Skip to main content

Full text of "Visualization of semileptonic form factors from lattice QCD"

See other formats

Visualization of semileptonic form factors from lattice QCD 









C. Bernard,! C. DeTar,^ M. Di Pierro,^ A.X. El-Khadra,-* R.T. Evans,"* E.D. Freeland,^ E. Gamiz,'* 
Steven Gottlieb,^ U.M. Heller,'^ J.E. Hetrick,^ A.S. Kronfeld,^ J. Laiho,* L. Levkova,^ P.B. Mackenzie,^ 
M. Okamoto,^ M.B. Oktay,^ J.N. Simone,^ R. Sugar,io D. Toussaint," and R.S. Van de Water^^ 

(Fermilab Lattice and MILC Collaborations) 

^Department of Physics, Washington University, St. Louis, Missouri, USA 
'^Physics Department, University of Utah, Salt Lake City, Utah, USA 
'^School of Computer Science, Telecommunications and Information Systems, DePaul University, Chicago, Illinois, USA 
'^Physics Department, University of Illinois, Urbana, Illinois, USA 
^Liberal Arts Department, The School of the Art Institute of Chicago, Chicago, Illinois, USA 
''Department of Physics, Indiana University, Bloomington, Indiana, USA 

''American Physical Society, Ridge, New York, USA 
^Physics Department, University of the Pacific, Stockton, California, USA 
Fermi National Accelerator Laboratory, Batavia, Illinois, USA 
'^"Department of Physics, University of California, Santa Barbara, California, USA 
''Department of Physics, University of Arizona, Tucson, Arizona, USA 
'^Physics Department, Brookhaven National Laboratory, Upton, New York, USA 

(Dated: August 26, 2009) 

Comparisons of lattice-QCD calculations of semileptonic form factors with experimental measure- 
ments often display two sets of points, one each for lattice QCD and experiment. Here we propose 
to display the output of a lattice-QCD analysis as a curve and error band. This is justified, because 
lattice-QCD results rely in part on fitting, both for the chiral extrapolation and to extend lattice- 
QCD data over the full physically allowed kinematic domain. To display an error band, correlations 
in the fit parameters must be taken into account. For the statistical error, the correlation comes 
from the fit. To illustrate how to address correlations in the systematic errors, we use the Becirevic- 
Kaidalov parametrization of the D nlf and D Klv form factors, and an analyticity-based fit 
for the B — > ttIu form factor /+. 

PACS numbers: 13.20.Fc,13.20.He,12.38.Gc 

The past several years have witnessed considerable im- 
provement in our understanding of semileptonic decays 
of D and B mesons. Measurements have advanced in 
accuracy from 6-20% on the normalization [l|J3, 0] and 
~ 10% on the shape [1] to 1% on both Mean- 
while, ab initio calculations in QCD with lattice gauge 
theory have become realistic [1, 0, H, |^ , now incorporat- 
ing the effects of sea quarks that were omitted in earlier 
work 0, [m, [H, [H, d. In this article, we discuss how 
to present both together, so that the agreement (or, in 
principle, lack thereof) is easy to assess. 

We focus on reactions mediated by electroweak vector 
currents, leading to pseudoscalar mesons, tt or K , in the 
final state. At the quark level, a heavy quark h decays 
into a daughter quark d (not necessarily the down quark) , 
with a spectator antiquark q. Writing the decay H — > 
Plv, the form factors are defined by 

(PIV^IH) = f+{q^){pH +PP - AT + Mq^)A^^ (1) 

denote the 3-momentum and energy of the final-state me- 
son in the rest frame of the initial state. The energy E 
is related to q"^ via 



Neglecting the lepton mass, < g'^ < qf^ax — {rriH—mp)'^ 
is kinematically allowed in the semileptonic decay. 

The form factors f+{q^) and fo{<l^) are related to /|| {E) 
^nd f^{E) by 

f+{q') = {2mH)-'/' [fi\iE) + {mn ~ E)f^iE)] , (4) 

[{mH^E)f\\iE)-plME)] , (5) 


where q = pn — pp is the 4-momentum of the lepton 
system, and = {pH+Pp)-qq^ / (f' = {'m% — mp)q'^/q'^. 
Equation ^ is general, applying to K ^ -kIv as well as to 
D and B decays. For lattice QCD, it is more convenient 
to express the transition matrix element as 

{P\V^\H) = ,/^[v^^f\\{E)+py^{E)\ , (2) 
where v — pn/mH, and p± = pp — Ev and E — v ■ pp 

with Eq. 13]) understood. Equations ([4]) and 
/_^_(0) = /o(0), as required in Eq. ^. 

Two aspects of lattice-QCD calculations are important 
here. First (as in all lattice-QCD calculations), it is com- 
putationally demanding to have a spectator quark with 
mass as small as those of the up and down quarks; for 
P — TT the same applies to the daughter quark. In re- 
cent unquenched calculations, the mass of the qq pseu- 
doscalar Pqq lies in the range 0.1m|- < 'rnp_^ < mj^. Sec- 
ond (of special importance in semileptonic decays), the 
calculations take place in a finite spatial volume, so the 
3-momentum takes discrete values. In typical cases the 


box-size L Ki 2.5 fm, so the smallest nonzero momentum 
P(i,o,o) = 27r(l,0,0)/L satisfies |P(i,o,o)l ~ 500 MeV. 

After generating numerical data at several values of 
{E,mp_J, the next step for lattice-QCD calculations is 
to carry out a chiral extrapolation, ?Tip_^ — > m^, of the 
data for and /|| [H, The chiral extrapolation 

must refle ct the fac t that the form factors are analytic 
in _E = \/p'^ + ^p, not p Note also that f± and 

/-!_ can be computed only with p =/= 0, hence E > mp 
or, equivalently, < (/nj^x- The statistical and dis- 
cretization uncertainties in [q^ , TOp_ ) start out small- 
est at q^^ Q corresponding to P(i,o,o)- A sensible chiral 

extrapolation will propagate this feature to /+(g^,m^). 
Similarly, the statistical and discretization uncertainties 
in /o(g^,mJ) are smallest near g^^x- 

When \p\a becomes too large, discretization effects 
grow out of control. Therefore, the kinematic domain 
of lattice-QCD calculations is limited to, these days, 
IpI ^ 1 GeV, with a corresponding upper limit on E 
and lower limit on q^ . To extend the form factor over 
the full physical kinematic domain, a parametrization of 
the q^ dependence is needed. 

One choice is the Becirevic-Kaidalov (BK) ansatz [Ig] 





where = q^/m^, {H* is the vector meson of flavor hd), 
and F, a, and (3 are free parameters to be fitted. A key 
feature of Eq. © is the built-in pole at q"^ = m%, , or 
E = —{mjjt. — mfj — rn?p)/2mH < 0, an indisputable fea- 
ture of the physical /+. Further singularities at higher 
negative energy are modeled by the BK parameters a 
and (3. A similar possibility is the Ball-Zwicky (BZ) 
ansatz IToj . which has one more parameter for /+ 
than BK. A shortcoming of these parametrizations is that 
comparisons of lattice-QCD and experimental slope pa- 
rameters can be misleading 0,[2l|, because lattice-QCD 
slopes are determined near q^ = g^^x; whereas experi- 
mental slopes are determined near q^ — 0. 

Another approach based on analyticity and unitarity 
is to write the form factors as 


1 ^ 




where (/>+,o are arbitrary, but suitable, functions, and the 
series coefficients are fit parameters. The variable 


^\-q^/t+ + ^l-to/t+' 

where i+ = {mn + mp)^ and to can be chosen to make 
1 2: 1 small for all kinematically allowed q^. Like BK and 

BZ, Eq. dH]) builds the H* pole into /+, but this approach 
is model independent because unitarity [13, [H, [23l] and 
heavy-quark physics [2l| impose bounds on X^fel^fcP' 
J2k l^fcP; a-iid because kinematics set \z\ < 1. Conse- 
quently, the series can be truncated safely, once addi- 
tional terms are negligible compared to other uncertain- 
ties in the analysis. 

In all approaches the output of an analysis of lattice- 
QCD form factors is a fit, usually a two-stage fit of chiral 
extrapolation followed by q'^ parametrization. Clearly, 
the final fit describes a curve, and the error matrix of 
the fit parameters describes an error band. Nevertheless, 
lattice-QCD results usually have been plotted as a set 
of points with error bars at fiducial values of q^ (or E). 
These points evoke the underlying discrete nature of the 
3-momentum p but, in general, the chosen values of q^ (or 
E) have nothing to do with the original discrete values 
of p. A plot with a curve plus error band exhibits the 
same information, while giving a visually superior sense 
of the correlations between points on the curve. 

The experimental measurements of /_|_(g^) come from 
counting events in bins of q^ and removing coupling and 
kinematic factors. The analysis inevitably entails some 
fitting, to correct for acceptance, etc., but the postfit bins 
of q^ faithfully mirror the input to such fits. 

If one would like to compare the calculations with the 
measurements, it is appealing to represent one as a curve 
with error band, and the other as points with error bars. 
Bearing the foregoing remarks in mind, it seems natural 
to draw the curve for lattice-QCD calculations. A few 
years ago, we prepared illustrative plots for D Klv 
with the Fermilab-MILC m, B lattice-QCD calculations 
and FOCUS [4] and Belle [2J] measurements. The intent 
was pedagogical, and we showed the plots at seminars 
and conferences (25| . 

Unfortunately, the error band in that effort was im- 
pressionistic, not rigorous. With the prospect of yet- 
more-precise results based on CLEO-c's full accumula- 
tion of 818 pb^^ [5], we now present a version that treats 
the error band as rigorously as possible. We also prepare 
plots for D — *■ 'kIv and B — > irlv. 

As before we shall base the plots for D decays on 
Ref. . The final result of this analysis consists of the 
BK parameters {F, a, [3) and the 3x3 error matrix. The 
full statistical error matrix is contained in a detailed, un- 
published description of a BK-based analysis of i? — > nlv 
form factors [2^. The best fit, statistical errors, and sys- 
tematic errors are tabulated in Table [H The statistical 
correlation matrices pij = CTfj/(o'^iO'|j)^^^ are tabulated 
in Table ini The correlations among systematic errors are 
discussed below. 

Propagating (correlated) fluctuations in F, a, and (3 
to the form factors, one flnds relative squared-errors 




^2 -2 
1 r,<^Fa q 

+ ^aa 




F 1-aq^ 

\ 1 — aq'^ 


' e \ 



Fj3 I3~q^ 

13^ \ 




These errors are plotted as a function of in Fig. [T] as 
solid curves. The relative statistical errors are smallest 
for such that 


It is illustrative to take a"^^ and a|n^ from Tables HI and HIl 
and solve Eqs. (fT3| and (fT4|) for g^. We call these values 

and g| and tabulate them, as well as. qj^^^^^ and q^^y., 
in Table IIIII As one can see from Fig. [T| and Table [ml 
the statistical error is smallest between g^j^ g p-j and gi^axi 
as expected. One may view this outcome as a check on 
the fitting procedures. 

One can reverse this strategy to determine the corre- 
lation between the systematic errors of F and the slope 
parameters. In the error budget of Ref. @ the largest 
systematic effect comes from discretization errors. These 
should be smallest around q^^ q and g^ax for /+ and /o, 
respectively, because those correspond to the smallest |p| 
yielding the respective matrix elements. This yields 


-0.198 {D 



-0.329 {D 
-0.533 {D 


and the dashed curves in Fig. [TJ 

It is customary to combine statistical and systematic 
uncertainties by adding the two cr^ (matrices). Carrying 
out this procedure leads to the curves and bands in Fig. [21 
The error bands seem to contradict the conventional wis- 
dom that the lattice-QCD uncertainties are smallest near 
9max- This is not entirely the case for the relative error, as 
seen in Fig.[Tl As increases, the relative errors decrease 
until hitting a minimum somewhere between q^^ ^ and 

9maxj is reasonable. The form factors rise faster than 
the relative errors drop, leading to the increasing absolute 
error seen in Fig. [21 These features are not an artifact of 
the BK parametrization, as we shall see below with the 
B -kIv form factor /+. 

Figure[2lis the first main result of this article. It shows 
the form factors /+ and /o for D Klv and D irlv. 
The lattice-QCD results are shown as curves (red for /+, 
blue for /o) with two errors bands, one statistical (orange 
for /+, gray for /o), the other systematic and statistical 
combined (yellow for /-i-, light blue for /o). Experimental 

measurements for 

are overlaid as points 

with error bars. It may require careful scrutiny to see 
which experiment is which, but a glance reveals how well 
the points and curves agree. The agreement is good for 
D ttIv and very good for D Klv. 

For the z expansion the propagation of errors is even 
simpler. Focusing on /+, one has from Eq. ([5]) 


^2 ^k+l 

O 1.1 ^ 


Z^k=0 "-k^ 


where the indices on correspond to those on the se- 
ries coefficients. The coefficients and error matrix for the 

TABLE I: Best-fit values of BK parameters with statistical 
and systematic errors, successively, in parentheses [^. [tI [2^. 



D^Klv 0.73(3)(7) 0.50(4)(7) 1.31(7)(13) 
D-^ttIp 0.64(3)(6) 0.44(4)(7) 1.41(6)(7) 

TABLE IL Statistical error correlation matrices pij 
(^fj /{'^'ii'^'jjy^^ of the BK parameters |26j |. 

































2 2 

q tni^* 

'max Z), 



— U 

— fa 

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 


'max D 




2 2 
q Im , 

FIG. 1: Relative errors vs . Solid (dashed) curves show 
the fitted statistical (estimated systematic) error for /+ (red 
curves) and /o (blue curves). Vertical lines show 9^1,0,0) 



TABLE III: Useful quantities for generating and assessing Figs. [l][2l and|3] 






9m ax 




9m ax 







D Klv 










D ~* nlf 










B Tvlf 








iV = 3 fit (to = 0.65g,^a^) are tabulated in Tablc[lVl The 
z-series fit was carried out after assigning g^-dependent 
systematic uncertainties, so Table HVl refers to the com- 
bined statistical and systematic errors of this analysis. 

This information, combined with the outer function 0+ 
0, is used to produce Fig. [31 the second main result 

FIG. 2: Form factors /+ (and /o) for semileptonic D decays, 
from lattice QCD 0, [2g, expressed as a red (blue) curve with 
an orange (gray) statistical error band and a yellow (light 
blue) combined error band. Error bands take correlations 
into account. Measurements of /+ are from Belle (green di- 
amonds) '2^, BaBar (magenta squares) (SJ, and CLEO-c 
(maroon triangles) ,2^, i2£] . The vertical line shows 

TABLE IV: Best-fit values at and correlation matrix pki of 
the 3-term z expansion of /+ for B nil/, with statistical 
and systematic errors combined 





















of this paper. Now the curve and error band conform 
with preconceptions, for several reasons. First, q^^ ^ is 

close to Qn-jaxi rather than in the middle of the kinematic 
range. Second, the chiral extrapolation in Ref. is less 
aggressive than that in Ref. [a], leading to a larger but 
more realistic error at 5^ = 0. The most striking aspect 
is that even though the absolute error in is increasing 
for > 0.8, the band remains narrow. The band simply 
conveys the point-to-point correlations better. 

This paper shows in detail how to compare semilep- 
tonic form factors from lattice-QCD and from experi- 
ments. For illustration we use the BK parametrization 

2 2 
1 . 

I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' 


lattice QCD [Fennilab/MILC, 08 1 1 .3640 [hep-lat]] 

■ experiment [BaBar, hep-ex/06 12020] 


■fmax ■ 

FIG. 3: Form factor /+ for B —> nlv expressed as a curve 
(red) from the best fit with a total error band (yellow) from 
taking correlations in the fit parameters into account over- 
laid with measurements of \Vub\f+/{3.38 x 10~^ 
(magenta squares) [30|]. The vertical line shows g, 

from BaBar 


max ■ 


for D Klv and D ttIu, and the z expansion for 
B ttIv. Clearly, the idea is more general. For ex- 
ample, an interesting prospect relevant to semileptonic 
form factors is to inject 3-momenta smaller that P(i,o,o) 
using "twisted" boundary conditions [3l|, [H, [s^l . That 
strategy should improve the accuracy of parameters in 
the chiral extrapolation and, hence, the BK, BZ, or z 
fits. The output of any fit could still be exhibited as 
outlined here, although one should bear in mind that su- 
perior visualization of a fitting procedure does not repair 
any shortcomings of the fit itself. 


We would like to thank Ian Shipsey for encouraging us 
to think carefully about the correlations in the systematic 
errors. We would like to thank Laurenz Widhalm for pro- 
viding the Belle data in numerical form 2^] , and Shi psey 
for the BaBar and CLEO D-decay data [13, IH, llj. 

Computations for this work were carried out in part on 
facilities of the USQCD Collaboration, which are funded 
by the Office of Science of the United States Depart- 
ment of Energy. This work was supported in part by 
the U.S. Department of Energy under Grants No. DE- 
FC02-06ER41446 (CD., L.L., M.B.O.), No. DE-FG02- 
91ER40661 (S.G.), No. DE-FG02-91ER40677 (A.X.K., 
R.T.E., E.G.), No. DE-FG02-91ER40628 (C.B., J.L.), 
and No. DE-FG02-04ER41298 (D.T.); by the Na- 
tional Science Foundation under Grants No. PHY- 
0555243, No. PHY-0757333, No. PHY-0703296 (C.D., 
L.L., M.B.O.), No. PHY-0555235 (J.L.), and No. PHY- 
0757035 (R.S.); and by Universities Research Associates 
(R.T.E., E.G.). This manuscript has been coauthored 
by an employee of Brookhaven Science Associates, LLC, 
under Contract No. DE-AC02-98CH10886 with the U.S. 
Department of Energy. Fermilab is operated by Fermi 
Research AUiance, LLC, under Contract No. DE-AC02- 
07CH11359 with the U.S. Department of Energy. 

[1] M. Ablikim et al. [BES Collaboration], Phys. Lett. B 

597, 39 (2004) [arXiv:hep-ex/0406028]. 
[2] G. S. Huang et al. [CLEO Collaboration], Phys. Rev. 

Lett. 94, 011802 (2005) [arXiv:hep-ex/0407035]. 
[3] J. M. Link et al. [FOCUS Collaboration], Phys. Lett. B 

607, 51 (2005) [arXiv:hep-ex/0410068]. 
[4] J. M. Link et al. [FOCUS Collaboration], Phys. Lett. B 

607, 233 (2005) [arXiv:hep-ex/0410037]. 
[5] D. Besson et al. [CLEO Collaboration], Phys. Rev. D 80, 

032005 (2009) [arXiv:0906.2983 [hep-ex]]. 
[6] C. Aubin et al. [Fermilab Lattice, MILC, and HPQCD 
Collaborations], Phys. Rev. Lett. 94, 011601 (2005) 
[7] M. Okamoto et al. [Fermilab Lattice, MILC, and HPQCD 
Collaborations], Nucl. Phys. Proc. Suppl. 140, 461 
(2005) [arXiv:hep-lat/0409116]. 
[8] E. Dalgic nee Gulez et al. [HPQCD Collaboration], Phys. 
Rev. D 73, 074502 (2006) [arXiv:hep-lat/0601021]; 75, 
119906(E) (2007). 
[9] J. A. Bailey et al, Phys. Rev. D 79, 054507 (2009) 
[arXiv:0811.3640 [hep-lat]]. 
[10] K. C. Bowler et al. [UKQCD Collaboration], Phys. Lett. 

B 486 (2000) 111 [arXiv:hep-lat/9911011]. 
[11] A. Abada et al, Nucl. Phys. B 619 (2001) 565 

[12] A. X. El-Khadra et al, Phys. Rev. D 64, 014502 (2001). 

[13] S. Aoki et al [JLQCD Collaboration], Phys. Rev. D 64 

(2001) 114505 [arXiv:hep-lat/0106024]. 
[14] J. Shigemitsu et al, Phys. Rev. D 66, 074506 (2002) 

[15] D. Becirevic, S. Prelovsek, and J. Zupan, Phys. Rev. D 

67, 054010 (2003) [arXiv:hep-lat/0210048]. 
[16] C. Aubin and C. Bernard, Phys. Rev. D 76, 014002 

(2007) [arXiv:0704.0795 [hep-lat]]. 
[17] L. Lellouch, Nucl. Phys. B 479, 353 (1996) [arXiv:hep- 

[18] D. Becirevic and A. B. Kaidalov, Phys. Lett. B 478, 417 

(2000) [arXiv:hep-ph/9904490]. 
[19] P. Ball and R. Zwicky, Phys. Rev. D 71, 014015 (2005) 

[20] R. J. Hill, Phys. Rev. D 73, 014012 (2006) [arXiv:hep- 


[21] T. Becher and R. J. Hill, Phys. Lett. B 633, 61 (2006) 

[22] C. Bourrely, B. Machet, and E. de Rafael, Nucl. Phys. B 

189, 157 (1981). 
[23] C. G. Boyd, B. Grinstein, and R. F. Lebed, Phys. 

Rev. Lett. 74, 4603 (1995) [arXiv:hep-ph/9412324]; 

C. G. Boyd and M. J. Savage, Phys. Rev. D 56, 303 

(1997) [arXiv:hep-ph/9702300]. 
[24] L. Widhalm et al [Belle Collaboration], Phys. Rev. Lett. 

97, 061804 (2006) [arXiv:hep-ex/0604049]; K. Abe et al 
[Belle Collaboration], arXiv:hep-ex/0510003. 

[25] A. S. Kronfeld et al [Fermilab Lattice, MILC, and 
HPQCD Collaborations], PoS LAT2GG5, 206 (2006) 
[Int. J. Mod. Phys. A 21, 713 (2006)] [arXiv:hep- 
lat/0509169]; A. S. Kronfeld [Fermilab Lattice, MILC, 
and HPQCD Collaborations], J. Phys. Conf. Ser. 46, 147 
(2006) [arXiv:hep-lat/0607011]. 

[26] M. Okamoto [for the Fermilab Lattice and MILC Col- 
laborations], unpublished (2005). A public report of this 
work is contained in Ref. 0). This line of analysis was 
superseded by Ref. Q. 

[27] B. Aubert et al [BaBar Collaboration], Phys. Rev. D 76, 
052005 (2007) [arXiv:0704.0020 [hep-ex]]. 

[28] D. Cronin-Hennessy et al. [CLEO Collaboration], Phys. 
Rev. Lett. 100, 251802 (2008) [arXiv:0712.0998 [hep-ex]]; 
S. Dobbs et al [CLEO Collaboration], Phys. Rev. D 77, 
112005 (2008) [arXiv:0712.1020 [hep-ex]]. 

[29] J. Y. Ge et al [CLEO Collaboration], Phys. Rev. D 79, 
052010 (2009) [arXiv:0810.3878 [hep-ex]]. 

[30] B. Aubert et al. [BaBar Collaboration], Phys. Rev. Lett. 

98, 091801 (2007) [arXiv:hep-ex/0612020]. 


[31] P. F. Bodaquc, Phys. Lett. B 593, 82 (2004) [arXiv:nucl- [33] J. M. Flynu, A. Jiittncr, and C. T. Sachrajda [UKQCD 

th/0402051]. Collaboration], Phys. Lett. B 632, 313 (2006) [arXiv:hep- 

[32] C. T. Sachrajda and G. Villadoro, Phys. Lett. B 609, 73 lat/0506016]. 
(2005) [arXiv:hep-lat/0411033].