPAC–2001-04
LBNL–48969
WSU–HEP–0101
hep-ph/0110317
October, 2001
SU (3) Breaking and D0 − D 0 Mixing
arXiv:hep-ph/0110317v2 14 Nov 2001
Adam F. Falk,1 Yuval Grossman,2 Zoltan Ligeti,3 and Alexey A. Petrov4
1
Department of Physics and Astronomy, The Johns Hopkins University
3400 North Charles Street, Baltimore, MD 21218
2
Department of Physics, Technion–Israel Institute of Technology
Technion City, 32000 Haifa, Israel
3
Ernest Orlando Lawrence Berkeley National Laboratory
University of California, Berkeley, CA 94720
4
Department of Physics and Astronomy
Wayne State University, Detroit, MI 48201
Abstract
The main challenge in the Standard Model calculation of the mass and width difference in the
0
D − D0 system is to estimate the size of SU (3) breaking effects. We prove that D meson mixing
occurs in the Standard Model only at second order in SU (3) violation. We consider the possibility
that phase space effects may be the dominant source of SU (3) breaking. We find that y = ∆Γ/2Γ
of the order of one percent is natural in the Standard Model, potentially reducing the sensitivity
to new physics of measurements of D meson mixing.
1
I.
INTRODUCTION
It is a common assertion that the Standard Model prediction for mixing in the D 0 − D 0
system is very small, making this process a sensitive probe of new physics. Two physical
parameters that characterize D 0 − D 0 mixing are
x≡
∆M
,
Γ
y≡
∆Γ
,
2Γ
(1)
where ∆M and ∆Γ are the mass and width differences of the two neutral D meson mass
eigenstates, and Γ is their average width. The D 0 − D 0 system is unique among the neutral
mesons in that it is the only one whose mixing proceeds via intermediate states with downtype quarks. The mixing is very slow in the Standard Model, because the third generation
plays a negligible role due to the smallness of |Vub Vcb| and the relative smallness of mb , and
so the GIM cancellation is very effective [1, 2, 3, 4, 5].
The current experimental upper bounds on x and y are on the order of a few times
10−2 , and are expected to improve significantly in the coming years. To regard a future
discovery of nonzero x or y as a signal for new physics, we would need high confidence that
the Standard Model predictions lie significantly below the present limits. As we will show,
in the Standard Model x and y are generated only at second order in SU(3) breaking, so
schematically
x , y ∼ sin2 θC × [SU(3) breaking]2 ,
(2)
where θC is the Cabibbo angle. Therefore, predicting the Standard Model values of x and
y depends crucially on estimating the size of SU(3) breaking. Although y is expected to
be determined by Standard Model processes, its value nevertheless affects significantly the
sensitivity to new physics of experimental analyses of D mixing [6].
At present, there are three types of experiments which measure x and y. Each is actually
sensitive to a combination of x and y, rather than to either quantity directly. First, there
is the D 0 lifetime difference to CP even and CP odd final states [7, 8, 9, 10, 11], which to
leading order measures
yCP =
τ (D → π + K − )
Am
− 1 = y cos φ − x sin φ
,
+
−
τ (D → K K )
2
(3)
where Am = |q/p|2 − 1 (see Eq. (5) for the definition of the neutral D mass eigenstates), and
φ is a possible CP violating phase of the mixing amplitude. Second, one can measure the
time dependence of doubly Cabibbo suppressed decays, such as D 0 → K + π − [12], which is
sensitive to the three quantities
(x cos δ + y sin δ) cos φ ,
(y cos δ − x sin δ) sin φ ,
x2 + y 2 ,
(4)
where δ is the strong phase between the Cabibbo allowed and doubly Cabibbo suppressed
amplitudes. A similar study for D 0 → K − π + π 0 also would be valuable, with the strong
phase difference extracted simultaneously from the Dalitz plot analysis [13]. Third, one can
search for D mixing in semileptonic decays [14], which is sensitive to x2 + y 2.
In a large class of models, the best hope to discover new physics in D mixing is to observe
the CP violating phase, φ12 = arg [M12 /Γ12 ] (see the definitions (7) and (8) below), which
is very small in the Standard Model. However, if y ≫ x, then the sensitivity of any physical
observable to φ12 is suppressed, since Am is proportional to x/y and φ is to (x/y)2, even
2
if new physics makes a large contribution to ∆M [6]. It is also clear from Eq. (4) that if
y is significantly larger than x, then δ must be known very precisely for experiments to be
sensitive to new physics in the terms linear in x and y. It may be possible to measure δ
with some accuracy at the planned τ -charm factory CLEO-c [15, 16].
There is a vast literature on estimating x and y within and beyond the Standard Model;
for a compilation of results, see Ref. [17]. Roughly, there are two approaches, neither of
which give very reliable results because mc is in some sense intermediate between heavy and
light. The “inclusive” approach is based on the operator product expansion (OPE). In the
mc ≫ Λ limit, where Λ is a scale characteristic of the strong interactions, ∆M and ∆Γ can
be expanded in terms of matrix elements of local operators [1, 2, 18]. Such calculations yield
< 10−3 . The use of the OPE relies on local quark-hadron duality, and on Λ/mc being
x, y ∼
small enough to allow a truncation of the series after the first few terms. The charm mass
may not be large enough for these to be good approximations, especially for nonleptonic D
decays. An observation of y of order 10−2 could be ascribed to a breakdown of the OPE
or of duality [18], but such a large value of y is certainly not a generic prediction of OPE
analyses. The “exclusive” approach sums over intermediate hadronic states, which may be
modeled or fit to experimental data [5, 19, 20]. Since there are cancellations between states
within a given SU(3) multiplet, one needs to know the contribution of each state with high
precision. However, the D is not light enough that its decays are dominated by a few final
states. In the absence of sufficiently precise data on many decay rates and on strong phases,
< 10−3 , Refs. [21, 22, 23]
one is forced to use some assumptions. While most studies find x, y ∼
−2
obtain x and y at the 10 level by arguing that SU(3) violation is actually of order unity,
but the source of the large SU(3) breaking is not made explicit.
In this paper, we compute the contribution to ∆Γ from SU(3) breaking from final state
phase space differences. This is a calculable source of SU(3) violation, which enhances the
rates to final states containing fewer strange quarks. In Sec. II we review the formalism of
D 0 − D 0 mixing. In Sec. III we give a general group theory proof that ∆M and ∆Γ are
only generated at second order in SU(3) breaking if SU(3) violation enters these quantities
perturbatively. In Sec. IV we discuss the estimates of SU(3) breaking using the “inclusive”
and “exclusive” analyses, and remind the reader of the shortcomings of each. Our main
results are found in Sec. V, namely the calculation of SU(3) breaking in ∆Γ from phase
space effects in two-, three- and four-body final states. We find that such effects are very
important, and can naturally account for ∆Γ/2Γ at the percent level. We extend the analysis
to intermediate resonances in Sec. VI. In Sec. VII we present our conclusions and ask whether
in light of our results it remains possible for the measurement of D mixing to probe new
physics.
II.
FORMALISM
We begin by reviewing the formalism for D 0 − D 0 mixing. The mass eigenstates DL and
DS are superpositions of the flavor eigenstates D 0 and D 0 ,
|DL,S i = p |D 0 i ± q |D0 i ,
(5)
where |p|2 + |q|2 = 1. In the Standard Model CP violation in D mixing is negligible, as is
CP violation in D decays both in the Standard Model and in most scenarios of new physics.
From here on we will assume that CP is a good symmetry. Then p = q, and |DL,S i become
3
CP eigenstates,
CP |D± i = ±|D± i ,
(6)
with the mass and width differences defined as ∆M ≡ mD+ − mD− and ∆Γ ≡ ΓD+ − ΓD− .
The off-diagonal element of the D 0 − D 0 mass matrix can be expressed as
M12 = hD 0 |Hw∆C=2 |D 0 i + P
Γ12 =
X
n
X
n
hD 0 |Hw∆C=1 |nihn|Hw∆C=1 |D 0 i
,
m2D − En2
(7)
ρn hD 0 |Hw∆C=1 |nihn|Hw∆C=1 |D 0i ,
(8)
where the sum is over all intermediate states, P denotes the principal value, and ρn is the
density of the state n. The first term in Eq. (7) comes from the local |∆C| = 2 operators
(box and dipenguin), which affect M12 only. The second term comes from the insertion of
two |∆C| = 1 operators. There is a contribution of this type to both M12 and Γ12 .
One can then express y in two equivalent ways, either as a sum over the states that are
common to D 0 and D0 ,
y=
i
1 X h 0
ρn hD |Hw |nihn|Hw |D0 i + hD0 |Hw |nihn|Hw |D 0 i ,
2Γ n
(9)
or as the difference in the decay rates of the two mass eigenstates
y=
i
1 X h
ρn |hD+ |Hw |ni|2 − |hD− |Hw |ni|2 .
2Γ n
(10)
A similar pair of expressions can be written for x,
X hD 0 |Hw |nihn|Hw |D0 i + hD0 |Hw |nihn|Hw |D 0 i
1
x =
hD 0 |Hw |D 0 i + P
,
Γ
m2D − En2
n
"
#
X |hD+ |Hw |ni|2 − |hD− |Hw |ni|2
1
=
.
hD 0 |Hw |D 0 i + P
Γ
m2D − En2
n
#
"
(11)
Note that x and y are generated by off-shell and on-shell intermediate states, respectively.
III.
SU (3) ANALYSIS OF D 0 − D0 MIXING
We now prove that D 0 − D 0 mixing arises only at second order in SU(3) breaking effects.
The proof is valid when SU(3) violation enters perturbatively. This would not be the case,
for example, if D transitions were dominated by intermediate states or single resonances
close to threshold. As we will see explicitly in Secs. V and VI, in such cases it is sometimes
possible for SU(3) violation to be enhanced substantially. Yet other than in these exceptional
situations, treating SU(3) violation perturbatively seems to us to be a mild assumption.
The quantities M12 and Γ12 which determine x and y depend on matrix elements with
the general structure
hD0 | Hw Hw |D 0 i ,
(12)
4
where in this section we let Hw denote specifically the ∆C = −1 part of the weak Hamiltonian. Let D be the field operator that creates a D 0 meson and annihilates a D 0 . Then the
matrix element may be written as
h0| D Hw Hw D |0i .
(13)
Let us focus on the SU(3) flavor group theory properties of this expression.
Since the operator D is of the form c̄u, it transforms in the fundamental representation
of SU(3), which we will represent with a lower index, Di . We use a convention in which the
correspondence between matrix indices and quark flavors is (1, 2, 3) = (u, d, s). The only
nonzero element of Di is D1 = 1. The ∆C = −1 part of the weak Hamiltonian has the flavor
structure (q̄i c)(q̄j qk ), so its matrix representation is written with a fundamental index and
two antifundamentals, Hkij . This operator is a sum of irreducible representations contained
in the product 3 × 3 × 3 = 15 + 6 + 3 + 3. In the limit in which the third generation is
neglected, Hkij is traceless, so only the 15 (symmetric on i and j) and 6 (antisymmetric on
i and j) representations appear. That is, the ∆C = −1 part of Hw may be decomposed as
1
(O15 + O6 ), where
2
¯
¯
+ s1 (ūc)(dd)
O15 = (s̄c)(ūd) + (ūc)(s̄d) + s1 (dc)(ūd)
¯
¯ ,
− s1 (s̄c)(ūs) − s1 (ūc)(s̄s) − s21 (dc)(ūs)
− s21 (ūc)(ds)
¯
¯
O6 = (s̄c)(ūd) − (ūc)(s̄d) + s1 (dc)(ūd)
− s1 (ūc)(dd)
¯ ,
¯
+ s21 (ūc)(ds)
− s1 (s̄c)(ūs) + s1 (ūc)(s̄s) − s21 (dc)(ūs)
(14)
ij
and s1 = sin θC ≈ 0.22. The matrix representations H(15)ij
k and H(6)k have nonzero
elements
H(15)ij
H213 = H231 = 1 ,
H212 = H221 = s1 ,
k :
H313 = H331 = −s1 ,
H312 = H321 = −s21 ,
(15)
13
31
12
21
H(6)ij
:
H
=
−H
=
1
,
H
=
−H
=
s
,
1
2
2
2
2
k
H313 = −H331 = −s1 ,
H312 = −H321 = −s21 .
We introduce SU(3) breaking through the quark mass operator M, whose matrix representation is Mji = diag(mu , md , ms ). Although M is a linear combination of the adjoint and
singlet representations, only the 8 induces SU(3) violating effects. It is convenient to set
mu = md = 0 and let ms 6= 0 be the only SU(3) violating parameter. All nonzero matrix
elements built out of Di , Hkij and Mji must be SU(3) singlets.
We now prove that D 0 − D 0 mixing arises only at second order in SU(3) violation, by
which we mean second order in ms . First, we note that the pair of D operators is symmetric,
and so the product Di Dj transforms as a 6 under SU(3). Second, the pair of Hw ’s is also
symmetric, and the product Hkij Hnlm is in one of the representations which appears in the
product
h
i
(15 + 6) × (15 + 6)
S
= (15 × 15)S + (15 × 6) + (6 × 6)S
′
(16)
′
= (60 + 24 + 15 + 15 + 6) + (42 + 24 + 15 + 6 + 3) + (15 + 6) .
A straightforward computation shows that only three of these representations actually appear in the decomposition of Hw Hw . They are the 60, the 42, and the 15′ (actually twice,
but with the same nonzero elements both times). So we have product operators of the form
DD = D6 ,
Hw Hw = O60 + O42 + O15′ ,
5
(17)
where the subscript denotes the representation of SU(3).
Since there is no 6 in the decomposition of Hw Hw , there is no SU(3) singlet which can
be made with D6 , and no SU(3) invariant matrix element of the form (13) can be formed.
This is the well known result that D 0 − D 0 mixing is prohibited by SU(3) symmetry.
Now consider a single insertion of the SU(3) violating spurion M. The combination
D6 M transforms as 6 × 8 = 24 + 15 + 6 + 3. Note that there is still no invariant to be made
with Hw Hw . It follows that D 0 − D 0 mixing is not induced at first order in SU(3) breaking.
With two insertions of M, it becomes possible to make an SU(3) invariant. The decomposition of DMM is
6 × (8 × 8)S = 6 × (27 + 8 + 1)
′
= (60 + 42 + 24 + 15 + 15 + 6) + (24 + 15 + 6 + 3) + 6 .
(18)
There are three elements of the 6 × 27 part which can give invariants with Hw Hw . Each
invariant yields a contribution proportional to s21 m2s . As promised, D 0 − D 0 mixing arises
only at second order in the SU(3) violating parameter ms .
IV.
ESTIMATING THE SIZE OF SU (3) BREAKING
We now turn to review some general estimates of the size of SU(3) breaking effects.
These effects can be approached from either an inclusive or an exclusive point of view. It is
instructive to see how SU(3) violation appears in each case.
A.
“Inclusive” approach
An elegant and concrete estimate of how SU(3) violation enters x and y is the short
distance analysis, first applied to D 0 − D 0 mixing by Georgi [1] and later extended by other
authors [2, 18]. We review it briefly, both to establish the contrast with our approach and
to recall the results. Let Λ be a scale characteristic of the strong interactions, such as
mρ or 4πfπ . In the limit mc ≫ Λ, the momentum flowing through the light degrees of
freedom in the intermediate state is large and an operator production expansion (OPE) can
be performed. For example, one can write
Z
n
o
1
0
Γ12 =
Im hD | i d4 x T Hw∆C=1 (x) Hw∆C=1 (0) |D 0i ,
(19)
2mD
where Hw∆C=1 is the |∆C| = 1 effective Hamiltonian. In the OPE, the time ordered product
in Eq. (19) can be expanded in local operators of increasing dimension; the higher dimension
operators are suppressed by powers of Λ/mc .
The leading contribution comes from the dimension-6 |∆C| = 2 four-quark operators
corresponding to the short distance box diagram,
O1 = ūα γµ PL cα ūβ γµ PL cβ ,
O2 = ūα γµ PL cβ ūβ γµ PL cα ,
O1′ = ūα PL cα ūβ PL cβ ,
O2′ = ūα PL cβ ūβ PL cα ,
(20)
where PL = 12 (1 − γ5 ). If one neglects QCD running between MW and mc , in which case O2
and O2′ do not contribute, one finds the simple expressions
∆Mbox
′
5 BD
(m2s − m2d )2
m2D
2
1
−
,
X
=
D
3π 2
m2c
4 BD (mc + mu )2
"
6
#
(21)
ratio
4-quark
6-quark
8-quark
∆M/∆Mbox
∆Γ/∆M
1
2
ms /m2c
Λ2 /ms mc
αs /4π
(αs /4π)(Λ2 /ms mc )2
β0 αs /4π
TABLE I: The enhancement of ∆M and ∆Γ relative to the box diagram at various orders in the
OPE. Λ denotes a hadronic scale around 4πfπ ∼ 1 GeV.
∆Γbox
′
4
5 BD
(m2s − m2d )2 m2s + m2d
m2D
=
1−
XD
.
3π
m2c
m2c
2 BD (mc + mu )2
"
(′)
#
(22)
(′)
where XD = Vcs2 Vcd2 G2F mD BD fD2 , and BD are bag factors for O1 , normalized to one in
vacuum saturation. Including leading logarithmic QCD effects enhances this estimate of
∆Γ by approximately a factor of two [24]. Eqs. (21) and (22) then lead to the estimates
xbox ∼ few × 10−5 ,
ybox ∼ few × 10−7 .
(23)
Neglecting md /ms , Eq. (22) is proportional to m6s . This factor comes from three sources:
(i) m2s from an SU(3) violating mass insertion on each quark line in the box graph; (ii)
m2s from an additional mass insertion on each line to compensate the chirality flip from
the first insertion; (iii) m2s to lift the helicity suppression for the decay of a scalar meson
into a massless fermion pair. The last factor of m2s is absent from Eq. (21) for ∆M; this
is why at leading order in the OPE ybox ≪ xbox . Higher order terms in the OPE are
important, because the chiral suppressions can be lifted by quark condensates instead of
by mass insertions, allowing ∆M and ∆Γ to be proportional to m2s . This is the minimal
suppression required by SU(3) symmetry, as we proved in Sec. III.
The order of magnitudes of the resulting contributions are summarized in Table I. In
the first line, the contributions to ∆M are normalized to ∆Mbox ; in the second line, the
contributions to ∆Γ are normalized to ∆M at each order. The contribution of 6-quark
operators to ∆M is enhanced compared to the 4-quark operators by Λ2 /mc ms . This can
be as much as an order of magnitude, if we identify the hadronic scale Λ as 4πfπ [25].
The second chiral suppression can also be lifted, but only at the price of adding a hard
gluon, so the contribution of 8-quark operators to ∆M compared to the 6-quark operators is
(αs /4π)(Λ2/mc ms ), which is of order unity.1 In the case of ∆Γ, higher dimension operators
are even more important [18]. A 6-quark operator, including a hard gluon to give an on-shell
intermediate state, lifts both a chiral suppression and the helicity suppression. The 8-quark
operators require a second intermediate particle to contribute to ∆Γ, which can be obtained
by splitting the gluon already present for ∆M into a quark pair [18], only costing a factor of
β0 αs /(4π) ∼ 1, where β0 = 11 − 23 nf = 9. Thus, the dominant contributions to x are from
6- and 8-quark operators, while the dominant contribution to y is from 8-quark operators.
With some assumptions about the hadronic matrix elements, the resulting estimates are
< 10−3 .
x∼y∼
1
(24)
We disagree with Ref. [18], in which it was claimed that x and y can arise at first order in ms . Such
contributions were claimed to come from pseudogoldstone loops which diverge in the infrared. However,
there are no such divergences because the π, K and η are coupled derivatively. Such a contribution would
also be in conflict with our proof in Sec. III that D mixing is second order in SU (3) violating effects.
7
> y. We emphasize that at this time
It is a general feature of OPE based analyses that x ∼
these methods are useful for understanding the order of magnitude of x and y, but not for
obtaining reliable quantitative results. For example, to turn the estimates presented here
into a systematic computation of x and y would require the calculation of almost two dozen
nonperturbative matrix elements.
B.
“Exclusive” approach
A long distance analysis of D mixing is complementary to the OPE. Instead of assuming
that the D meson is heavy enough for duality to hold between the partonic rate and the sum
over hadronic final states, here one assumes that D transitions are dominated by a small
number of exclusive processes, which are examined explicitly. This is particularly interesting
for studying ∆Γ, which depends on real final states in D decays.
For a long distance analysis, it is useful to express the width difference directly in terms
of observable decay rates. From Eq. (9), we find
y=
X
ηCKM (n) ηCP (n) cos δn
n
q
B(D 0 → n) B(D0 → n) ,
(25)
where δn is the strong phase difference between the D 0 → n and D0 → n amplitudes. In
decays to many-body final states, the strong phases may have different values in different
regions of the Dalitz plot, in which case the sum is supplemented by an integral over the
Dalitz plot for each final state. The CKM factor is ηCKM = (−1)ns , where ns is the number of
s and s̄ quarks in the final state. For example, ηCKM (K + K − ) = +1 and ηCKM (K + π − ) = −1.
The factor ηCP = ±1 is determined by the CP transformation of the final state, CP |f i =
¯ which is well-defined since |f i and |fi
¯ are in the same SU(3) multiplet. This factor
ηCP |fi,
is the same for the whole multiplet. For example, ηCP = +1 for the decays to K + K − ,
and therefore to all decays into two pseudoscalars. For states where different partial waves
contribute with different CP parities, ηCP is determined separately for each partial wave.
For example, ηCP (ρ+ ρ− ) = +1 for ρ+ ρ− in a relative s or d wave, and −1 in a p wave.
Finally, it is convenient to assemble the final states into SU(3) multiplets and write
y=
X
a
ya ,
ya = ηCP (a)
X
ηCKM (n) cos δn
n∈a
q
B(D 0 → n) B(D0 → n) ,
(26)
where a indexes complete SU(3) multiplets. By multiplets we refer to the SU(3) representation of the entire final state, not of the individual mesons and baryons.
In practice, we cannot use Eq. (26) to get a reliable estimate of y, since the doubly Cabibbo
suppressed rates have large errors, and there are very little data on strong phase differences.
To proceed further, we would be forced to introduce model dependent assumptions about
the amplitudes and/or their strong phases. For example, in two-body D decays to charged
pseudoscalars (π + π − , π + K − , K + π − , K + K − ), the SU(3) violation can enter through the
decay rates or the strong phase difference. We know experimentally that in some of these
rates the SU(3) breaking is sizable; for example B(D 0 → K + K − )/B(D 0 → π + π − ) ≃ 2.8 [26].
Such effects were the basis for the claim in Ref. [21] that SU(3) is simply inapplicable to
D decays. In contrast, we know very little about the strong phase δ which vanishes in the
> 0.8, but it is also
SU(3) limit; Ref. [27] presented a model calculation resulting in cos δ ∼
8
possible to obtain much larger values for δ [22]. Using Eq. (26), the value of ya corresponding
to the U-spin doublet of charged π and K is
0
+ −
0
+
−
q
B(D 0 → K − π + ) B(D 0 → K + π − ) .
(27)
The experimental central values, allowing for D mixing in the doubly Cabibbo suppressed
rates, yield yπK ≃ (5.76 − 5.29 cos δ) × 10−3 [6]. For small δ there is an almost perfect
cancellation even though the ratios of the individual rates significantly violate SU(3). In
the “exclusive” approach, x is obtained from y by use of a dispersion relation, and one
generally finds x ∼ y.
At this stage, one cannot use the exclusive approach to predict either x or y. Any
estimate of their sizes depends on computing SU(3) breaking effects. While this problem
is not tractable in general, one source of SU(3) breaking in y, from final state phase space,
can be calculated with only minimal and reasonable assumptions. We will estimate these
effects in the next section.
yπK = B(D → π π ) + B(D → K K ) − 2 cos δ
V.
SU (3) BREAKING FROM PHASE SPACE
We now turn to the contributions to y from on-shell final states. There is a contribution
to the D 0 width difference from every common decay product of D 0 and D0 . In the SU(3)
limit, these contributions cancel when one sums over complete SU(3) multiplets in the final
state. The cancellations depend on SU(3) symmetry both in the decay matrix elements and
in the final state phase space. While there are certainly SU(3) violating corrections to both
of these, it is extremely difficult to compute the SU(3) violation in the matrix elements in a
model independent manner.2 However, with some mild assumptions about the momentum
dependence of the matrix elements, the SU(3) violation in the phase space depends only on
the final particle masses and can be computed. In this section we estimate the contributions
to y solely from SU(3) violation in the phase space.3 We will find that this source of SU(3)
violation can generate y of the order of a percent.
The mixing parameter y may be written in terms of the matrix elements for common
final states for D 0 and D0 decays,
1X
y=
[P.S.]n hD0 | Hw |nihn| Hw |D 0 i ,
Γ n
Z
(28)
where the sum is over distinct final states n and the integral is over the phase space for state
n. Let us now perform the phase space integrals and restrict the sum to final states F which
transform within a single SU(3) multiplet R. The result is a contribution to y of the form
X
1
hD0 | Hw ηCP (FR )
|niρn hn| Hw |D 0 i ,
Γ
n∈FR
2
3
(29)
The SU (3) breaking in matrix elements may be modest even in cases such as D → K + K − and D → π + π − ,
for which the ratio of measured rates appears to be very far from the SU (3) limit [28].
The phase space difference alone can explain the large SU (3) breaking between the measured D → K ∗ ℓν̄
and D → ρℓν̄ rates, assuming no SU (3) breaking in the form factors [29]. Recently it was shown that the
lifetime ratio of the Ds and D0 mesons may also be explained this way [30].
9
where ρn is the phase space available to the state n. In the SU(3) limit, all the ρn are the
same for n ∈ FR , and the quantity in braces above is an SU(3) singlet. Since the ρn depend
only on the known masses of the particles in the state n, incorporating the true values of ρn
in the sum is a calculable source of SU(3) breaking.
This method does not lead directly to a calculable contribution to y, because the matrix
elements hn|Hw |D 0 i and hD 0 |Hw |ni are not known. However, CP symmetry, which in the
Standard Model and almost all scenarios of new physics is to an excellent approximation
conserved in D decays, relates hD0 |Hw |ni to hD 0|Hw |ni. Since |ni and |ni are in a common
SU(3) multiplet, they are determined by a single effective Hamiltonian. Hence the ratio
yF,R
| Hw |niρn hn|Hw |D 0 i
=
=P
0
0
n∈FR hD | Hw |niρn hn|Hw |D i
n∈FR hD
P
0
| Hw |niρn hn|Hw |D 0 i
P
0
n∈FR Γ(D → n)
n∈FR hD
P
0
(30)
is calculable, and represents the value which y would take if elements of FR were the only
channel open for D 0 decay. To get a true contribution to y, one must scale yF,R to the total
branching ratio to all the states in FR . This is not trivial, since a given physical final state
typically decomposes into a sum over more than one multiplet FR . The numerator of yF,R
is of order s21 while the denominator is of order 1, so with large SU(3) breaking in the phase
space the natural size of yF,R is 5%.
In this analysis, phase space is the only source of SU(3) violation which we will include.
Of course, there are other SU(3) violating effects, such as in matrix elements and final state
interaction phases. The purpose of our calculation is to explore the rough size of SU(3)
violation in exclusive contributions to y. We assume that there is no cancellation with other
sources of SU(3) breaking, or between the various multiplets which occur in D decay, that
would reduce our result for y by an order of magnitude. This is equivalent to assuming that
the D meson is not heavy enough that duality can be expected to enforce such cancellations.
We begin by computing yF,R for D decays to states F = P P consisting of a pair of
pseudoscalar mesons such as π, K, η. We neglect η −η ′ mixing throughout this analysis, and
we have checked that this simplification has a negligible effect on the numerical results. Since
P P is symmetric in the two mesons, it must transform as an element of (8 ×8)S = 27 + 8 + 1.
In principle, there are three possible amplitudes for D 0 → P P , one with the pair in a 27
and Hw in a 15,
ij
A27 (P P27 )km
(31)
ij Hk Dm ,
one with the pair in an 8 and and Hw in a 15,
k ij
A15
8 (P P8 )i Hk Dj ,
(32)
and one with the pair in an 8 and and Hw in a 6,
A68 (P P8 )ki Hkij Dj .
(33)
However, the product Hkij Dj with (ij) symmetric (the 15) is proportional to Hkij Dj with
6
(ij) antisymmetric (the 6), and the linear combination A8 ≡ A15
8 − A8 is the only one which
appears. Thus there are effectively two invariant amplitudes. There is no SU(3) invariant
amplitude to produce the final state in an singlet. Note that since we are assuming SU(3)
symmetry in the matrix elements, such final states do not appear in our analysis.
10
It is straightforward to use these invariants in Eq. (30) to compute yF,R . As an example,
for yP P,8 we obtain
1
1
1
1
Φ(η, η) + Φ(π 0 , π 0 ) + Φ(η, π 0 ) + Φ(π + , π − ) + Φ(K + , K − ) − Φ(η, K 0 )
2
2
3
6
1
1
1
0
+
−
−
+
0
0
0
0
− Φ(η, K ) − Φ(K , π ) − Φ(K , π ) − Φ(K , π ) − Φ(K , π )
6
2
2
−1
1
1
,
(34)
×
Φ(η, K 0 ) + Φ(K − , π + ) + Φ(K 0 , π 0 ) + O(s21 )
6
2
yP P,8 = s21
where Φ(P1 , P2 ) is the phase space integral for the decay into mesons P1 and P2 . In a twobody decay, Φ(P1 , P2 ) is proportional to |~p |2ℓ+1 , where ~p and ℓ are the spatial momentum
and orbital angular momentum of the final state particles. For D 0 → P P , the decay is into
an s wave. It is straightforward to compute the required ratios from the known pseudoscalar
masses,
yP P,27 = −0.00071 s21 = −3.4 × 10−5 .
yP P,8 = −0.0038 s21 = −1.8 × 10−4 ,
(35)
These effects are no larger than one finds in the inclusive analysis. This is not surprising,
since as in the parton picture, the final states are far from threshold.
Next we turn to final states of the form P V , consisting of a pseudoscalar and a vector
meson. Note that three-body final states 3P can resonate through P V , and so are partially
included here. In this case there is no symmetry between the mesons, so in principle all
representations in the combination 8 × 8 = 27 + 10 + 10 + 8S + 8A + 1 can appear. √
For
¯
2,
simplicity, we take the quark content of the φ and ω respectively to be s̄s and (ūu + dd)/
and consider only the combination which appears in the SU(3) octet. We have checked that
reasonable variations of the φ − ω mixing angle have a negligible effect on our numerical
results. For each representation, there is a single invariant, up to the same degeneracy for
the 8 as in the P P case. Along with the analogues of Eqs. (31)–(33) with coefficients B27
and B8 ≡ B815 − B86 , we have the new invariants
for Hw in a 15, and
ij
B10 (P V10 )ijk Hm
Dn ǫkmn
(36)
B10 (P V10 )ijk Hilm Dj ǫklm
(37)
for Hw in a 6. It turns out that these two invariants are proportional to each other. As
before, the SU(3) singlet final state is not produced.
Both because one of the particles is more massive, and because the decay is now into a
p wave, the phase space dependence is stronger than for the P P final state. We obtain the
ratios
yP V,8S = 0.031 s21 = 0.15 × 10−2 ,
yP V,10 = 0.020 s21 = 0.10 × 10−2 ,
yP V,27 = 0.040 s21 = 0.19 × 10−2 .
yP V,8A = 0.032 s21 = 0.15 × 10−2 ,
yP V,10 = 0.016 s21 = 0.08 × 10−2 ,
(38)
For any representation of the final state, the effects are less than one percent.
For the V V final state, decays into s, p and d waves are all possible. Bose symmetry
and the restriction to zero total angular momentum together imply that only the symmetric
11
Final state representation
PP
PV
(V V )s-wave
(V V )p-wave
(V V )d-wave
8
27
8S
8A
10
10
27
8
27
8
27
8
27
yF,R /s21
yF,R (%)
−0.0038
−0.00071
0.031
0.032
0.020
0.016
0.040
−0.081
−0.061
−0.10
−0.14
0.51
0.57
−0.018
−0.0034
0.15
0.15
0.10
0.08
0.19
−0.39
−0.30
−0.48
−0.70
2.5
2.8
TABLE II: Values of yF,R for two-body final states. This represents the value which y would take
if elements of FR were the only channel open for D 0 decay.
SU(3) combinations appear. Because some V V final states, such as φK ∗ , lie near the D
threshold, the inclusion of vector meson widths is quite important. Our model for the
resonance line shape is a Lorentz invariant Breit-Wigner normalized on 0 ≤ m < ∞,
m2 Γ2R
f (m; mR , ΓR ) = N(mR , ΓR )
,
(m2 − m2R )2 + m2 Γ2R
(39)
where mR and ΓR are the mass and width of the vector meson, and m2 is the square of its
four-momentum in the decay. For s wave decays, we find the ratios
yV V,8 = −0.081 s21 = −0.39 × 10−2 ,
yV V,27 = −0.061 s21 = −0.30 × 10−2 ,
(40)
while for p wave decays we find
yV V,8 = −0.10 s21 = −0.48 × 10−2 ,
yV V,27 = −0.14 s21 = −0.70 × 10−2 ,
(41)
yV V,27 = 0.57 s21 = 2.8 × 10−2 .
(42)
and for d waves,
yV V,8 = 0.51 s21 = 2.5 × 10−2 ,
With these heavier final states and with the higher partial waves, we see that effects at
the level of a percent are quite generic. The vector meson widths turn out to be quite
important; if they were neglected, the results in the p- and d-wave channels would be larger
by approximately a factor of three. The finite widths soften the SU(3) breaking which
otherwise would be induced by a sharp phase space boundary. We have checked that our
results are not very sensitive to variations in the line shape used to model the vector meson
widths. Again, 4P and P P V final states can resonate through V V , so they are partially
included here. Our results for two-body final states are summarized in Table II.
As we go to final states with more particles, the combinatoric possibilities begin to proliferate. We will consider the final states 3P and 4P , and for concreteness require that
12
the pseudoscalars be found in a totally symmetric 8 or 27 representation of SU(3). This
assumption is convenient, because the phase space integration is much simpler if it can be
performed symmetrically. These final states should be representative; we have no reason to
believe that this choice selects final state multiplets for which phase space effects are particularly enhanced or suppressed. Note that 3 × (15 + 6) contains no representation larger
than a 27.
In contrast to the two-body case, for three-body final states the momentum dependence
of the matrix elements is no longer fixed by the conservation of angular momentum. The
simplest assumption is to take a momentum independent matrix element, with all three final
state particles in an s wave. In that case, we find
y3P,8 = −0.48 s21 = −2.3 × 10−2 ,
y3P,27 = −0.11 s21 = −0.54 × 10−2 .
(43)
Note that the SU(3) violation is smaller for the larger multiplets, as more final states enter
the sum. It may be that the 8 is in some sense an unusually small representation for three or
more particles, and that this mode enhances the SU(3) violation by providing fewer distinct
final states among which cancellations can occur. The enhancement of y3P,8 over y3P,27 is not
a peculiarity of s wave decays. We have also considered other matrix elements; for example,
if one of the mesons has angular momentum ℓ = 1 in the D 0 rest frame (balanced by the
combination of the other two mesons), then the ratios become
y3P,8 = −1.13 s21 = −5.5 × 10−2 ,
y3P,27 = −0.074 s21 = −0.36 × 10−2 .
(44)
Alternatively, we could imagine introducing a mild “form factor suppression,” with a weight
such as Πi6=j (1 − m2ij /Q2 )−1 , where m2ij = (pi + pj )2 , and Q = 2 GeV is a typical resonance
mass. The result then changes to
y3P,8 = −0.44 s21 = −2.1 × 10−2 ,
y3P,27 = −0.13 s21 = −0.64 × 10−2 .
(45)
Finally, we have studied the final state with four pseudoscalars, with the mesons in an
overall symmetric 8 or a symmetric 27. We take a momentum independent matrix element.
There are actually two symmetric 27 representations; we call the 27 the representation of
ij
i
the form Rkl
= [Mm
Mkm Mnj Mln + symmetric − traces] and the 27′ the one of the form
ij
i
Mnm Mkn Mlj + symmetric − traces]. Then we find
Rkl
= [Mm
y4P,8 = 3.3 s21 = 16 × 10−2 ,
y4P,27′ = 1.9 s21 = 9.2 × 10−2 .
y4P,27 = 2.2 s21 = 11 × 10−2 .
(46)
Here the partial contributions to y are very large, of the order of 10%. This is not surprising,
since 4P final states containing more than one strange particle are close to D threshold, and
the ones containing no pions are kinematically inaccessible. There is no reason to expect
SU(3) cancellations to persist effectively in this regime. Our results for 3P and 4P final
states are summarized in Table III.
Formally, one could construct y from the individual yF,R by weighting them by their D 0
branching ratios,
X
1X
y=
yF,R
Γ(D 0 → n) .
(47)
Γ F,R
n∈FR
However, the data on D decays are neither abundant nor precise enough to disentangle the
decays to the various SU(3) multiplets, especially for the three- and four-body final states.
13
Final state representation
(3P )s-wave
(3P )p-wave
(3P )form-factor
4P
8
27
8
27
8
27
8
27
27′
yF,R /s21
yF,R (%)
−0.48
−0.11
−1.13
−0.07
−0.44
−0.13
3.3
2.2
1.9
−2.3
−0.54
−5.5
−0.36
−2.1
−0.64
16
9.2
11
TABLE III: Values of yF,R for three- and four-body final states.
Final state
fraction
PP
PV
(V V )s-wave
(V V )d-wave
3P
4P
5%
10%
5%
5%
5%
10%
TABLE IV: Total D 0 branching fractions to classes of final states, rounded to nearest 5% [26].
Nor have we computed yF,R for all or even most of the available representations. Instead,
we can only estimate individual contributions to y by assuming that the representations
for which we know yF,R to be typical for final states with a given multiplicity, and then to
scale to the total branching ratio to those final states. The total branching ratios of D 0
to two-, three- and four-body final states can be extracted from Ref. [26]. The results are
presented in Table IV, where we round to the nearest 5% to emphasize the uncertainties in
these numbers. Close to half of all D 0 decays are accounted for in this table; the rest are
decays to other modes such as P P V , decays to states with SU(3) singlet mesons, decays
to higher resonances, semileptonic decays, and other suppressed processes. Based on data
in the channel K 0∗ ρ0 , the V V final state is dominantly CP even, consistent with an equal
distribution between s and d wave decays (although favoring a small s wave enhancement).
We estimate the contribution to y from a given type of final state by taking the product
of the typical yF,R found in our calculation with the approximate branching ratios given in
Table IV. Such estimates are necessarily crude, but they are sufficient to give a sense of
the order of magnitude of y which is to be expected. While in most cases the contributions
are small, of the order of 10−3 or less, we observe that D 0 decays to nonresonant 4P states
naturally contribute to y at the percent level. The reason for such unusually large SU(3)
violating effects in y is that approximately 10% of D 0 decays are to final states for which
the complete SU(3) multiplets are not kinematically accessible.
It should be noted that for D decays to final states so close to threshold, our argument
that D mixing is second order in SU(3) violation is inapplicable, because its underlying
14
assumption that SU(3) violation enters perturbatively is not met. In particular the proof
fails near D threshold, if the decay is either to weakly decaying final states or to hadrons
with widths Γ which are smaller than ms . In either case, the phase space available for
the decay can vary rapidly on the scale of ms , spoiling the analytic expansion. For decays
to hadronic resonances, Γ/ms is a small parameter which is not analytic as ms → 0. For
decays to long lived mesons, the Θ-functions which fix the boundaries of phase space are
not analytic functions of their arguments, which in turn depend on ms through the masses
of the final state hadrons. In this way, the generic m2s /m2D suppression is lifted and we find
larger SU(3) violation in y just at the point that the conditions of the proof are not satisfied.
We will see a similar failure of SU(3) cancellations when we study D mixing induced by
resonances in Sec. VI.
We have not considered all possible final states which might give large contributions to
y. In particular, the branching ratio for D 0 → K − a+
1 is (7.3 ± 1.1)% [26], even though this
final state is quite close to D threshold. Unfortunately, the identities of the SU(3) partners
of the a1 (1260), which has J P C = 1++ , are not well established. While it is natural to
identify the K1 (1400) as the corresponding strange axial vector meson, and the f1 (1285)
as the analogue of the ω, there is no natural candidate for the s̄s analogue of the φ. The
size of yP V ∗ is quite sensitive to this choice, as well as to the value taken for the poorly
measured width of the a1 . If we take the s̄s state to be the f1 (1420), and Γ(a1 ) = 400 MeV,
we find yP V ∗ ,8S = 1.8%. If instead we take the f1 (1510), we find yP V ∗ ,8S = 1.7%. With
Γ(a1 ) = 250 MeV, these numbers become 2.5% and 2.4%, respectively. Although it is clear
that percent level contributions to y are possible from SU(3) violation in this channel, the
data are still too poor to draw firm conclusions.
On the basis of this analysis, in particular as applied to the 4P final state, we would
conclude that y on the order of a percent would be completely natural. Anything an order
of magnitude smaller would require significant cancellations which do not appear naturally
in this framework. Cancellations would be expected only if they were enforced by the OPE,
that is, if the charm quark were heavy enough that the “inclusive” approach were applicable.
The hypothesis underlying the present analysis is that this is not the case.
VI.
SU (3) BREAKING FROM NEARBY RESONANCES
One interesting feature of the D 0 is that there are excited mesons with masses close to
mD . As a result, it would not be unnatural for K resonances to play an important role in
D decays. This possibility has already been explored in the literature [20, 27, 31, 32] . Here
we explore SU(3) breaking in the resonance contribution to D mixing.
We are interested in the process D 0 → R → D 0 , where R is a resonance with mass mR
and width ΓR . Only spin zero resonances are relevant. The contribution of a single state to
the D mass and width differences is given by
yRres = ηR
|HR |2
ΓR
,
2
2 2
Γ (mD − mR ) + m2D Γ2R
xres
R = ηR
2|HR |2
m2D − m2R
. (48)
Γ mD (m2D − m2R )2 + m2D Γ2R
where |HR |2 ≡ hD0 |HW |RihR|HW |D 0i parameterizes the couplings of R to D 0 and D 0 , and
ηR is the CP eigenvalue of the SU(3) multiplet within which the resonance resides. If we
assume the absence of direct CP violation in D decays, then hD0 |HW |Ri may be related to
15
hR|HW |D 0 i by SU(3). The ratio
xres
2(m2D − m2R )
R
=
,
yRres
mD ΓR
(49)
is independent of HR . Significant contributions to x and y from the resonance R are possible
< mD ΓR .
only if m2D − m2R ∼
As a concrete example, consider the K ∗ (1950), a positive parity excited kaon which,
because of its large width, may play an important role in mediating D 0 → K − π + . Fitting the
K ∗ (1950) contribution to the observed D → Kπ rates, one finds that resonance mediation
could be as large as the usual quark tree amplitude [27]. We can estimate an upper bound on
the contribution of K ∗ (1950) to y by assuming that the resonance is completely responsible
for D → Kπ. The limit is given by
|∆Γ|
hD0 |HW |KH ihKH |HW |D 0 i
B(D 0 → Kπ)
≤
×
,
Γ
hD 0 |HW |KH ihKH |HW |D 0 i B(KH → Kπ)
(50)
where we denote the K ∗ (1950) by KH . With B(D 0 → Kπ) ≃ 6% and B(KH → Kπ) ≃ 52%,
we find |y| ≤ 0.06 s21 ≃ 3 × 10−3 . If D mixing is mediated by a resonance, then we expect x
and y to be roughly of the same size.
This upper bound is too generous, because we have not included the suppression from
SU(3) cancellations. Note that our proof of Sec. III, that SU(3) violation appears only at
second order in ms , applies only so long as ms ≪ ΓR . While this must be true in the limit
mR ∼ mD → ∞, in which case ΓR scales as mc , the ratio ms /ΓR may not be small for
resonances near the physical D mass. Therefore, SU(3) cancellations may be less effective
for resonances than for real final states.
The resonances in question fall into a positive parity 8 of SU(3), consisting of states
which we will denote (πH , KH , ηH ). If these states were degenerate and had equal widths,
their contributions to D mixing would cancel. A measure of the actual effectiveness of this
cancellation is the contribution of the entire multiplet relative to that of the KH . The SU(3)
partners of the K ∗ (1950) have not been conclusively identified. Instead of speculating, we
will explore the efficiency of SU(3) cancellations qualitatively by taking the simple mass
relations
1
mπH = mKH − ms ,
mηH = mKH + ms ,
(51)
3
and assuming that the widths of the πH and ηH are the same as Γ(KH ) ≃ 200 MeV. Then
res
y res = yK
−
H
1 res 3 res
y − y .
4 πH 4 ηH
(52)
res
For ms = 150 MeV, we find y res /yK
= 0.27. The cancellations are somewhat less effective
H
res
res
res
for x , with x /xKH = 0.50. We see that even for the K ∗ (1950), likely to be the most
favorable for inducing a large effect, SU(3) cancellations reduce the contributions to xres
and y res . We conclude that it would be quite unlikely for resonances to make a contribution
to y at the level of one percent.
VII.
CONCLUSIONS
The motivation most often cited in searches for D 0 −D 0 mixing is the possibility of observing a signal from new physics which may dominate over the Standard Model contribution.
16
But to look for new physics in this way, one must be confident that the Standard Model
prediction does not already saturate the experimental bound. Previous analyses based on
< 10−3 , while naı̈ve estimates based
short distance expansions have consistently found x, y ∼
on known SU(3) breaking in charm decays allow an effect an order of magnitude larger.
Since current experimental sensitivity is at the level of a few percent, the difference is quite
important.
In this paper we have performed a general SU(3) analysis of the contributions to y. We
proved that if SU(3) violation may be treated perturbatively, then D 0 − D0 mixing in the
Standard Model is generated only at second order in SU(3) breaking effects. Within the
exclusive approach, we identified an SU(3) breaking effect, SU(3) violation in final state
phase space, which can be calculated with minimal model dependence. We found that phase
space effects alone can provide enough SU(3) breaking to induce y ∼ 10−2 . Large effects
in y appear for decays to final states close to D threshold, where an analytic expansion in
SU(3) violation is no longer possible.
We believe that this is an important result. Despite the large uncertainties, this is the first
model independent calculation to give y close to the present experimental bounds. While
some degree of cancellation is possible between different multiplets, as would be expected in
the mc → ∞ limit, or between SU(3) breaking in phase space and in matrix elements, it is
not known how effective these cancellations are. The most reasonable assumption in light
of our analysis is that they are not significant enough to result in an order of magnitude
suppression of y. Therefore, any future discovery of a D meson width difference should not
by itself be interpreted as an indication of the breakdown of the Standard Model.
However, our analysis does not amount to a Standard Model calculation of y. First, we
have considered only SU(3) breaking from phase space, and have not included any symmetry
breaking in the matrix elements. Second, we have not calculated the contributions from all
final states. Had we done so, we would still need very precise experimental data in order to
disentangle the various SU(3) multiplets and combine the results into an overall value of y.
Third, we have assumed that the charm quark is not heavy enough for duality to enforce
significant cancellations between the various nonleptonic D decay channels, although some
degree of cancellation is to be expected.
The implication of our results for the Standard Model prediction for x is less apparent.
> y, it is not clear
While analyses based on the “inclusive” approach generally yield x ∼
what the “exclusive” approach predicts. The effect of SU(3) breaking in phase space in x
is softer than in y, so one would expect x < y from our analysis. Thus if x > y is found
experimentally, it may still be an indication of a new physics contribution to x, even if y is
also large. On the other hand, if y > x then it will be hard to find signals of new physics, even
if such contributions dominate ∆M. The linear sensitivity to new physics in the analysis of
the time dependence of D 0 → K + π − is from x′ = x cos δ + y sin δ and y ′ = y cos δ − x sin δ
instead of x and y. If y > x, then δ would have to be known precisely for these terms to be
sensitive to new physics in x.
There remain large uncertainties in the Standard Model predictions of x and y, and
values near the current experimental bounds cannot be ruled out. Therefore, it will be
difficult to find a clear indication of physics beyond the Standard Model in D 0 − D 0 mixing
measurements. We believe that at this stage the only robust potential signal of new physics
in D 0 − D 0 mixing is CP violation, for which the Standard Model prediction is very small.
Unfortunately, if y is larger or much larger than x, then the observable CP violation in
D 0 − D0 mixing is necessarily small, even if new physics dominates x. Therefore, searching
17
for new physics and CP violation in D 0 − D 0 mixing should aim at precise measurements
of both x and y, and at more complicated analyses involving the extraction of the strong
phase in the time dependence of doubly Cabibbo suppressed decays.
Acknowledgments
It is a pleasure to acknowledge helpful discussions with Yossi Nir, Helen Quinn and
Martin Savage. We thank the Aspen Center for Physics for hospitality while portions of this
work were completed. A.F. was supported in part by the U.S. National Science Foundation
under Grant PHY–9970781, and is a Cottrell Scholar of the Research Corporation. Y.G. was
supported in part by the Israel Science Foundation under Grant No. 237/01-1, and by the
Technion V.P.R Fund – Harry Werksman Research Fund. Z.L. was supported in part by
the Director, Office of Science, Office of High Energy and Nuclear Physics, Division of High
Energy Physics, of the U.S. Department of Energy under Contract DE-AC03-76SF00098.
The work of Y.G. and Z.L. was also supported in part by the United States–Israel Binational
Science Foundation (BSF) through Grant No. 2000133. A.P. thanks the Cornell University
Theory Group, where part of this work was performed.
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
H. Georgi, Phys. Lett. B297, 353 (1992).
T. Ohl, G. Ricciardi and E. H. Simmons, Nucl. Phys. B403, 605 (1993).
A.A. Petrov, Phys. Rev. D56, 1685 (1997).
A. Datta and M. Khambakhar, Zeit. Phys. C27, 515 (1985).
J. Donoghue, E. Golowich, B. Holstein and J. Trampetic, Phys. Rev. D33, 179 (1986).
S. Bergmann, Y. Grossman, Z. Ligeti, Y. Nir and A.A. Petrov, Phys. Lett. B486, 418 (2000).
E.M. Aitala et al., E791 Collaboration, Phys. Rev. Lett. 83, 32 (1999).
J.M. Link et al., FOCUS Collaboration, Phys. Lett. B485, 62 (2000).
D. Cronin-Hennessy et al., CLEO Collaboration, hep-ex/0102006.
K. Abe et al., BELLE Collaboration, BELLE-CONF-0131.
B. Aubert et al., BaBar Collaboration, hep-ex/0109008.
R. Godang et al., CLEO Collaboration, Phys. Rev. Lett. 84, 5038 (2000).
G. Brandenburg et al., CLEO Collaboration, Phys. Rev. Lett. 87, 071802 (2001).
E.M. Aitala et al., E791 Collaboration, Phys. Rev. D57, 13 (1998).
J. P. Silva and A. Soffer, Phys. Rev. D61, 112001 (2000).
M. Gronau, Y. Grossman and J. L. Rosner, Phys. Lett. B508, 37 (2001).
H.N. Nelson, hep-ex/9908021.
I. Bigi and N. Uraltsev, Nucl. Phys. B592, 92 (2000).
F. Buccella, M. Lusignoli and A. Pugliese, Phys. Lett. B379, 249 (1996).
E. Golowich and A.A. Petrov, Phys. Lett. B427, 172 (1998).
L. Wolfenstein, Phys. Lett. B164, 170 (1985).
P. Colangelo, G. Nardulli and N. Paver, Phys. Lett. B242, 71 (1990).
T.A. Kaeding, Phys. Lett. B357, 151 (1995).
A.A. Petrov, hep-ph/0009160: E. Golowich, “CP violation in the charm sector,” Workshop
on CP Violation, Adelaide, Australia, July 3-8, 1998.
[25] A. Manohar and H. Georgi, Nucl. Phys. B 234, 189 (1984).
18
[26]
[27]
[28]
[29]
[30]
[31]
[32]
D.E. Groom et al., Particle Data Group, Eur. Phys. J. C15, 1 (2000).
A.F. Falk, Y. Nir and A.A. Petrov, JHEP 12, 019 (1999).
M.J. Savage, Phys. Lett. B257, 414 (1991).
Z. Ligeti, I.W. Stewart and M.B. Wise, Phys. Lett. B420, 359 (1998).
S. Nussinov and M.V. Purohit, hep-ph/0108272.
E. Golowich, Phys. Rev. D24, 676 (1981).
M. Gronau, Phys. Rev. Lett. 83, 4005 (1999).
19