Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
DOI: 10.1103/PhysRevD.95.115008
115008-2
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
pffiffiffiffiffiffiffiffiffiffiffiffiffi
c̄Γb heavy quark current onto HQET, cΓb → hDjc̄σ μν bjBi ¼ i mB mD ½hT ðv0μ vν − v0ν vμ Þ; ð10dÞ
⃖
cv0 ½Γ − iDΓ=ð2m ⃗
c Þ þ ΓiD=ð2m b Þ þ bv .
1
115008-3
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
ĥV ¼ 1 þ α̂s CV 1 þ εc ðL̂2 − L̂5 Þ þ εb ðL̂1 − L̂4 Þ; well below, as in the small-velocity limit [30,31]). In the
MS scheme fermions do not decouple for μ < m, intro-
w−1
ĥA1 ¼ 1 þ α̂s CA1 þ εc L̂2 − L̂5 ducing artificially large corrections in the running, com-
wþ1 pensated by corresponding spurious terms in the β function
w−1 computed without integrating out heavy quarks [32].
þ εb L̂1 − L̂4 ;
wþ1
ĥA2 ¼ α̂s CA2 þ εc ðL̂3 þ L̂6 Þ; C. Decay rates and form factor ratios
ĥA3 ¼ 1 þ α̂s ðCA1 þ CA3 Þ þ εc ðL̂2 − L̂3 þ L̂6 − L̂5 Þ The B → DðÞ lν̄ differential rates have the well-known
expressions in the SM,
þ εb ðL̂1 − L̂4 Þ;
ĥP ¼ 1 þ α̂s CP þ εc ½L̂2 þ L̂3 ðw − 1Þ þ L̂5 − L̂6 ðw þ 1Þ dΓðB → Dlν̄Þ G2F jV cb j2 η2EW m5B 2
¼ ðw − 1Þ3=2
þ εb ðL̂1 − L̂4 Þ; dw 48π 3
× r3D ð1 þ rD Þ2 GðwÞ2 ; ð18aÞ
w−1
ĥT 1 ¼ 1 þ α̂s CT 1 þ ðCT 2 − CT 3 Þ þ εc L̂2 þ εb L̂1 ;
2
dΓðB → D lν̄Þ G2F jV cb j2 η2EW m5B 2
wþ1 ¼ ðw − 1Þ1=2 ðw þ 1Þ2
ĥT 2 ¼ α̂s ðCT 2 þ CT 3 Þ þ εc L̂5 − εb L̂4 ; dw 48π 3
2
× r3D ð1 − rD Þ2
ĥT 3 ¼ α̂s CT 2 þ εc ðL̂6 − L̂3 Þ: ð15Þ
4w 1 − 2wrD þ r2D
× 1þ F ðwÞ2 ;
In Eqs. (14) and (15), the relations for the SM currents—that w þ 1 ð1 − rD Þ2
is, hþ , h− , hV , hA1 , hA2 , and hA3 —agree with the literature, ð18bÞ
e.g., Refs. [16,20]. Because of Luke’s theorem, the
OðΛQCD =mc;b Þ corrections to hþ , hS , hA1 , and hT 1 vanish where rDðÞ ¼ mDðÞ =mB and ηEW ≃ 1.0066 [33] is the
at zero recoil. To the best of our knowledge, the expressions electroweak correction. In addition,
for hT and hT 1;2;3 cannot be found in the literature. For hT 2 and
hT 3 , which start at order ΛQCD =mc;b, the partial results used 1 − rD
GðwÞ ¼ hþ − h ; ð19aÞ
in the literature (e.g., Ref. [28]) kept and left out terms, 1 þ rD −
which are both order OðΛQCD =mc;b Þ.
w−1
The scalar and vector matrix elements in B → D tran- F ðwÞ2 ¼ h2A1 2ð1 − 2wrD þ r2D Þ 1 þ R1
sitions, and the pseudoscalar and axial vector ones in wþ1
B → D , are related by the equations of motion
þ ½ð1 − rD Þ þ ðw − 1Þð1 − R2 Þ2
−1
½mb ðμÞ − mc ðμÞhDjcbjBi ¼ hDjcqbjBi; 4w
× ð1 − rD Þ2 þ ð1 − 2wrD þ r2D Þ ;
−½mb ðμÞ þ mc ðμÞhD jcγ 5 bjBi ¼ hD jcqγ 5 bjBi; ð16Þ wþ1
ð19bÞ
in which mQ ðμÞ are the MS quark masses at a common
scale μ, obeying and the form-factor ratios are defined as
115008-4
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
To include the lepton mass suppressed terms, one factorially growing terms, it is convenient to reorganize
sometimes defines [28,34] additional form factor ratios the perturbation series in terms of a suitable short-distance
mass scheme, instead of the pole mass. We use the 1S
hA3 − rD hA2 scheme [40–42], which has been tested in the calculations
R3 ðwÞ ¼ ;
hA1 of numerous observables. (Using the MS mass yields a
hA1 ðw þ 1Þ − hA3 ðw − rD Þ − hA2 ð1 − wrD Þ poorly behaved perturbation series, for the reasons men-
R0 ðwÞ ¼ : ð22Þ tioned at the end of Sec. II B. Other possible short-distance
ð1 þ rD ÞhA1 mass schemes include the potential-subtracted (PS) mass
[43] or the kinetic mass [44].)
All contributions of R0;3 ðwÞ are proportional to m2l . (The The 1S scheme defines m1S b as half of the perturbatively
authors of Ref. [34] defined R3 ¼ hA3 =hA1 .) They are not computed ϒð1SÞ mass. It is related to the pole mass
linearly independent from R1;2 ðwÞ, as there are only three as m1S 2
b ¼ mb ð1 − 2αs =9 þ Þ [40–42], so that we
form factor ratios in B → D lν̄ in the SM. In the heavy may treat the pole mass as the function mb ðm1S b Þ¼
quark limit, R3 ðwÞ ¼ R0 ðwÞ ¼ 1. At Oðεc;b ; αs Þ, the SM 1S 2
mb ð1 þ 2αs =9 þ Þ. Neglecting higher-order terms,
predictions are
as done throughout this paper, is a good approximation
2 in all cases where they are known, including the evalu-
R3 ðwÞ ¼ 1 þ α̂s ðCA3 − rD CA2 Þ − ðε L̂ þ εc L̂5 Þ ation of R1;2 [22]. We adopt the inputs [45],
wþ1 b 4
þ εc ½L̂6 ð1 − rD Þ − L̂3 ð1 þ rD Þ;
CA ðrD − wÞ − ð1 − rD wÞCA2 m1S
b ¼ ð4.71 0.05Þ GeV;
R0 ðwÞ ¼ 1 þ α̂s 3 δmbc ¼ mb − mc ¼ ð3.40 0.02Þ GeV; ð24Þ
1 þ rD
2ðw − rD Þ
þ ðε L̂ þ εc L̂5 Þ
ð1 þ rD Þð1 þ wÞ b 4 from fits to inclusive B → Xc lν̄ spectra and other
1 − rD determinations of m1Sb . We eliminate mc using mc ¼
þ εc L̂3 ðw − 1Þ − L̂6 ðw þ 1Þ : ð23Þ 1S
mb ðmb Þ − δmbc , and extract Λ̄ via
1 þ rD
115008-5
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
R3 ð1Þ ≃ 1.19 − 0.26ηð1Þ − 1.20χ̂ 2 ð1Þ; For B → Dlν̄, jz j ≤ 0.032. The unitarity constraints sug-
gest a form factor parametrization of the form
R0 ð1Þ ≃ 1.09 þ 0.25ηð1Þ;
R03 ð1Þ ≃ −0.08 − 1.20χ̂ 02 ð1Þ þ 0.13ηð1Þ − 0.26η0 ð1Þ;
GðwÞ
R00 ð1Þ ≃ −0.18 þ 0.87χ̂ 2 ð1Þ þ 0.06ηð1Þ þ 0.25η0 ð1Þ: ð28Þ ≃ 1 − 8a2 ρ2 z þ ðV 21 ρ2 − V 20 Þz2 : ð30Þ
Gðw0 Þ
III. COMBINED FIT TO B → D lν̄ AND B → Dlν̄ Here w0 ¼ 2a2 − 1 ≃ 1.28 is defined such that z ðw0 Þ ¼ 0,
A. Parametrization of the w dependence while V 21 ≃ 57. and V 20 ≃ 7.5 are obtained numerically
from Ref. [48]. The uncertainty in the coefficient of the z2
Unitarity and analyticity provide strong constraints on term in Eq. (30) may be sizable [48]. However, the impact
the shapes of the B → DðÞ lν̄ form factors [46–51]. It is of this term on the physical fit results is expected to
common to employ a parametrization of the B → Dlν̄ form be small.
factor GðwÞ, defined in Eq. (19), via the conformal mapping The leading-order Isgur-Wise function, ξðwÞ, may be
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi
zðwÞ ¼ ð w þ 1 − 2Þ=ð w þ 1 þ 2Þ. Unitarity con- extracted from the parametrization in Eq. (30) by using
straints yield, e.g., GðwÞ=Gð1Þ≃18ρ2 zþð51:ρ2 −10:Þz2 − Eqs. (14) and (13). Keeping terms to Oðεc;b ðw − 1ÞÞ, we
ð252:ρ2 −84:Þz3 , in which ρ2 ¼ −G0 ð1Þ=Gð1Þ is a slope can approximate the subleading Isgur-Wise functions as
parameter [48]. The convergence of this expansion may be
optimized by parametrizing it in a way that minimizes the
range of the expansion parameter, via χ̂ 2 ðwÞ ≃ χ̂ 2 ð1Þ þ χ̂ 02 ð1Þðw − 1Þ; χ̂ 3 ðwÞ ≃ χ̂ 03 ð1Þðw − 1Þ;
pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi ηðwÞ ≃ ηð1Þ þ η0 ð1Þðw − 1Þ; ð31Þ
w þ 1 − 2a 1 þ rD 1=2
z ðwÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi ; a¼ pffiffiffiffiffi : ð29Þ
w þ 1 þ 2a 2 rD since χ̂ 3 ð1Þ ¼ 0. One finds at Oðεc;b ; αs Þ,
ξðwÞ 2 2 2 2 0 1 − rD
≃ 1 − 8a ρ̄ z þ z V 21 ρ̄ − V 20 þ ðεb − εc Þ 2Ξη ð1Þ
ξðw0 Þ 1 þ rD
h i
þ ðεb þ εc Þ Ξ½12χ̂ 03 ð1Þ − 4χ̂ 2 ð1Þ − 16½ða2 − 1ÞΞ − 16a4 χ̂ 02 ð1Þ
0
CV 3 ðw0 Þ þ rD CV 2 ðw0 Þ C0V ðw0 Þ þ rD C0V 2 ðw0 Þ
þ α̂s Ξ CV 1 ðw0 Þ þ þ 2a2 ðΞ − 32a2 Þ 3
1 þ rD 1 þ rD
00 00
CV ðw0 Þ þ rD CV 2 ðw0 Þ
− 64a6 3 − 32a4 C00V 1 ðw0 Þ ; ð32Þ
1 þ rD
where Ξ ¼ 64a4 ρ̄2 − 16a2 − V 21 . The slope parameter Eq. (32), this implies that constraining ξðwÞ in itself does
ρ̄2 ¼ −ξ0 ðw0 Þ=ξðw0 Þ is related to the slope ρ2 ¼ not constrain ηð1Þ, which is the largest unknown contri-
−G0 ðw0 Þ=Gðw0 Þ via bution in R1;2 ð1Þ.
This expression for ξðwÞ, combined with the HQET
h i
expansions in Eqs. (14) and (15), allows one to parametrize
ρ̄2 − ρ2 ¼ ðεb þ εc Þ 12χ̂ 03 ð1Þ − 4χ̂ 2 ð1Þ − 16ða2 − 1Þχ̂ 02 ð1Þ
all B → DðÞ form factors in terms of six parameters: ρ̄2 ,
1 − rD χ̂ 2 ð1Þ, χ̂ 02 ð1Þ, χ̂ 03 ð1Þ, ηð1Þ and η0 ð1Þ. The normalizations of
þ 2ðεb − εc Þη0 ð1Þ
1 þ rD the form factors are also fixed by Eq. (32), and thus jV cb j
may be determined from a global fit to overall rates without
rD CV 2 ðw0 Þ þ CV 3 ðw0 Þ
þ α̂s þ C0V 1 ðw0 Þ using lattice results.
1 þ rD
rD C0V 2 ðw0 Þ þ C0V 3 ðw0 Þ B. QCD sum rule inputs
þ 2a2 : ð33Þ
1 þ rD The subleading Isgur-Wise functions have only been
calculated using model-dependent methods, and are not yet
Enforcing ξð1Þ ¼ 1, one may directly extract ξðw0 Þ via available from lattice QCD. The two-loop QCD sum
evaluation of Eq. (32) at the zero recoil point, z ðw ¼ 1Þ ¼ rule (QCDSR) calculations [23–25] imply that the sub-
ð1 − aÞ=ð1 þ aÞ, and thereby obtain a properly normalized leading Isgur-Wise function ηðwÞ is approximately con-
parametrization for ξðwÞ. Since ηð1Þ does not appear in stant. The functions χ̂ 2;3 , which parametrize corrections
115008-6
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
from the chromomagnetic term in the subleading HQET C. Fit scenarios
Lagrangian, are small, in agreement with quark model A simultaneous fit of the six parameters ρ̄2 , χ̂ 2 ð1Þ, χ̂ 02 ð1Þ,
intuition. χ̂ 03 ð1Þ, ηð1Þ, and η0 ð1Þ to the B → DðÞ lν̄ rates can be carried
The QCD sum rule results are obtained at a fixed scale.
out with the present data. Such a fit fixes both the shapes
The scale dependence can be removed from χ̂ 2;3 by
defining “renormalization improved” functions, χ̂ ren and normalizations of the B → DðÞ lν̄ rates, without any
2;3 [16].
theory input other than the HQET expansion. However, one
These are obtained by multiplying the results of
expects large uncertainties at present, because of the limited
Refs. [23,24] for χ̂ 2;3 by ½αs ðΛÞ3=β0 ∼ 1.4, where Λ ∼ experimental precision and the number of subleading
1 GeV and β0 ¼ 9 for three light flavors. For these HQET parameters. One may instead use QCD sum rule
renormalized subleading Isgur-Wise functions, we use predictions and/or lattice QCD results to constrain the fit,
increasing sensitivity to ρ̄2 . The fit propagates the uncer-
χ̂ ren
2 ð1Þ ¼ −0.06 0.02; χ̂ 0ren
2 ð1Þ ¼ 0 0.02; tainties on the subleading Isgur-Wise functions into the fit
χ̂ 0ren
3 ð1Þ ¼ 0.04 0.02; ηð1Þ ¼ 0.62 0.2; result, and allows the data to further constrain the sub-
leading contributions.
η0 ð1Þ ¼ 0 0.2: ð34Þ Our fit relies on the HQET predictions and unitarity
constraints to determine the ratios and shapes of the form
These central values reproduce L̂1…6 in Ref. [48], often factors. The form factors at zero recoil, Gð1Þ and F ð1Þ,
used to predict R1;2 and RðDðÞ Þ. have been computed in LQCD, providing state-of-the-art
We assign relatively large uncertainties, to permit predictions for the normalizations of the B → DðÞ lν̄ rates.
assessment of possible pulls of the experimental data from The most precise lattice QCD predictions at zero recoil are
these QCDSR predictions. Replacing χ̂ 2;3 with χ̂ ren 2;3 , the [52,53]
Wilson coefficient of the chromomagnetic operator
receives a corresponding αs ðμÞ3=β0 factor at the matching Gð1ÞLQCD ¼ 1.054ð8Þ; F ð1ÞLQCD ¼ 0.906ð13Þ; ð35Þ
pffiffiffiffiffiffiffiffiffiffiffiffi
scale μ ¼ mb mc , partly canceling the above ½αs ðΛÞ3=β0
enhancement. For ease of comparison with the literature we where we combined the quoted systematic and statistical
ignore this, as it can be viewed as a higher-order correction, uncertainties. Although these normalizations may be
and is in any case covered by the large assigned uncer- expected to drop out of the predictions for RðDðÞ Þ, they
tainties. We ignore correlations in the QCDSR results do influence the fit to the differential decay distributions
(arising from the common calculational method), which and hence the resulting form factor ratios. Making use of
is conservative. these lattice constraints leads to our first fitting scenario:
Using Eq. (34) in Eq. (21) yields expressions for R1;2 ðwÞ (i) Rescale the B → D and B → D form factors in the
as polynomials in (w − 1), with the coefficients and their fit by Gð1ÞLQCD =Gð1Þ and F ð1ÞLQCD =F ð1Þ, respec-
uncertainties correlated by HQET. In Ref. [48], the central tively, such that the rates at w ¼ 1 agree with the
values in Eq. (34) were used to write R1;2 ðwÞ as quadratic lattice predictions. We refer to this fit as “Lw¼1 .”
polynomials, without quoting any theory uncertainties on Measurements of the rate normalizations are, however,
their slopes and curvatures. It subsequently become stan- subject to relatively large systematic uncertainties. For
dard practice in experimental jV cb j and R1;2 measurements example, the calibration of the hadronic tagging efficiency
to fit for R1;2 ð1Þ, while fixing R01;2 ð1Þ and R001;2 ð1Þ to their produces systematic uncertainties of the order of a few
quoted central values [48]. Such an approach is inconsistent percent [54]. To compare the best-fit shapes without lattice
with the simultaneous use of the HQET constraints and the constraints and such systematic effects, we consider a
QCDSR results. For example, the present world average second scenario:
central value, R1 ð1Þ ≃ 1.4, cannot simultaneously satisfy (ii) Allow the normalizations of the B → Dlν̄ and B →
the HQET prediction for R1 ð1Þ in Eq. (26) and the QCDSR D lν̄ rates to float independently. This approach
expectation ηð1Þ > 0, which holds at the 3σ level, and is only uses B → DðÞ lν̄ shape information to constrain
used elsewhere in the same fit. A consistent treatment of the form factors, but no theory input for the
these form factor ratios is absent from the derivations of the normalizations at zero recoil, and is independent
state-of-the-art predictions for RðDðÞ Þ in the SM [except of lattice information. We refer to this fit as “NoL.”
for lattice QCD (LQCD) RðDÞ predictions] and in the For each fit, we apply (relax) the QCDSR constraints,
presence of new physics [28,34]. exploring a “constrained” (“unconstrained”) fit. The
We now proceed to assess the importance of obeying the QCDSR constrained fits are denoted with a suffix
HQET relations between different form factors, and of “þSR.” Both Lw¼1 and NoL fits alter the overall normal-
including the uncertainties in the QCDSR predictions in izations the B → Dlν̄ and B → D lν̄ rates, but leave the
Eq. (34). These effects will be important in the future, to HQET expansions of the form factors unchanged. Thus,
systematically improve the SM predictions. they can be considered as introducing an extra source of
115008-7
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
heavy quark symmetry breaking in the normalizations (to All fits explored in this paper use the unitarity con-
effectively account for higher-order effects), while still straints. The consequences of relaxing the unitarity con-
preserving the form factor relations independently in straints between the slope and the curvature terms in
Eqs. (14) and (15). Eq. (30) will be explored in detail elsewhere [55].
Since lattice QCD predictions are also available for
w ≥ 1 for the B → Dlν̄ form factors f þ ðwÞ and f 0 ðwÞ, it is D. Data and fit details
possible to obtain a prediction for the slope parameter, ρ̄2 ,
from them. This leads to a third fit approach, namely: To determine the leading and subleading Isgur-Wise
(iii) Extract ξðwÞ, including the slope parameter ρ̄2, by functions and jV cb j, we carry out a simultaneous fit of the
fitting to the w ≥ 1 lattice QCD data for B → D, and available B → DðÞ lν̄ spectra. There are only two measure-
apply it simultaneously with the LQCD normalization ments [54,56] which provide kinematic distributions fully
of B → D at w ¼ 1. We refer to this fit as “Lw≥1 .” corrected for detector effects. The measured recoil and
In a “theory only” version of this fit, denoted by decay angle distributions are analyzed simultaneously by
“th∶Lw≥1 þ SR,” one fully constrains the B → DðÞ lν̄ constructing a standard χ 2 function. Common uncertainties
differential rates without any experimental input; the (tagging efficiency, reconstruction efficiencies, number of
only fit is to lattice data and QCDSR constraints. For B-meson pairs) should be treated as fully correlated
the “Lw≥1 þ SR” fit, we combine the w ≥ 1 B → D and between the two measurements and we construct a covari-
w ¼ 1 B → D lattice data with QCDSR constraints and ance using Table IV in Ref. [56] and Table IV in Ref. [54].
the experimental information, to include all available While Ref. [56] provides a full breakdown of the total
information and explore possible tensions. We summarize uncertainty for each measured w bin, Ref. [54] only
the inputs of the various fit scenarios pursued in this paper provides a breakdown for the total branching fraction.
in Table I. To construct the desired covariance between both mea-
surements, we thus assume that there is no shape depend-
ence on the tagging and reconstruction efficiency
TABLE I. Summary of theory and data inputs for each fit uncertainty of Ref. [54]. Comparing this with the mild
scenario. All use the HQET predictions to order OðΛQCD =mc;b Þ dependence on these error sources in Ref. [56], this seems a
and Oðαs Þ, as well as the unitarity constraints. fair approximation of the actual covariance. To take into
account the uncertainties of m1S b and δmbc , we introduce
Lattice QCD
both as nuisance parameters into the fit, assuming Gaussian
Fit QCDSR F ð1Þ fþ;0 ð1Þ f þ;0 ðw > 1Þ Belle Data constraints with uncertainties given in Eq. (24). The χ 2
Lw¼1 ✓ ✓ ✓ function is numerically minimized and uncertainties are
Lw¼1 þ SR ✓ ✓ ✓ ✓ evaluated using the usual asymptotic approximations by
NoL ✓ scanning the Δχ 2 ¼ χ 2scan − χ 2min contour to find the þ1
NoL þ SR ✓ ✓ crossing point, which provides the 68% confidence level.
Lw≥1 ✓ ✓ ✓ ✓ The constraints from lattice QCD predictions and/or QCD
Lw≥1 þ SR ✓ ✓ ✓ ✓ ✓
sum rules are incorporated into the fit assuming (multi-
th∶Lw≥1 þ SR ✓ ✓ ✓ ✓
variate) Gaussian errors and are added to the χ 2 function.
TABLE II. Summary of the results for the fit scenarios considered. The correlations are shown in Appendix B.
115008-8
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
The full fit results are shown in Table II. The “Lw¼1 ” the QCDSR constraints in the “NoL þ SR” fit yields results
unconstrained fit, i.e., using only the lattice normalizations close to those in the “Lw¼1 þ SR” fit.
at w ¼ 1, yields In the “th∶Lw≥1 þ SR” scenario, which uses no exper-
imental data, fitting the parametrized ξðwÞ to the six lattice
jV cb j ¼ ð38.8 1.2Þ × 10−3 ; ð36Þ points for f þ;0 ðwÞ in Table III and F ð1Þ in Eq. (35), results
in a slope parameter
to be compared with the current world average [29]
jV cb j¼ð42.20.8Þ×10−3 and jV cb j¼ð39.20.7Þ×10−3 , ρ̄2 ¼ 1.24 0.08: ð38Þ
from inclusive and exclusive b → clν̄l decays, respectively.
The uncertainties of the subleading Isgur-Wise parameters The fitted w spectra are shown in Fig. 1 (gray curves),
are sizable. There is no sensitivity to disentangle η0 ð1Þ from together with the lattice data points. The χ 2 of the fit is 7.4,
ρ̄2 , so we fix η0 ð1Þ to be zero for all QCDSR unconstrained corresponding to a fit probability of 11% with 7 − 3 ¼ 4
fits. Including the QCDSR constraints in the “Lw¼1 þ SR” degrees of freedom. The value for the slope is in good
fit yields agreement with the slope obtained from the QCDSR
constrained and unconstrained “Lw¼1 ” and “NoL” fits.
jV cb j ¼ ð38.5 1.1Þ × 10−3 ; ð37Þ In the “Lw≥1 ” fit, all six lattice points for f þ;0 ðwÞ in
Table III and F ð1Þ in Eq. (35) are fitted together with the
resulting in almost the same jV cb j value. The normalization available experimental information. Once again, η0 ð1Þ is
of ηð1Þ is comparable between these two fits, at about half
fixed to zero, as it is strongly correlated with ρ̄2 . The fit has
the value of the QCDSR expectation. Both fits have
χ 2 ¼ 49, corresponding to a fit probability of 43%. For
reasonable χ 2 values, corresponding to fit probabilities
jV cb j, this fit yields
of 64% each.
Neglecting all subleading ΛQCD =mc;b contributions in
the “Lw¼1 ” fit results in a poorer overall χ 2 . The value of jV cb j ¼ ð39.1 1.1Þ × 10−3 ; ð39Þ
jV cb j decreases slightly, jV cb j ¼ ð38.2 1.1Þ × 10−3 , with
χ 2 ¼ 62.6 for 48 d.o.f., corresponding to a fit probability of which is slightly higher than the “Lw¼1 ” result. The value of
8%, which is still an acceptable fit. The slope parameter ρ̄2 is also higher.
becomes ρ̄2 ¼ 0.93 0.05, below those obtained including In the “Lw≥1 þ SR” fit, the QCDSR constraints are
included, so that all theory and experimental information
the ΛQCD =mc;b corrections. The uncertainty of ρ̄2 is notice-
ably smaller due to the smaller number of degrees of is incorporated. The resulting differential B → DðÞ lν̄
freedom in this fit. The value of jV cb j is only weakly distributions are shown in Fig. 2, overlaid with the
affected by this shift in ρ̄2 . experimental data, as well as the predictions for the B →
In the “NoL” fits, using no LQCD inputs, we use only DðÞ τν̄ differential rates. The fit has χ 2 ¼ 53.8, correspond-
shape information to disentangle ρ̄2 from the subleading ing to a fit probability of 44%. For jV cb j the fit gives
contributions, while allowing the B → Dlν̄ and B → D lν̄
channels to each have arbitrary normalizations (these fits jV cb j ¼ ð39.3 1.0Þ × 10−3 : ð40Þ
cannot determine jV cb j). This results in large uncertainties
in the QCDSR unconstrained fit. Again, η0 ð1Þ and ρ̄2 are This is higher than the “Lw¼1 þ SR” result, because the
strongly correlated, so the former is fixed at zero. Including value of ρ̄2 is also higher.
1.3 0.95
Lw SR SM
1.2 1
0.90 th:Lw 1 SR SM
1.1
f w
f0 w
0.85
1.0
0.80
0.9
0.8 0.75
1.00 1.05 1.10 1.15 1.00 1.05 1.10 1.15
w w
FIG. 1. The “th∶Lw≥1 þ SR” fit of the form factors fþ;0 to the lattice points listed in Table III is shown (gray solid line). The dashed
gray lines correspond to the 68% errors. The dark blue line shows the f þ;0 best fit for “Lw≥1 þ SR”, using lattice points, experimental
information, and QCDSR constraints. The blue band displays the corresponding 68% C.L. of this fit.
115008-9
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
TABLE III. The predictions for the form factors f þ;0 at A more detailed study of these effects, in particular the
w ¼ 1.0, 1.08, 1.16 using the synthetic data results of Ref. [53]. extraction of jV cb j, will be presented elsewhere [55]. A first
The correlations can be found in Table VII in Ref. [53]. comparison with the Caprini-Lellouch-Neubert (CLN)
Form factor w ¼ 1.0 w ¼ 1.08 w ¼ 1.16
parametrization [48], as implemented by previous exper-
imental studies, can be done by considering the results for
fþ 1.19940.0095 1.09410.0104 1.00470.0123 the form factor ratios R1 and R2 , defined in Eq. (20).
f0 0.90260.0072 0.86090.0077 0.82540.0094 Figure 3 shows the extracted values of R1;2 ð1Þ for all fit
scenarios. The results agree with each other and with the
world average of R1 ð1Þ and R2 ð1Þ [9] shown by black
The correlation matrices for all fits are shown in ellipses, up to a mild 1σ tension. Firm conclusions are
Appendix B. In the “Lw¼1 ” and “Lw≥1 ” type fits, moderate difficult to reach, as it is impossible to assess how the
correlations are seen between jV cb j, Gð1Þ, and F ð1Þ, as experimental results would change, had the uncertainties in
expected. The correlations are sizable in these fits between the quadratic polynomials used to fit R1;2 ðwÞ been properly
ρ̄2 and the subleading Isgur-Wise functions. included. When the QCDSR constraints are used, the
40
d B D l Ν l d w 1015 GeV
35 B D l Νl
30 B D Τ ΝΤ
25
20
15
10
5
80
70 B D l Ν l 20
B D l Ν l
60 B D Τ Ν Τ
15
50
40
10
30
20 5
10
1.0 1.1 1.2 1.3 1.4 1.5 1.0 0.5 0.0 0.5 1.0
w cosΘ l
d B D l Ν l d cosΘ v 1015 GeV
6
d B D l Ν l d Χ 1015 GeV
20 5 B D l Ν l
B D l Ν l
15 4
3
10
2
5
1
FIG. 2. The measured B̄ → DðÞ lν̄ decay distributions [54,56] compared to the best-fit contours (dark blue curves) for the
“Lw≥1 þ SR” fit, using LQCD at all w and QCDSR constraints. The blue bands show the 68% C.L. regions. The orange curves and
bands show the central values and the 68% C.L. regions of the fit predictions for dΓðB̄ → DðÞ τν̄Þ=dw.
115008-10
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
FIG. 3. The SM predictions for R1 ð1Þ and R2 ð1Þ for the fits imposing (left) or not imposing (right) the QCDSR constraints in Eq. (34).
The black ellipse shows the world average of the data [9]. The fit scenarios are described in the text and in Table I, and the fit results are
shown in Table II. All contours correspond to 68% C.L. in two dimensions (Δχ 2 ¼ χ 2scan − χ 2min ¼ 2.3).
central values satisfy R1 ð1Þ < 1.34, as required by the account the full covariance of the theory prediction and the
HQET prediction in Eq. (26) and the constraint ηð1Þ > 0. world average measurement. The tension between our most
precise “Lw≥1 þ SR” fit and the data is 3.9σ, with a p-value
E. RðDðÞ Þ and new physics of 11.5 × 10−5 , to be compared with 8.3 × 10−5 quoted by
HFAG [9]. The precision of this prediction is limited by that
Using the fitted values for ρ̄2, χ̂ 2 ð1Þ, χ̂ 02 ð1Þ, χ̂ 03 ð1Þ, ηð1Þ,
of the input measurements and LQCD inputs, and can be
and η0 ð1Þ, one can predict RðDðÞ Þ in the SM and for any
systematically improved with new data from Belle II
new physics four-fermion interaction. Figure 4 and
or LHCb.
Table IV summarize the predicted values of RðDðÞ Þ in To derive a SM prediction for RðD Þ, the authors of
the SM for the seven fit scenarios considered. Our fit results Ref. [34] used the measured R2 ð1Þ form factor ratio [9] and
for RðDÞ are in good agreement with other predictions in the QCDSR predictions to obtain R0 ð1Þ ¼ 1.14 0.11. In
the literature [57,58]. All our fits using lattice QCD inputs comparison, our “Lw≥1 þ SR” fit results yield
yield RðD Þ above those in Ref. [34]. This slightly eases
the disagreement with the world average measurement [9].
The significance is calculated from χ 2 statistics, taking into R0 ð1Þ ¼ 1.17 0.02; R3 ð1Þ ¼ 1.19 0.03: ð41Þ
FIG. 4. The SM predictions for RðDÞ and RðD Þ, imposing (left) or not imposing (right) the QCDSR constraints (see Table IV). Gray
ellipses show other SM predictions (last three rows of Table IV). The black ellipse shows the world average of the data [9]. The contours
are 68% C.L. (Δχ 2 ¼ 2.3), hence the nearly 4σ tension.
115008-11
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
TABLE IV. The RðDÞ and RðD Þ predictions for our fit NP couplings are permitted to have an arbitrary phase,
scenarios, the world average of the data, and other theory generating allowed regions rather than single contours. We
predictions. The fit scenarios are described in the text and in display the allowed regions generated for the “NoL þ SR”
Table I. The bold numbers are our most precise predictions. best-fit values, the “Lw≥1 þ SR” best-fit values, and for
Scenario RðDÞ RðD Þ Correlation leading-order contributions only, i.e., αs , εc;b → 0, with
ρ̄2 ¼ 1.24. The small variation between the “NoL þ SR”
Lw¼1 0.292 0.005 0.255 0.005 41%
and “Lw≥1 þ SR” regions illustrates the good consistency
Lw¼1 þ SR 0.291 0.005 0.255 0.003 57%
NoL 0.273 0.016 0.250 0.006 49% of the predictions obtained with and without LQCD. In
NoL þ SR 0.295 0.007 0.255 0.004 43% each plot, we also include for comparison the correspond-
Lw≥1 0.298 0.003 0.261 0.004 19% ing contours (dashed lines) produced by a NP OV − OA
Lw≥1 þ SR 0.299 0.003 0.257 0.003 44% coupling. The latter rescales RðDÞ and RðD Þ keeping their
th∶Lw≥1 þ SR 0.306 0.005 0.256 0.004 33% ratio fixed. Solid dots indicate the SM point for each case.
Data [9] 0.403 0.047 0.310 0.017 −23% For scalar currents, if NP only contributes to OS (OP ) then
only RðDÞ [RðD Þ] is affected in accordance with Eq. (10b)
Refs. [53,57,59] 0.300 0.008 [Eq. (11a)], respectively. We plot the allowed regions for
Ref. [58] 0.299 0.003
the OS OP linear combinations, which are also motivated
Ref. [34] 0.252 0.003
by specific NP models.
FIG. 5. The allowed ranges of RðDÞ − RðD Þ, due to one of the new physics operators in addition to the SM: OS − OP (top left),
OS þ OP (top right), OV þ OA (bottom left), and OT (bottom right).
115008-12
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
or omit lattice QCD and QCD sum rule predictions, we and Robert Michaud for the kind hospitality in Houston
constrained the leading and subleading Isgur-Wise functions. where part of this work was carried out and inspiring
We thus obtained strong constraints on all form factors, and conversations over good wine and food. Z. L. and M. P.
predictions for the form factor ratios R1;2 as well as RðDðÞ Þ, were supported in part by the U.S. Department of Energy
both in the SM and in arbitrary NP scenarios, valid at Oðαs Þ under contract DE-AC02-05CH11231. D. R. acknowl-
and OðΛQCD =mc;b Þ. Our most precise prediction for edges support from the University of Cincinnati.
RðDðÞ Þ, in the “Lw≥1 þ SR” fit, using the experimental
data and all lattice QCD and QCDSR inputs is APPENDIX A: THE Oðαs Þ CORRECTIONS
RðDÞ ¼ 0.299 0.003; In this appendix we summarize the explicit expressions
for the CΓ ðwÞ functions defined in Eq. (9), calculated in
RðD Þ ¼ 0.257 0.003; ð42Þ Ref. [27]. The following results use the MS scheme and
with a correlation of 44%. The same fit also yields jV cb j ¼ correspond to matching from QCD onto HQET at
pffiffiffiffiffiffiffiffiffiffiffiffi
ð39.3 1.0Þ × 10−3 , which is in good agreement with μ ¼ mc mb :
existing exclusive determinations. All possible b → c cur-
rent form factors were derived at OðΛQCD =mc;b Þ and Oðαs Þ, 1
CS ¼ ½2zðw − wz ÞΩðwÞ − ððw − 1Þðz þ 1Þ2 rðwÞ
including those for a tensor current, previously unavailable zðw − wz Þ
in the literature at this order. A lattice QCD calculation of the − ðz2 − 1Þ ln zÞ; ðA1aÞ
subleading Isgur-Wise functions, or even just those which
arise from the chromomagnetic term in the subleading 1
HQET Lagrangian (χ 2;3 ), would be important to reduce CP ¼ ½ðz − 1Þ½ðwðz3 − ð3 þ 2wÞz2 þ z − 1Þ
2z2 ðw − wz Þ2
hadronic uncertainties in both SM and NP predictions, com-
plementary to a long-awaited lattice calculation of RðD Þ. þ ðz2 þ 3ÞzÞrðwÞ þ ðz2 − 1Þ ln z
At the current level of experimental precision, our pre- − 2zðwz − wÞðz − 1 þ ðz þ 1Þz ln zÞ
dictions agree up to mild tensions with previous results, which
neglected the HQET relations for the uncertainties of the þ 4z2 ðw − wz Þ2 ΩðwÞ; ðA1bÞ
OðΛQCD =mc;b Þ terms. Our fit results are consistent with one
1
another, and at the current level of precision we find no CV 1 ¼ ½2ðw þ 1Þðð3w − 1Þz − z2 − 1ÞrðwÞ
inconsistencies between the data, lattice QCD results, and 6zðw − wz Þ
QCD sum rule predictions. Our fit using all available lattice þ ð12zðwz − wÞ − ðz2 − 1Þ ln zÞ
QCD and QCD sum rule inputs and HQET to order
þ 4zðw − wz ÞΩðwÞ; ðA1cÞ
Oðαs ; ΛQCD =mc;b Þ yields the most precise combined predic-
tion for RðDÞ and RðD Þ to date. However, in principle, our fit −1
need not require either lattice or sum rule input, and its CV 2 ¼ ½ðð4w2 þ 2wÞz2 − ð2w2 þ 5w − 1Þz
precision can be improved simply as the statistics of future 6z2 ðw− wz Þ2
data increases. − ðw þ 1Þz3 þ 2ÞrðwÞ þ zð2ðz − 1Þðwz − wÞ
The (moderate) tension between the measurements of
þ ðz2 − ð4w − 2Þz þ ð3 − 2wÞÞ ln zÞ; ðA1dÞ
jV cb j from inclusive and exclusive semileptonic decays
probably cannot be resolved with current data. Under- 1
standing how the inclusive rate is made up from a sum of CV 3 ¼ ½ðð2w2 þ 5w − 1Þz2 − ð4w2 þ 2wÞz
6zðw − wz Þ2
exclusive channels has been unclear from the data for a long
time [60], and puzzles remain even in light of BABAR and − 2z3 þ w þ 1ÞrðwÞ þ ð2zðz − 1Þðwz − wÞ
Belle measurements [61,62]. A more detailed examination of þ ðð3 − 2wÞz2 þ ð2 − 4wÞz þ 1Þ ln zÞ; ðA1eÞ
the effects of the unitarity constraints and the precision
extraction of jV cb j is the subject of ongoing work [55]. We 1
are also implementing the full angular distributions of the CA1 ¼ ½2ðw − 1Þðð3w þ 1Þz − z2 − 1ÞrðwÞ
6zðw − wz Þ
measurable particles [63,64] into a software package,
HAMMER [65,66], based on the state-of-the-art HQET þ ð12zðwz − wÞ − ðz2 − 1Þ ln zÞ
predictions for all six B → D; D ; D decay modes. þ 4zðw − wz ÞΩðwÞ; ðA1fÞ
ACKNOWLEDGMENTS −1
CA2 ¼ ½ðð4w2 − 2wÞz2 þ ð2w2 − 5w − 1Þz
We thank Marat Freytsis, Ben Grinstein and Aneesh 6z2 ðw − wz Þ2
Manohar for helpful conversations. F. B. was supported by þ ð1 − wÞz3 þ 2ÞrðwÞ þ zð2ðz þ 1Þðwz − wÞ
the Deutsche Forschungsgemeinschaft (DFG) Emmy-
Noether Grant No. BE 6075/1-1. F. B. thanks Kim Scott þ ðz2 − ð4w þ 2Þz þ ð2w þ 3ÞÞ ln zÞ; ðA1gÞ
115008-13
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
1 2 2
CA3 ¼ ½ð2z3 þ ð2w2 − 5w − 1Þz2 CS ð1Þ ¼ − ; CP ð1Þ ¼ ;
6zðw − wz Þ2 3 3
þ ð4w2 − 2wÞz − w þ 1ÞrðwÞ þ ð2zðz þ 1Þðwz − wÞ 4 1þz 2ð1 − z þ z ln zÞ
CV 1 ð1Þ ¼ − − ln z; CV 2 ð1Þ ¼ − ;
3 1−z 3ð1 − zÞ2
− ðð2w þ 3Þz2 − ð4w þ 2Þz þ 1Þ ln zÞ; ðA1hÞ
2zð1 − z þ ln zÞ
CV 3 ð1Þ ¼ ;
1 3ð1 − zÞ2
CT 1 ¼ ½ðw − 1Þðð4w þ 2Þz − z2 − 1ÞrðwÞ 8 1þz
3zðw − wz Þ CA1 ð1Þ ¼ − − ln z;
3 1−z
þ ð6zðwz − wÞ − ðz2 − 1Þ ln zÞ
2½3 − 2z − z2 þ ð5 − zÞz ln z
þ 2zðw − wz ÞΩðwÞ; ðA1iÞ CA2 ð1Þ ¼ − ;
3ð1 − zÞ3
2z½1 þ 2z − 3z2 þ ð5z − 1Þ ln z
2 CA3 ð1Þ ¼ ;
CT 2 ¼ ½ð1 − wzÞrðwÞ þ z ln z; ðA1jÞ 3ð1 − zÞ3
3zðw − wz Þ
8 4ð1 þ zÞ
CT 1 ð1Þ ¼ − − ln z; CT 2 ð1Þ ¼ 2CV 2 ð1Þ;
2 3 3ð1 − zÞ
CT 3 ¼ ½ðw − zÞrðwÞ þ ln z; ðA1kÞ CT 3 ð1Þ ¼ −2CV 3 ð1Þ: ðA4Þ
3ðw − wz Þ
and CT 4 ¼ 0. Here z ¼ mc =mb , and the functions Finally, for arbitrary matching scale μ, one should add to
Eq. (A1) the terms
w
ΩðwÞ ≡ pffiffiffiffiffiffiffiffiffiffiffiffiffi ½2Li2 ð1 − w− zÞ − 2Li2 ð1 − wþ zÞ ðμ2 Þ ðm mc Þ 1
2 w2 − 1 CS;P ¼ CS;Pb − ½2wrðwÞ þ 1 lnðmc mb =μ2 Þ; ðA5aÞ
3
þ Li2 ð1 − w2þ Þ − Li2 ð1 − w2− Þ
ðμ2 Þ ðmb mc Þ 2
− wrðwÞ ln z þ 1; ðA2Þ CV 1 ;A1 ¼ CV 1 ;A − ½wrðwÞ − 1 lnðmc mb =μ2 Þ; ðA5bÞ
1
3
R0
where Li2 ðxÞ ¼ lnð1 − tÞ=tdt is the dilogarithm, and ðμ2 Þ ðm mc Þ 1
x CT 1 ¼ CT 1 b − ½2wrðwÞ − 3 lnðmc mb =μ2 Þ; ðA5cÞ
3
ln wþ pffiffiffiffiffiffiffiffiffiffiffiffiffi
rðwÞ ≡ pffiffiffiffiffiffiffiffiffiffiffiffiffi ; w ≡ w w2 − 1; and all other CΓj ¼ CΓj b
ðμ2 Þ ðm mc Þ
, for j ≥ 2.
w2 − 1
1
wz ≡ ðz þ 1=zÞ: ðA3Þ APPENDIX B: DULL CORRELATIONS
2
The correlation matrices for the fit scenarios are given in
At the zero recoil point, w ¼ 1, Tables V–XI.
115008-14
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
TABLE VI. The correlations of the “Lw¼1 þ SR” fit scenario.
115008-15
BERNLOCHNER, LIGETI, PAPUCCI, and ROBINSON PHYSICAL REVIEW D 95, 115008 (2017)
TABLE X. The correlations of the “Lw≥1 þ SR” fit scenario.
[1] N. Isgur and M. B. Wise, Phys. Lett. B 232, 113 (1989). [13] M. E. Luke, Phys. Lett. B 252, 447 (1990).
[2] N. Isgur and M. B. Wise, Phys. Lett. B 237, 527 (1990). [14] M. Neubert and V. Rieckert, Nucl. Phys. B382, 97 (1992).
[3] J. P. Lees et al. (BABAR Collaboration), Phys. Rev. Lett. [15] A. V. Manohar and M. B. Wise, Heavy Quark Physics
109, 101802 (2012). (Cambridge University Press, Cambridge, England, 2000).
[4] J. P. Lees et al. (BABAR Collaboration), Phys. Rev. D 88, [16] M. Neubert, Phys. Rep. 245, 259 (1994).
072012 (2013). [17] A. F. Falk, H. Georgi, B. Grinstein, and M. B. Wise,
[5] M. Huschle et al. (Belle Collaboration), Phys. Rev. D 92, Nucl. Phys. B343, 1 (1990).
072014 (2015). [18] J. D. Bjorken, in Gauge Bosons and Heavy Quarks: Pro-
[6] A. Abdesselam et al. (Belle Collaboration), arXiv: ceedings, 18th SLAC Summer Institute on Particle Physics
1603.06711. (SSI 90), July 16–27, 1990, edited by J. F. Hawthorne (SLAC,
[7] A. Abdesselam et al. (Belle Collaboration), arXiv: Stanford, 1991), p. 167.
1608.06391. [19] A. F. Falk, Nucl. Phys. B378, 79 (1992).
[8] R. Aaij et al. (LHCb Collaboration), Phys. Rev. Lett. 115, [20] A. F. Falk and M. Neubert, Phys. Rev. D 47, 2965 (1993).
111803 (2015); 115, 159901(E) (2015). [21] M. E. Luke and A. V. Manohar, Phys. Lett. B 286, 348
[9] Y. Amhis et al. (Heavy Flavor Averaging Group Collabo- (1992).
ration), arXiv:1612.07233, and updates at http://www.slac [22] B. Grinstein and Z. Ligeti, Phys. Lett. B 526, 345 (2002);
.stanford.edu/xorg/hfag/. 601, 236(E) (2004).
[10] M. Freytsis, Z. Ligeti, and J. T. Ruderman, Phys. Rev. D 92, [23] M. Neubert, Z. Ligeti, and Y. Nir, Phys. Lett. B 301, 101
054018 (2015). (1993).
[11] J. Charles, S. Descotes-Genon, Z. Ligeti, S. Monteil, M. [24] M. Neubert, Z. Ligeti, and Y. Nir, Phys. Rev. D 47, 5060
Papucci, and K. Trabelsi, Phys. Rev. D 89, 033016 (2014). (1993).
[12] F. U. Bernlochner and Z. Ligeti, Phys. Rev. D 95, 014022 [25] Z. Ligeti, Y. Nir, and M. Neubert, Phys. Rev. D 49, 1302
(2017). (1994).
115008-16
COMBINED ANALYSIS OF SEMILEPTONIC B DECAYS … PHYSICAL REVIEW D 95, 115008 (2017)
[26] A. F. Falk and B. Grinstein, Phys. Lett. B 249, 314 (1990). [49] C. G. Boyd and M. J. Savage, Phys. Rev. D 56, 303
[27] M. Neubert, Nucl. Phys. B371, 149 (1992). (1997).
[28] M. Tanaka and R. Watanabe, Phys. Rev. D 87, 034028 [50] R. J. Hill, Proceedings, 4th Conference on Flavor Physics
(2013). and CP Violation (FPCP 2006): Vancouver, British
[29] C. Patrignani et al. (Particle Data Group Collaboration), Columbia, Canada, April 9–12, 2006, eConf C060409,
Chin. Phys. C 40, 100001 (2016). 027 (2006).
[30] M. A. Shifman and M. B. Voloshin, Yad. Fiz. 47, 801 (1988) [51] C. Bourrely, I. Caprini, and L. Lellouch, Phys. Rev. D 79,
[Sov. J. Nucl. Phys. 47, 511 (1988)]. 013008 (2009); 82, 099902(E) (2010).
[31] C. G. Boyd, B. Grinstein, and A. V. Manohar, Phys. Rev. D [52] J. A. Bailey et al. (Fermilab Lattice and MILC Collabora-
54, 2081 (1996). tions), Phys. Rev. D 89, 114504 (2014).
[32] A. V. Manohar, Lect. Notes Phys. 479, 311 (1997). [53] J. A. Bailey et al. (Fermilab Lattice and MILC Collabora-
[33] A. Sirlin, Nucl. Phys. B196, 83 (1982). tions), Phys. Rev. D 92, 034506 (2015).
[34] S. Fajfer, J. F. Kamenik, and I. Nisandzic, Phys. Rev. D 85, [54] A. Abdesselam et al. (Belle Collaboration),
094025 (2012). arXiv:1702.01521.
[35] M. Neubert and C. T. Sachrajda, Nucl. Phys. B438, 235 [55] F. Bernlochner, Z. Ligeti, M. Papucci, and D. J. Robinson
(1995). (to be published).
[36] M. E. Luke, A. V. Manohar, and M. J. Savage, Phys. Rev. D [56] R. Glattauer et al. (Belle Collaboration), Phys. Rev. D 93,
51, 4924 (1995). 032006 (2016).
[37] M. Beneke and V. M. Braun, Nucl. Phys. B426, 301 (1994). [57] S. Aoki et al. (Flavour Lattice Averaging Group Collabo-
[38] I. I. Y. Bigi, M. A. Shifman, N. G. Uraltsev, and A. I. ration), Eur. Phys. J. C 77, 112 (2017), and updates at http://
Vainshtein, Phys. Rev. D 50, 2234 (1994). itpwiki.unibe.ch/flag/.
[39] M. Beneke, V. M. Braun, and V. I. Zakharov, Phys. Rev. [58] D. Bigi and P. Gambino, Phys. Rev. D 94, 094008 (2016).
Lett. 73, 3058 (1994). [59] H. Na, C. M. Bouchard, G. P. Lepage, C. Monahan, and J.
[40] A. H. Hoang, Z. Ligeti, and A. V. Manohar, Phys. Rev. Lett. Shigemitsu (HPQCD Collaboration), Phys. Rev. D 92,
82, 277 (1999). 054510 (2015); 93, 119906(E) (2016).
[41] A. H. Hoang, Z. Ligeti, and A. V. Manohar, Phys. Rev. D 59, [60] J. D. Richman and P. R. Burchat, Rev. Mod. Phys. 67, 893
074017 (1999). (1995).
[42] A. H. Hoang, Phys. Rev. D 61, 034005 (1999). [61] F. U. Bernlochner, Z. Ligeti, and S. Turczyk, Phys. Rev. D
[43] M. Beneke, Phys. Lett. B 434, 115 (1998). 85, 094033 (2012).
[44] A. Czarnecki, K. Melnikov, and N. Uraltsev, Phys. Rev. [62] F. U. Bernlochner, D. Biedermann, H. Lacker, and T. Luck,
Lett. 80, 3189 (1998). Eur. Phys. J. C 74, 2914 (2014).
[45] Z. Ligeti and F. J. Tackmann, Phys. Rev. D 90, 034021 [63] Z. Ligeti, M. Papucci, and D. J. Robinson, J. High Energy
(2014). Phys. 01 (2017) 083.
[46] C. G. Boyd, B. Grinstein, and R. F. Lebed, Nucl. Phys. [64] R. Alonso, A. Kobach, and J. Martin Camalich, Phys. Rev.
B461, 493 (1996). D 94, 094021 (2016).
[47] C. G. Boyd, B. Grinstein, and R. F. Lebed, Phys. Rev. D 56, [65] S. Duell, F. Bernlochner, Z. Ligeti, M. Papucci, and D.
6895 (1997). Robinson, Proc. Sci., ICHEP2016 (2016) 1074.
[48] I. Caprini, L. Lellouch, and M. Neubert, Nucl. Phys. B530, [66] F. Bernlochner, S. Duell, Z. Ligeti, M. Papucci, and D. J.
153 (1998). Robinson (to be published).
115008-17