Skip to main content

Full text of "Electromagnetic splittings of hadrons from improved staggered quarks in full QCD"

See other formats












Electromagnetic splittings of hadrons from 
improved staggered quarks in full QCD 

S. Basak*, at A. Bazavov, h C. Bernard, c C. DeTar/ W. Freeman, fo Steven Gottlieb/ 
U.M. Heller/ J.E. Hetrick/ J. Laiho/ L. Levkova, rf J. Osborn/ R. Sugar/ 1 and 
D. Toussaint'' (MILC Collaboration) 

a Department of Physics, Indiana University 

Bloomington, IN 47405, USA 
b Physics Department, University of Arizona 

Tucson, AZ 85721, USA 

c Physics Department, Washington University 

St. Louis, MO 63130, USA 

Physics Department, University of Utah 

Salt Lake City, UT 84112, USA 
e American Physical Society 

One Research Road, Box 9000, Ridge, NY 11961-9000, USA 
f Physics Department, University of the Pacific 
Stockton, CA 95211, USA 

8 Argonne Leadership Computing Facility, Argonne National Laboratory 
Argonne, IL 60439, USA 

h Physics Department, University of California 
Santa Barbara, CA 93106, USA 


We present our initial study of the electromagnetic splittings of charged and neutral mesons, 
and the violation of Dashen's theorem. Hadron masses are calculated on MILC Nf = 2 + 1 QCD 
ensembles at lattice spacing w 0.15 fm, together with quenched non-compact U{\) configurations. 
The ff{a 2 ) tadpole improved staggered quark (asqtad) action is used both for the sea quarks 
and for six different valence quark masses. Chiral extrapolations are performed using partially 
quenched chiral perturbation theory incorporating electromagnetic corrections. 

The XXVI International Symposium on Lattice Field Theory 
July 14- 19, 2008 
Williamsburg, Virginia, USA 

* Speaker. 

'Presently at NISER, Bhubaneswar, Orissa 751005, India 

© Copyright owned by the author(s) under the terms of the Creative Commons Attribntion-NonCommercial-ShareAlike Licence. 

http: //pos . sissa . it/ 

Electromagnetic splittings 

S. Basak 

1. Introduction 

Lattice QCD has long been used to determine nonperturbative hadronic properties; the spec- 
trum, in particular, has been extensively investigated. However, electromagnetic and isospin- 
breaking effects are usually not included in those investigations. Although the strong interaction 
is blind to the electromagnetic charges, we do in fact have mass splittings of mesons and baryons 
in the real world which depend on both isospin-breaking and electromagnetic interactions. Fur- 
ther, it has been pointed out that the current evaluation of the light quark masses, particularly the 
ratio mu/md, suffers significant uncertainties coming from the electromagnetic contributions to the 
masses of K and especially K mesons [jl|]. 

It is therefore important to include electromagnetic effects in the lattice QCD simulations 
for more realistic spectrum calculations. The pioneering attempt to calculate charged and neutral 
pion splittings and the light quark masses from lattice QCD with electromagnetic interactions was 
by Duncan, Eichten and Thacker [Q]. The photon fields were introduced in non-compact form 
and were treated in the quenched approximation. The pseudoscalar meson masses were calcu- 
lated with the Wilson action and at different values of electric charge. Recently, Blum et al. [§] 
have calculated electromagnetic contributions to % and K mesons mass splittings and determined 
the light quark masses with domain wall quarks and Nf = 2 QCD configurations at the physical 
value of electric charge. They have also estimated the correction to the Dashen's theorem [Q] at 
&(am q ), and found A em = 0.337(40) or 0.264(43) depending on their fitting range. Electromag- 
netic splittings of 71 and p mesons have also been calculated using a RG-improved gauge action 
and a mean-field improved clover quark action at two different volumes and three lattice spacings 
in Ref. [Qj. Electromagnetic effects on baryons have been discussed in Refs. [||, [7]]; and calculation 
with unquenched photon fields in Ref. [§]. 

In this work, we study the electromagnetic mass splittings of pseudoscalar mesons in the pres- 
ence of a quenched electromagnetic field and calculate the correction to the Dashen's theorem, 
parametrized following the chiral perturbation theory calculation of Urech [|9p. The lattice data is 
fitted using partially quenched chiral perturbation theory formulas at &{p A , e 2 m q ) given by Bijnens 

et al. [ICQ. We have used a non-compact action for the photon fields following [g]. The valence 

quarks are asqtad [ |11[ ] staggered quarks, and the configurations are the MILC TV/ = 2 + 1 QCD 

configurations at lattice spacing 0.15 fm, with varying sea quark masses [12]. 

2. Lattice QCD with electromagnetic interaction 

The mass differences among the members of hadron isomultiplets arise from two sources: (a) 
strong isospin-breaking contributions due to the difference in masses of the light quarks and (b) 
different electromagnetic charges of the quarks. Apart from a small isospin-breaking contribution 
of order {m c { — m u ) 2 , the n ± — 71° mass splitting is almost purely electromagnetic in origin; whereas 
it is the isospin-breaking contributions that dominate for K° — K. Dashen's theorem [Q] summa- 
rizes the electromagnetic effects on meson masses: in the chiral limit to ff(e ), electromagnetic 
interactions modify the leading order M 2 ± and M\± masses while the n° and K° masses remain 


Electromagnetic splittings 

S. Basak 


Ml ± =2mB Q +A^e 2 (q u -q d ) 2 , 


M 2 ± = {m + m s )B +A^e 2 (q u - q s 




where q u , q c {, and q s are the u, d, and s quark charges in units of e. At &{e 2 m), however, M 2 ±0 
and M 2 K± can receive large and different contributions, and the correction to Dashen's theorem can 
parametrized as, 

AM 2 

AM% — AM^ 




or as A^, defined by AM^ = (1 + A^) AM 2 . The partially quenched chiral perturbation theory rele- 
vant for QCD + QED with 2+ 1 dynamical flavors at NLO has recently been worked out by Bijnens 
et al. [10]; it can be used to perform fits to the lattice M 2 K data in order to determine the electro- 
magnetic low energy constants. The pure electromagnetic correction, relevant for calculating the 
correction to Dashen's theorem, can be computed from the expression, 

AM 2 | em = Ml s (xi,X3,qi,q3)-M 2 s (xi,Xi,l3,q3) 

- M% B (xuXuquq3)+M% s (xi,Xu<i3,q3)- 


In this notation [ 1C], the normalized quark mass is Xi = 2Bo'W;, where Bq is related to the quark- 

antiquark vacuum expectation value in the chiral limit, and M 2 f .(x\,X3^i^3) denotes the squared 
pseudoscalar meson mass having valence quark masses Xi>X3 and valence charges qi,qs in units 
of electron charge e. In the isospin limit m u = mj, when Xi = m u-, X3 = m s and the quark charges 
<?i = <7n; <?3 = <?.v> we have AMp = AM 2 | em . We fit the lattice data for AM^ according to 


A{X\3 ~ Xn) + ^2 

+ $?3 





where, Xij = (Xi + an d WQ the constants to be determined from the fit. For the scale /i 
we have used lOOOMeV. The sea-quark dependence of the log-terms appears through Xif-> where 
the sea-quark index / is summed over all the sea quarks. Terms with the sea-quark charges con- 
tributing to AM^ do not involve unknown low energy constants (LECs) at this order and hence are 
computable without lattice simulation [|To[]. In our expression we have included the parameters 
^2 3 4, which are not present in Ref. [jlCP, in front of the logarithm terms. This is because we ex- 
pect the continuum expression to be modified by discretization effects on the quite coarse lattices 
we are using. Finite volume effects, due to the masslessness of the photon, may also be rather 
significant. At this stage, therefore, we can obtain at best a rough estimate of the quantity AM^; a 
precise determination will require lattice data at finer lattice spacings and larger volumes. 


Electromagnetic splittings 

S. Basak 

3. Numerical results 

Phenomenologically relevant quantities, such as the mass-squared differences of charged and 
neutral mesons and the correction to Dashen's theorem, can be calculated from partially quenched 
lattice QCD where photons are treated in the quenched approximation. Therefore, we can use 
the existing SU (3) configurations generated without dynamical photons. For unquenched QCD 
gluon configurations, we have used four ensembles of existing a 0.15fm MILC lattices with 2+1 
dynamical flavors, the details of which are provided in Table 1. The lattice spacing is set by the 

static quark potential [12]. 


JS = 10/g 2 

am /am s 

a (fm) 


20 3 x 48 





16 3 x 48 



« 0.15 


16 3 x 48 





16 3 x 48 





Table 1: N f = 2+ 1 MILC lattices used in this project. 

The quenched photon configurations {A„(n)} are generated from the non-compact U(l) action, 



subjected to Coulomb gauge fixing d-A^n) = 0. Additional global gauge fixing is done to en- 
sure that Gauss's law is satisfied. In momentum space, the action is Gaussian distributed, and the 
scalar and vector potentials [A (p), A(p)] can be generated independently of each other from Gaus- 
sian distributed random numbers. The coordinate-space Coulomb-gauge photon configurations are 
recovered by FFT The necessary valence quark propagators are calculated with wall sources in 
these SU (3) X U(l) background fields. We calculate the pseudoscalar meson propagator with nine 
different valence quark masses, O.lm^ <m q < 1.0m' s . All the fits we report here are obtained by 
correlated # 2 -fit; the errors are obtained from a jackknife analysis. The meson masses are extracted 
from the exponential fall-off of the meson propagators in the time range 9 - 24, taking into account 
the correlations among the time slices. We have ignored the contribution from the disconnected 
diagram that effects the neutral pion mass. 

We first try to estimate the &{e 2 ) and @{e 2 m) contribution to pseudoscalar masses at three 
different (physical and non-physical) values of electric charge, 

M\{e £ 0) -M\{e = 0) = M>e 2 (q u -q d f + , 

(e 2 m) 


In Fig. H, we plot the dependence of % mass-squared splittings corresponding to Eq. (3.2). A 
straight line fit describes the &{e 2 ) behavior fairly well over the full range of electric charges that 
are examined. In a further test to ascertain the &(e 2 m) or higher order effects, in Fig. we plot 
the variation of the same pion mass-squared splittings against quark masses. The slope of each of 
line, corresponding to different electric charges, nicely matches to its e 2 value. 


Electromagnetic splittings 

S. Basak 


Figure 1: The e dependence of mass-squared electromagnetic splittings of the pion. 

Figure 2: The ff(e 2 m) behavior of mass-squared electromagnetic splittings of the pion. 

In a series of four plots in Fig. f|, we show the meson mass splittings between mesons com- 
puted with physical electric charge e 2 = 47ia em and with e 2 = 0. The labels q\q~2 in the plots denote 
quark charges, i.e., ud indicates quarks with charge q u = 2e/3 and qj = —le/3. The points using 
the same symbol and color are obtained by varying the q2 quark mass. The lines are merely to 
guide the eye. We have not yet fit the squared meson masses to the Bijnens form based on partially 
quenched chiral perturbation theory (and hence not yet extracted the electromagnetic LECs) [|10|]. 
These LECs will be important in determining the physical isomultiplet splittings in mesons and 
therefore in calculating quark masses. 

The electromagnetic LECs can be used to obtain the correction to Dashen's theorem, but we 

have estimated this correction directly by chiral extrapolation of Eq. ( |2.3| ) using Eq. (|2.4|). We put 
<7i = Qu, <?3 = <?.v in Eq. ( |2.3| ) and extrapolate/interpolate to %\ = 2Bom u , X3 = 2Z?o m s- Figure || is 
our preliminary result for the violation of Dashen's theorem, where we have plotted the AM|, for 
two ensembles, /3 = 6.566 and 6.572, as a function of M 2 , obtained with e 2 = 0, in physical units. 
The data from the remaining two QCD ensembles at heavier quark mass are still being analyzed. 

We have determined the coefficients ^ in Eq. (2.4) using our lattice data for pseudoscalar masses 


Electromagnetic splittings 

S. Basak 

0.007 I 

3 0.006 - 

| 0.005 - 

S 0.004 - 

i „ 0.003 - 
T 0.002 - 
e" 0.001 - 


P = 6.566, m/m =0.00484/0.0484 

iJ 1 






6 0.006 


■| 0.005 
S 004 
n „ 0.003 

7 0.002 

g" oooi 

P s 6.572. mjta = 0.0097/0.0484 




m + m 





0.007 r 

3 0.006 - 


■g 0.005 - 

S 0.004 - 

« 0.003 - 
7 0.002 - 
e" 0.001 - 


P = 6.586, m/in = 0.0194/0.0484 

o : 

0.04 0.06 
m + m- 



3 0.006 

•| 0.005 
^ 0.004 
n . 0.003 
? 0.002 

a" o.ooi 

P = 6.600, m/m = 0.0290/0. (I4S4 

m . m 3) ^ 5 



. 5 $ 

I I I I \ 

0.04 0.06 0.08 

m + m- 

Figure 3: The mass-squared splittings between pseudoscalar mesons with physical value of electric charge 
e 2 = 4-KOem and e 2 = 0. Each panel corresponds to different j3 and light sea-quark mass am\ for fixed 
sea-strange quark mass am' s — 0.0484. The labels q\q2 denote the quark charges. We write s instead of d 
whenever m qi is equal to m' s , the strange sea-quark mass in the simulation. 

Xij, after which chiral extrapolation is performed at the experimental masses for n and K. Our 
preliminary result for the deviation from Dashen's theorem is 7.0 x < AM^ / (GeV 2 ) < 1.8 x 
10~ 3 . This may be compared with the estimate in Ref [1C]: 1.07 x 10~ 3 . Although we find that 
our value is consistent with the phenomenological estimate, the error in our result is large and does 
not yet include the systematic errors due to discretization and finite volume effects. We hope to 
reduce the error by increasing the statistics and extending the calculation to finer lattices and larger 
volumes. Fitting the meson splittings in Fig. |5] to determine the electromagnetic LECs directly may 
also provide better control over our errors. 

4. Conclusions and outlook 

In this work we have calculated the electromagnetic mass splittings of pseudoscalar masses 
with 0{a 2 ) improved staggered fermions ona» 0.015 fm Nf ■ = 2 + 1 MILC SU (3) ensembles and 
Coulomb-gauge U(l) configurations. We have determined the correction to Dashen's theorem at 
the physical point (in the isospin limit), but with relatively large errors, i.e., a wide range for AM^. 
Our determination of meson and baryon isomultiplet splittings and quark masses is in progress. 
Our immediate goal is using finer lattices to reduce the error in the correction to Dashen's theorem 
enough to bring down the present error in m u /md- 


Electromagnetic splittings 

S. Basak 



J5- 0.0000 


Figure 4: Correction to Dashen's theorem, the difference of electromagnetic contributions between K and 
% masses, as a function of the LO % mass squared (equivalent to the % mass squared with e 1 = 0). 


[1] MILC Collaboration: C. Aubin et al, Phys. Rev. D70, 114501 (2004) [hep-lat/0407028 ] . 

[2] A. Duncan, E. Eichten, H. Thacker, Phys. Rev. Lett. 76, 3894 (1996) [hep-lat / 9602005 ] . 

[3] T. Blum et al, Phys. Rev. D76, 1 14508 (2007) [0708 . 0484 (hep-lat) ] . 

[4] R.F. Dashen, Phys. Rev. 183, 1245 (1969). 

[5] ] Y. Namekawa, Y. Kikukawa, PoS (LAT2005) 090, 2005 [hep-lat/ 050 912 0] . 

[6] A. Duncan, E. Eichten, H. Thacker, Phys. Lett. B409, 387 (1997) [ hep-lat/ 9 60 7 32 ] . 

[7] T. Doi et al, PoS (LAT2006) 174, 2006 [hep-lat/0610095] . 

[8] A. Duncan, E. Eichten, R. Sedgewick, Phys. Rev. D71, 094509 (2005) [ hep-lat / 0405014]. 

[9] R. Urech, Nucl. Phys. B433, 234(1995) [hep-ph/9405341] . 

[10] J. Bijnens,N. Danielsson, Phys. Rev. D75, 014505 (2007) [hep-lat / 61 127 ] . 

[11] C. Bernard etal, Phys. Rev. D58, 014503 (1998) [hep-lat/9712010] . 

[12] C. Bernard et al, PoS (LAT2007) 090, 2007 [ 07 1 . 1 1 1 8 (hep-lat )] . 



P = 6.566 (m/n^ = 
P = 6.572 (m/m s = 







m ~ (GeVT