Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Gluonic observables and the scalar spectrum of twelve-flavour QCD
Enrico Rinaldi
on behalf of the LatKMI collaboration
Higgs Centre for Theoretical Physics, School of Physics and Astronomy University of Edinburgh
Lattice 2013, Mainz
Kobayashi-Maskawa Institutefor the Origin of Particles and the Universe素粒子宇宙起源研究機構
素粒子宇宙起源研究機構
Kobayashi-Maskawa Institute for the Origin of Particles and the Universe
Kobayashi-Maskawa Institutefor the Origin of Particles and the Universe素粒子宇宙起源研究機構
素粒子宇宙起源研究機構
Kobayashi-Maskawa Institute for the Origin of Particles and the Universe
LatKMI collaboration
Y. Aoki, T. Aoyama, M. Kurachi, T. Maskawa, K. Miura, K.-i. Nagai, H. Ohki,
E. Rinaldi, K. Yamawaki, T. Yamazaki
A. Shibata
Kobayashi-Maskawa Institutefor the Origin of Particles and the Universe素粒子宇宙起源研究機構
素粒子宇宙起源研究機構
Kobayashi-Maskawa Institute for the Origin of Particles and the Universe
Kobayashi-Maskawa Institutefor the Origin of Particles and the Universe素粒子宇宙起源研究機構
素粒子宇宙起源研究機構
Kobayashi-Maskawa Institute for the Origin of Particles and the Universe
Motivations
• a Higgs boson has been found at LHC
• a “composite” Higgs-like scalar can also reproduce the LHC data
• this is realised in Walking Technicolor models
• interest in finding the lower edge of the Conformal Window
• a light scalar appears as pseudo-NG boson
ATLAS, PLB716 (2012) 1-29
CMS, PLB716 (2012) 30-61
Motivations
• a Higgs boson has been found at LHC
• a “composite” Higgs-like scalar can also reproduce the LHC data
• this is realised in Walking Technicolor models
• interest in finding the lower edge of the Conformal Window
• a light scalar appears as pseudo-NG boson
_(µ)
µ
_(µ)
µ
_(µ)
µ
Nf
QCD-like
“walking”
conformal
Lattice simulations of twelve-flavour QCD
• SU(3) gauge field + 12 dynamical fermions
• gauge action → tree level Symanzik
• fermion action → HISQ (Highly Improved Staggered Quarks)
• modified HMC based on MILC code (v7)
• fixed gauge coupling → β = 4.0
• four volumes → V=183, 243, 303, 363
• six bare fermion masses → amf = 0.05 - 0.16
• long trajectories → >15000 MD units
L T amf confs.18 24 0.06 309018 24 0.08 525018 24 0.10 513018 24 0.12 520018 24 0.16 420024 32 0.05 944024 32 0.06 1400024 32 0.08 1534024 32 0.10 948030 40 0.05 750030 40 0.06 1400030 40 0.08 1145030 40 0.10 500036 48 0.05 510036 48 0.06 6000
Glueball spectroscopy
• the glueball spectrum is well studied in SU(N) Yang-Mills theories
• the lightest glueballs are starting to be investigated in full LQCD
• this talk focus on many-flavour QCD
• typically heavy mass, determined by ΛQCD
• typically noisy and require large statistic
21
TABLE XXI: The final glueball spectrum in physical units.In column 2, the first error is the statistical uncertainty com-ing from the continuum extrapolation, the second one is the1% uncertainty resulting from the approximate anisotropy.In column 3, the first error comes from the combined uncer-tainty of r0MG, the second from the uncertainty of r!1
0 =410(20) MeV
JPC r0MG MG (MeV)0++ 4.16(11)(4) 1710(50)(80)2++ 5.83(5)(6) 2390(30)(120)0!+ 6.25(6)(6) 2560(35)(120)1+! 7.27(4)(7) 2980(30)(140)2!+ 7.42(7)(7) 3040(40)(150)3+! 8.79(3)(9) 3600(40)(170)3++ 8.94(6)(9) 3670(50)(180)1!! 9.34(4)(9) 3830(40)(190)2!! 9.77(4)(10) 4010(45)(200)3!! 10.25(4))(10) 4200(45)(200)2+! 10.32(7)(10) 4230(50)(200)0+! 11.66(7)(12) 4780(60)(230)
In the tensor channel, the glueball matrix element isextrapolated to 1.0±0.2 GeV3 in the continuum, which isthe average of results of E and T2 channels. In the calcu-lation, it is found that in the lattice spacing range we use,the glueball mass and matrix elements are approximatelyindependent of the lattice spacing, this implies that thelattice artifacts might be neglected here. If the renor-malization constant ZT ! 0.52(15) of the tensor operatordoes not change much in the range of lattice spacing andapplies to all the ! values in this work, the renormalizedmatrix element of tensor operator is 0.52 ± 0.19 GeV3,which is in agreement with the prediction 0.35 GeV3 fromthe tensor dominance model [17] and QCD sum rule [18]for the tensor mass around 2.2 GeV.
VII. Conclusion
The glueball mass spectrum and glueball-to-vacuummatrix elements are calculated on anisotropic lattices inthis work. The calculations are carried out at five latticespacings as’s which range from 0.22 fm to 0.10 fm. Dueto the implementation of the improved gauge action andimproved gluonic local operators, the lattice artifacts arehighly reduced. The finite volume e!ects are carefullystudied with the result that they can be neglected on thelattices we used in this work.
As to the glueball spectrum, we have carried out cal-culations similar to the previous work [4] on much largerand finer lattices, so that the liability of the continuumlimit extrapolation is reinforced. Our results of the glue-
ball spectrum is summarized in Tab. XXI and Fig. 16.After the non-perturbative renormalization of the local
gluonic operators, we finally get the matrix elements ofscalar(s), pseudoscalar(p), and tensor operator (t) with
0
2
4
6
8
10
12
--+--+++ 0
1
2
3
4
5
r 0 M
G
MG
(GeV
)
0++
2++
3++
0-+
2-+
0+-
1+-
2+-
3+- 1--2--3--
FIG. 16: The mass spectrum of glueballs in the pure SU(3)gauge theory. The masses are given both in terms of r0
(r!10 = 410 MeV) and in GeV. The height of each colored
box indicates the statistical uncertainty of the mass.
the results
s = 15.6 ± 3.2 (GeV)3
p = 8.6 ± 1.3 (GeV)3
t = 0.52 ± 0.19 (GeV)3, (62)
where the errors of s and t come mainly from the errorsof the renormalization constants ZS and ZT . The moreprecise calculation of ZS and ZT will be carried out inlater work.
Acknowledgments
This work is supported in part by U.S. Departmentof Energy under grants DE-FG05-84ER40154 and DE-FG02-95ER40907. The computing resources at NERSC(operated by DOE under DE-AC03-76SF00098) and SC-CAS (Deepcomp 6800) are also acknowledged. Y. Chenis partly supported by NSFC (No.10075051, 10235040)and CAS (KJCX2-SW-N02). C. Morningstar is also sup-ported by NSF under grant PHY-0354982.
[1] B. Berg and A. Billoire, Nucl. Phys. B221, 109 (1983). [2] G. Bali, et al. (UKQCD Collaboration), Phys. Lett. B
Chen et al., PRD73(2006)
0
0.5
1
1.5
2
2.5
3
am
Ground StatesExcitations
A1++ A1
-+ A2+- E++ E-+ T1
+-T1++ T1
-+ T2++ T2
--T2-+
Lucini, Rago, ER, JHEP08(2010)
Glueball spectroscopy
• the glueball spectrum is well studied in SU(N) Yang-Mills theories
• the lightest glueballs are starting to be investigated in full LQCD
• this talk focus on many-flavour QCD
• typically heavy mass, determined by ΛQCD
• typically noisy and require large statistic
Gregory, Irving, Lucini, McNeile, Rago, Richards, ER, JHEP10(2012)
Glueball spectroscopy
• the glueball spectrum is well studied in SU(N) Yang-Mills theories
• the lightest glueballs are starting to be investigated in full LQCD
• this talk focus on many-flavour QCD
• typically heavy mass, determined by ΛQCD
• typically noisy and require large statistic
Gregory, Irving, Lucini, McNeile, Rago, Richards, ER, JHEP10(2012)
Glueball spectroscopy: operators
• eigenstates of the Hamiltonian are classified according to the irreducible representations of the cubic group
• suitable gauge-invariant operators must be constructed that respect the symmetries
• vacuum contributions must be subtracted in the scalar case
• improved operators are obtained by blocking and smearing algorithms Lucini, Rago, ER
JHEP08(2010)
JHEP08(2010)119
J A1 A2 E T1 T2
0 1 0 0 0 0
1 0 0 0 1 0
2 0 0 1 0 1
3 0 1 0 1 1
4 1 0 1 1 1
Table 1. Subduced representations J ! GO of the octahedral group up to J = 4. This tableillustrates the spin content of the irreducible representations of GO in terms of the continuum J .
3 Extracting glueball masses
In this section we present the construction of our operators and we review the general
methodology for extracting glueball masses. While the standard variational procedure and
the construction of operators in irreducible representations of the cubic lattice group is well
known, this is, up to our knowledge, the first systematic attempt of inserting scattering
and torelon operators into the variational set, in order to rule out from the spectrum
contributions of these spurious states.
Symmetries of the lattice spectrum. At finite lattice spacing a the continuum rota-
tion group is not an exact symmetry of the system. The full continuum rotational symmetry
is dynamically restored only when a " 0. On the lattice, eigenstates of the Hamiltonian
have to fall into the irreducible representations of the octahedral point group GO, the sym-
metry group of the cube. The octahedral point group has 5 irreducible representations A1,
A2, E, T1 and T2 respectively with dimensions 1, 1, 2, 3, 3.
Since we are interested in the glueball spectrum of the gauge theory in the continuum,
we need to consider GO as a subgroup of the complete rotation group SO(3): irreducible
representations of SO(3) are decomposed in terms of those of GO. Irreducible representa-
tions of integer spin J in SO(3) restricted to GO are referred to as subduced representations
J ! GO. When considered as a representation of GO, the (2J +1) degeneracy of the contin-
uum spin J state is split onto di!erent irreducible representations of GO. A simple example
of this kind of pattern is the spin 2 (tensor) glueball, whose 5 polarisations are seen on
the lattice as di!erent states, 2 in the E and 3 in the T2 representation of GO. Due to
the breaking of continuum rotational symmetry on the lattice, the aforementioned pattern
of degeneracies is exact in the limit a " 0, but it is only approximate at finite a. Com-
paring the measured glueball spectrum with the expected pattern of degeneracy can give
information on the relevance of lattice artifacts.
Near the continuum limit, it is possible to identify the masses of spin J glueballs by
matching the patterns of degeneracies of the subduced representations J ! GO from the
degeneracy coe"cients. We report these coe"cients up to J = 4 in table 1. For any given
operator O on the lattice, we define a rotation transformation as Ri(O) where the index i
labels all the elements of the group GO.
Since a generic representation of the group will not be irreducible, in order to create
states that transform only in a given symmetry channel, we will need to create an appro-
– 4 –
{A1(1), A2(1), E(2), T1(3), T2(3)}
OG
(t) =1
L3
X
x2L
3
Tr
0
@Y
l2W(x)
Ul
1
A O(R)G (t) =
24X
↵=1
a(R)↵ R↵ [OG(t)]
O(A1)(t)� h0|O(A1)|0i
• basis of operators ➔
• matrix of correlators ➔
• generalised eigenvalue problem ➔
• ground state correlator fit (α=0) ➔
Glueball spectroscopy: variational analysis
Cij(t) =X
⌧
h0|O†i (⌧ + t)Oj(⌧)|0i
{O1(t), . . . ,On(t)}
Cij(t)v↵j = �↵v↵i
h�†↵(t)�↵(0)i = |c↵|2
⇣e�m↵t + e�m↵(T�t)
⌘
�↵(t) =nX
i=1
v↵i Oi(t)
• the effective mass plateaux is used to determine the fitting window on the correlator
Example: scalar glueball effective mass
The fitted mass is shown on top of the corresponding effective mass
enabling us to establish the goodness of the estimate
0 1 2 3 4 5t/a
0
0.2
0.4
0.6
0.8
1
am
eff
L=18 #5130 cfgs. 171 bins
L=24 #9480 cfgs. 237 bins
L=30 #5000 cfgs. 100 bins
Scalar glueball effective mass !=4.0 amf=0.10
Finite-size effects appear to be negligible at larger fermion masses
Results: scalar glueball spectrum
• The scalar glueball mass decreases with the bare fermion mass
• it is expected to become independent of mf in the quenched limit
• at light quark mass and on small volumes it is consistent with the mass of a fermionic bound state [talk by T.Yamazaki 3F]
• finite volume effects and mixing with a light scalar fermionic state need to be better investigated 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
amf
0
0.2
0.4
0.6
0.8
1
aM
G
L=18L=24L=30L=36! (L=30)" (L=30)
Pseudoscalar and vector data are taken from LatKMI, PRD86(2012)054506
Results: scalar glueball spectrum
• The scalar glueball mass decreases with the bare fermion mass
• it is expected to become independent of mf in the quenched limit
• at light quark mass and on small volumes it is consistent with the mass of a fermionic bound state [talk by T.Yamazaki 3F]
• finite volume effects and mixing with a light scalar fermionic state need to be better investigated
Pseudoscalar and vector data are taken from LatKMI, PRD86(2012)054506Scalar data are taken from LatKMI, arxiv:1305.6006
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
0
0.2
0.4
0.6
0.8
1
aM
G
L=18L=24L=30L=36! (L=30)" (L=30)# (L=30)
Example: spatial torelon effective mass (amf=0.06)
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
1
1.2
am
eff
L=18L=24L=30L=36am=0.1142(79) [7:11]
am=0.273(31) [7:12]
am=0.533(37) [4:7]
am=0.657(52) [3:5]
amtor
(L) = a2�L� ⇡
3L� ⇡2
18L3
1
a2�• The torelon ground state mass
is obtained from a variational analysis of smeared spatial Polyakov loop correlators
• The string tension is extracted from the L dependence of the torelon ground state mass
• A stable effective mass plateaux is found for each volume and the fitted mass is shown on top of the effective mass data
• The formula applied for the largest volumes implies a common string tension value
Example: spatial torelon effective mass (L=24)
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
1
1.2
am
eff
amf=0.05
amf=0.06
amf=0.08
amf=0.10
am=0.158(11) [7:12]
am=0.273(31) [7:12]
am=0.615(29) [3:6]
am=0.820(82) [3:5]
• Torelon effective masses on a fixed volume show the bare fermion mass dependence
• When the bare quark mass increases we obtain a larger torelon mass
amtor
(L) = a2�L� ⇡
3L� ⇡2
18L3
1
a2�
Results: string tension
• The string tension is obtained for the full range of simulated masses and volumes with 1% to 4% statistical errors
• The finite-size effects seem under control as date on two volumes with compatible string tension exist for all light fermion masses
• There is a clear dependence of the string tension on the bare fermion mass
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
0
0.1
0.2
a!
1/2
L=18
L=24L=30
L=36
Scaling of the string tension
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
0
0.1
0.2
a!
1/2
L=18L=24L=30L=36(L=30): 0.05(1)+1.5(2)*(am
f)
(L=30): 0.9(2)*(amf)^(0.66(9))
ap� = c am
11+�
fap� = c0 + c1 amf
Scaling of the string tension
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
0
0.1
0.2
a!
1/2
L=18L=24L=30L=36(L=30): 0.05(1)+1.5(2)*(am
f)
(L=30): 0.9(2)*(amf)^(0.66(9))
ap� = c am
11+�
fap� = c0 + c1 amf
γ∼0.5 is compatible with the mesonic spectrum
c0 = 0.05 implies confinement in the chiral limit
Scaling of the string tension
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
0
0.1
0.2
a!
1/2
L=18L=24L=30L=36(L=30): 0.05(1)+1.5(2)*(am
f)
(L=30): 0.9(2)*(amf)^(0.66(9))
ap� = c am
11+�
fap� = c0 + c1 amf
Relevant region to study the chiral behaviour
Region with higher order corrections
Ratios of gluonic and mesonic scales
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
1
1.5
2
2.5
3
3.5
4
m!/"
1/2
L=18L=24L=30L=36
Fermionic pseudoscalar state and string tension
Ratios of gluonic and mesonic scales
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18am
f
1
1.5
2
2.5
3
3.5
4
m!/"
1/2
L=18L=24L=30L=36
Fermionic pseudoscalar state and string tension
Ratios of gluonic and mesonic scales
0 0.02 0.04 0.06 0.08 0.1 0.12am
f
0
0.5
1
1.5
2
2.5
3
3.5
4
m!/!
1/2
L=24L=30L=36SU(3) pure gauge
Flavour-singlet fermionic scalar state and string tension
Ratios of gluonic and mesonic scales
0 0.02 0.04 0.06 0.08 0.1 0.12am
f
0
0.5
1
1.5
2
2.5
3
3.5
4
m!/!
1/2
L=24L=30L=36SU(3) pure gauge
Flavour-singlet fermionic scalar state and string tension
Conclusions
• we study the scalar spectrum in many-flavour QCD, focused on Nf=12
• scalar glueball and scalar meson interpolating operators are used
• both glueball operators and mesonic operators couple to this state
• important to understand the mixing of such contributions
• we measure the string tension in Nf=12 QCD and its fermion mass dependence
• it is difficult to extrapolate the results to the chiral limit
• ratios of fermionic pseudoscalar and scalar state with the string tension are compatible with a constant value, down to the chiral limit
Outlook
• larger volumes at light quark masses are necessary to carefully estimate finite size errors
• possible discretisation effects need to be studied by simulating at different lattice spacing
• lighter quark masses are needed to better identify scaling properties of the string tension
• a variational analysis including both gluonic and mesonic operators in the scalar channel is needed: it will give informations about the mixing between the two kind of operators
• a similar study is in progress on Nf=8 QCD configurations [talks by K.Nagai and H.Ohki 1F]
Thank you
Appendix
additional slides with results
Comparison with gluonic observables Preliminary
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
am
eff
gluonic operators only
fermionic operators only
Scalar spectrum L=18 T=24 !=4 amf=0.06
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
am
eff
gluonic operators only
fermionic operators only
Scalar spectrum L=24 T=32 !=4 amf=0.06
L = 18 and amf = 0.06 L = 24 and amf = 0.06
Comparison with gluonic observables Preliminary
L = 18 and amf = 0.08 L = 24 and amf = 0.08
0 2 4 6t/a
0.2
0.4
0.6
0.8
1
am
eff
gluonic operators only
fermionic operators only
Scalar spectrum L=18 T=24 !=4 amf=0.08
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
1
am
eff
gluonic operators only
fermionic operators only
Scalar spectrum L=24 T=32 !=4 amf=0.08
Comparison with gluonic observables Preliminary
L = 30 and amf = 0.06 L = 30 and amf = 0.08
0 2 4 6 8 10t/a
0
0.2
0.4
0.6
0.8
1
am
eff
gluonic operators only
fermionic operators only
0 2 4 6 8 10 12t/a
0
0.2
0.4
0.6
0.8
1
am
eff
gluonic operators only
fermionic operators only