From Decay to Complete Breaking: Pulling the Strings in SU(2) Yang-Mills Theory M. Pepea and U.-J. Wieseb,c a INFN, Istituto Nazionale di Fisica Nucleare, Sezione di Milano-Bicocca Edificio U2, Piazza della Scienza 3 - 20126 Milano, Italy b Center for Research and Education in Fundamental Physics, Institute for Theoretical Physics, Bern University, Sidlerstr. 5, 3012 Bern, Switzerland c Institute for Theoretical Physics, ETH Zu¨rich, Schafmattstr. 32, CH-8093 Zu¨rich, Switzerland Westudy{2Q+1}-stringsconnectingtwostaticchargesQin(2+1)-dSU(2)Yang-Millstheory. Whilethefundamental{2}-stringbetweentwochargesQ= 1 isunbreakable,theadjoint{3}-string 2 connectingtwochargesQ=1canbreak. Whena{4}-stringisstretchedbeyondacriticallength,it decays into a {2}-string by gluon pair creation. When a {5}-string is stretched, it first decays into a{3}-string,whicheventuallybreakscompletely. Theenergyofthescreenedchargesattheendsof 9 a string is well described bya phenomenological constituent gluon model. 0 0 PACSnumbers: 11.15.Ha,12.38.Aw,12.38.Gc. 2 n a Studies of the strings connecting two static color be characterized by its N-ality k = 0,1,...,N 1 . { − } J chargesprovidevaluable insightsinto the physicsofcon- Strings connecting external charges with N-ality k = 0 6 6 finement in SU(N) Yang-Mills theories. The properties are known as k-strings. These strings are unbreakable 1 of the string connecting a static quark-anti-quark pair and have a k-dependent string tension, which may or withchargesinthe fundamental N and N represen- may not be proportional to the Casimir operator of the ] { } { } t tations are described by a low-energy effective bosonic corresponding representation [20, 21]. a string theory. While the string tension σ determines l - the quark-anti-quark potential V(r) =σr at asymptotic Since it is easiest to simulate numerically, in this let- p ter we study the dynamics of strings in (2+1)-d SU(2) e distances, the massless modes corresponding to trans- h verse fluctuations of the string give rise to the universal Yang-Millstheory whichhas the center Z(2). Other the- [ ories in (3+1) dimensions or with other gauge groups Lu¨scher term proportional to 1/r [1, 2], as well as to a are expected to show similar behavior. Here we investi- 1 diverging string thickness proportional to logr [3]. The v effectivestringtheoryalsomakesdetailedpredictionsfor gate the strings connecting two static charges Q in the 0 the excited states of the string [4]. Lattice gauge the- SU(2) representation 2Q + 1 , which we refer to as { } 1 2Q+1 -strings,notto be confused with k-strings. The ory provides us with a powerful tool for investigating 5 { } 2Q+1 -stringswithintegerQhavek =0andwilleven- the string dynamics using Monte Carlo simulations. In 2 { } tuallybreakatlargedistances,whilethe 2Q+1 -strings 1. this way, the linearly rising quark-anti-quark potential with half-integer Q have k = 1 and are u{nbreak}able. At has been calculated at large distances [5]. By develop- 0 asymptoticdistancesall 2Q+1 -stringsconnectinghalf- 9 ing ahighlyefficientmulti-levelsimulationtechnique [6], { } integer charges have the same tension σ as the funda- 0 Lu¨scher and Weisz have studied the universal 1/r-term : in the quark-anti-quarkpotential at large distances [7]. mental 2 -string. Since SU(2) Yang-Mills theory has v { } no dynamical fundamental charges, the static charges Q i X In theories with dynamical fundamental charges, the atthetwoendsofa 2Q+1 -stringcanonlybescreened r confining string connecting two static color charges can by dynamical gluons{. When} a pair of gluons is created a breakduetothecreationofdynamicalcharge-anti-charge from the vacuum, the external sources are screened and pairswhichscreentheexternalstaticsources. Numerical thus reduced to Q 1. As a consequence, the 2Q+1 - evidence for charge screening was obtained in a lattice − { } string decays to a 2Q 1 -string and abruptly reduces gauge-Higgs model in which the dynamical fundamen- its tension accordi{ngly−[22}, 23]. While some numerical tal charges are scalars [8]. Direct evidence for string evidenceforstringdecaywaspresentedin[24], usingthe breaking was first observed in the SU(2) gauge-Higgs multi-level simulation technique of [6], we are able, for model [9, 10] and later also in the Z(2) gauge-Higgs the first time, to investigate string decay in detail. model [11]. While the fundamental string is unbreak- able in SU(N) Yang-Mills theory, the string connecting We consider (2+1)-d SU(2) Yang-Mills theory on a static adjoint charges can break due to pair-creation of cubic lattice using the standard Wilson plaquette action dynamical gluons. This effect has been investigated in forlinkvariablesinthefundamentalrepresentation. The [12, 13, 14, 15, 16, 17, 18]. Numericalevidence for string external color charges Q at the two ends of a string are breakinginlatticeQCDwithdynamicalquarkshasbeen represented by Polyakov loops Φ (x) in the 2Q+ 1 Q { } observed in [19]. The center symmetry of SU(N) Yang- representation wrapping around the Euclidean time di- Mills theory is Z(N). Consequently, each SU(N) rep- rection. The corresponding potential V (r) between the Q resentation and hence each external static charge can static sources is extracted from the Polyakovloop corre- 2 lator 4 Φ (0)Φ (r) exp( βV (r)). (1) h Q Q i∼ − Q 3.5 In orderto ensure a goodprojectiononthe groundstate 3 of the string, we have simulated at inverse temperatures as large as β = 64 in lattice units. The spatial lattice 2.5 size was L = 32 and the bare gauge coupling was cho- sen as 4/g2 = 6.0 which puts the deconfinement phase 2 transition at β 4. While this is a moderate coupling, c ≈ we are confident that our results remain unchanged, at 1.5 fit least qualitatively, in the continuum limit. The values Q = 1/2 Q = 3/2 of the simulated Polyakov loop correlators range from 1 10−8 to 10−135. Measuring such small signals would be 2 4 6 8 10 12 14 16 5 completely impossible without the Lu¨scher-Weisz multi- level simulation technique. We have slightly refined this 4.5 method by applying the segmentation of the lattice not only to slabs in time, but also to blocks in space. By 4 carefully tuning the parameters of the multi-level algo- rithm, we havebeen able to extractthe potentials V (r) 3.5 Q forthe 2 -, 3 -, 4 -,and 5 -strings. Asshowninthe { } { } { } { } 3 top panel of figure 1, at distance r 8, the 4 -string ≈ { } decays, thus reducing its tension to the one of the fun- 2.5 fit damental 2 -string. Similarly, the bottom panel shows Q = 1 { } that,atdistancer 6,the 5 -stringdecaysandreduces Q = 2 ≈ { } 2 its tension to the one of the adjoint 3 -string. Only at 2 4 6 8 10 12 14 16 { } r 10 the string breaks completely, at about the same ≈ FIG. 1: Top: Potential V(r) of two static color charges with distanceastheadjoint 3 -string. Notunexpectedly,the Q = 1 (squares) and Q = 3 (stars), shifted by a constant { } 2 2 tensionofastringisthesame,nomatterwhetheritcon- foramoreconvenientcomparisonoftheslopes. Bottom: The nects screened or unscreened external charges Q. same for Q=1 (squares) and Q=2 (stars). The lines are a A fit of the fundamental potential to fit of the multi-channel model to the Monte Carlo data. The π horizontal band at 2M0,2 = 4.84(2) corresponds to twice the V (r)=σr +2M + (1/r3), (2) mass of a source of charge Q=2 obtained from the measure- 1/2 − 24r O ment of a single Polyakov loop. works very well and yields the asymptotic string tension σ = 0.06397(3). In particular, the Monte Carlo data Here c is the coefficient of a sub-leading 1/r correc- are in excellent agreement with the predicted coefficient Q π of the Lu¨scher term. The “mass” contribution of tion which does not necessarily assume the asymptotic −an24external charge Q = 12 to the total energy of the Lcou¨nstcrhiberutvioanlueof−an2π4o.rigTinhael c“hmaarsgse”QM+Q,nn tdheastcrhibasesbethene system is given by M = 0.109(1). This “mass” itself screened to the value Q by n gluons. Just as the is not physical because it contains ultra-violet divergent “mass” M = M , the “masses” M themselves pieces. Since string decay occurs at moderate distances, 1/2,0 Q,n are not physical, because they again contain ultra-violet its typical energy scale is not wellseparatedfrom Λ . QCD divergent contributions. However, the mass differences Consequently, unlike the string behavior at asymptotic distances, string decay can not be addressed in a fully ∆Q,n = MQ−1,n+1−MQ,n are physical since the diver- gent pieces then cancel. The 3 - and 4 -string are de- systematiclow-energyeffective stringtheory. Inparticu- { } { } scribed by the two-channel Hamiltonians H and H , lar,unlikethe stringtensionσ ofthe unbreakablefunda- 1 3/2 while the 5 -string is described by the three-channel mental string, the tension σQ of an ultimately unstable { } Hamiltonian H with 2Q+1 -string (with Q 1) is not defined unambigu- 2 { } ≥ ously. Herewedefineσ byafitoftheMonteCarlodata Q E (r) A tcoonasisdiemrptlheeph2eQno+m1en-osltorginicgalams aodmelu.ltIin-chthainsnmelosdyeslt,ewme. H1(r)=(cid:18) 1,A0 E0,1(r) (cid:19), A channel co{ntaining}a 2Q+1 -string connecting two H (r)= E3/2,0(r) B , chargesQ, whichresulted{ fromsc}reening a largercharge 3/2 (cid:18) B E1/2,1(r) (cid:19) Q+n by n gluons, has the energy E (r) C 0 2,0 cQ H2(r)= C E1,1(r) A . (4) E (r)=σ r +2M . (3) Q,n Q − r Q,n 0 A E0,2(r) 3 Q σQ σQ/σ 4Q(Q+1)/3 0.4 fit 1/2 0.06397(3) 1 1 0.35 Q = 1/2 Q = 3/2 1 0.144(1) 2.25(2) 8/3 0.3 3/2 0.241(5) 3.77(8) 5 2 0.385(5) 6.02(8) 8 0.25 0.2 TABLE I: Fitted values of the string tensions σ . The ratio 0.15 Q σ /σ with σ =σ is compared with the value 4Q(Q+1)/3 Q 1/2 0.1 representing Casimir scaling. 0.05 Q MQ,0 MQ−1,1 MQ−2,2 ∆Q,0 ∆Q−1,1 0 2 4 6 8 10 12 14 1/2 0.109(1) — — — — 0.7 fit 1 0.37(3) 1.038(1) — 0.67(3) — 0.6 Q = 1 3/2 0.72(5) 1.32(5) — 0.60(5) — Q = 2 0.5 2 1.04(3) 1.71(3) 2.42(1) 0.67(3) 0.71(3) 0.4 TABLE II: Fitted values of the “mass” M of an original 0.3 Q,n chargeQ+nthathasbeenscreenedtothevalueQbyngluons, 0.2 together with the differences ∆Q,n =MQ−1,n+1−MQ,n. 0.1 0 Here A, B, and C are decay amplitudes which we as- -0.1 sume to be r-independent. The potential V (r) is the Q 2 4 6 8 10 12 14 energy of the ground state of H . Figure 2 compares Q FIG. 2: Top: Forces F(r) that the {2}- and {4}-string ex- the forces F(r) = dV(r)/dr in the 2 -, 3 -, 4 -, − { } { } { } ert on the external charges Q = 1 (squares) and Q = 3 and 5 -stringcaseswiththeresultsofthemulti-channel 2 2 { } (stars), respectively. Bottom: Thesameforthe{3}-and{5}- model. The tensions σ listed in table 1 have been de- Q stringconnectingexternal chargesQ=1(squares) andQ=2 termined by a fit to the Monte Carlo data. The sim- (stars), respectively. The lines represent the fit of the multi- ple model works rather well. It is interesting to note channel model to the Monte Carlo data. that the ratios σ /σ do not obey Casimir scaling, i.e. Q they are not equal to 4Q(Q+1)/3. The “masses” M Q,n are listed in table 2. Remarkably, within the error bars, stretched further and further, individual strands eventu- the mass differences ∆Q,0 =MQ−1,1 MQ,0 all take the allyrupture,therebyabruptlyreducingthetensionofthe − same value M = 0.65(5), independent of Q. We in- cable. While strand rupture is well-known in the mate- G terpret M as a constituent gluon mass which in units rialscienceofcentimeterthicksteelcableswithatension G of the string tension takes the value M /√σ = 2.6(2). of about 105 Newton, we have seen that a similar pro- G It should be pointed out that, in contrast to the string cessoccursfortheconfiningstringsinnon-Abeliangauge tension, M is not unambiguously defined. It just re- theories which have about the same tension but are 13 G sults from the fit parameters of the phenomenological orders of magnitude thinner. Whether a strand picture model. The value ∆ = M M = 0.71(3) in- may correctly describe the actual anatomy of decaying 1,1 0,2 1,1 − dicates that the addition of a second constituent gluon 2Q+1 -strings is an interesting question that will re- { } costs an energy slightly larger than M . Interestingly, quirefurtherinvestigationswhichgobeyondthescopeof G the mass of two constituent gluons 2M = 1.3(1) is the present letter. G close to the 0+ glueball mass M0+ = 1.198(25) ob- It wouldbe interesting to investigate string decay and tained in [25] at the same value of the bare coupling. string breaking for other SU(N) gauge theories. In MG also sets the distance scale for string decay and SU(3) Yang-Mills theory the 3 -string connecting a { } string breaking. A leading order estimate for the dis- quark in the 3 with an anti-quark in the 3 repre- { } { } tance at which the 4 -string decays into the 2 -string sentationis unbreakable,whilethe 8 -stringconnecting { } { } { } is r 2MG/(σ3/2 σ1/2) = 7.3(6), while the distance two adjointsourcescan breakby pair creationof gluons. ≈ − at which the 3 - and the 5 -string ultimately break is Whena 6 -stringisstretched,theexternalsourceinthe { } { } { } estimated to be around r 2MG/σ1 =9.0(7). 6 -representationwilleventuallybescreenedtoa 3 by ≈ { } { } String decay can be viewed as a quantum analogue of a gluon. The correspondingstring decay should be anal- the classical process of strand rupture in a cable con- ogoustothe decayofthe 4 -stringinSU(2)Yang-Mills { } sisting of a bundle of strands. When such a cable is theory discussed above. In analogy to the 5 -string in { } 4 SU(2),the 10 -stringinSU(3)Yang-Millstheoryisex- Using the Lu¨scher-Weisz multi-level algorithm, study- { } pectedtodecaytoanadjoint 8 -string,beforeitbreaks ingstringdecayinSU(3),SU(4),Sp(2),G(2),andother { } completely at larger distances. In QCD with dynami- Yang-Mills theories is interesting and definitely feasible. cal quarks, strings can also decay by quark-anti-quark One may also ask whether string decay can be studied pair creation. Due to its Z(4) center symmetry, SU(4) analytically in supersymmetric theories. Yang-Mills theory has two distinct unbreakable strings, connecting external charges either in the 4 and 4 M.P. acknowledges useful discussions with F. Gliozzi { } { } or in the 6 -representation. For external sources with and J. Greensite. This work is supported in part by { } non-trivial N-ality, one then expects cascades of string funds provided by the Schweizerischer Nationalfonds decays down to the 4 -string for k = 1,3 and down to (SNF). The “Centerfor ResearchandEducationin Fun- { } the 6 -string for k =2. damental Physics” at Bern University is supported by { } Studying gauge groups other than SU(N) would also the “Innovations-und KooperationsprojektC-13”of the be interesting. For example, all Sp(N) gauge theories Schweizerische Universit¨atskonferenz (SUK/CRUS). have the same center Z(2). The first Lie group in this sequence is Sp(1) = SU(2) = Spin(3), while the sec- ond is Sp(2) = Spin(5), the universal covering group of SO(5). In Sp(2) Yang-Mills theory, only the fundamen- tal 4 -string is absolutely stable. As usual, the adjoint [1] M. Lu¨scher, K. Symanzik, and P. Weisz, Nucl. Phys. { } B173 (1980) 365. 10 -string can break by pair creation of gluons. The { } [2] M. Lu¨scher, Nucl. Phys. B180 (1981) 317. representation 5 is center-neutral. Since in Sp(2) { } [3] M.Lu¨scher,G.Mu¨nster,andP.Weisz,Nucl.Phys.B180 (1981) 1. 5 10 = 5 10 35 , (5) [4] M. Lu¨scher and P. Weisz, JHEP 0407 (2004) 014. { }⊗{ } { }⊕{ }⊕{ } [5] G. Bali and K. Schilling, Phys. Rev.D46 (1992) 2636. a single gluon can screen a charge 5 only to a 10 [6] M. Lu¨scher and P. Weisz, JHEP 0109 (2004) 010. { } { } or a 35 . We expect the unstable 5 -string to have a [7] M. Lu¨scher and P. Weisz, JHEP 0207 (2004) 049. { } { } smaller tensionthanthe adjoint 10 -stringor the 35 - [8] W. Bock et al., Z. Phys.C45 (1990) 597. { } { } string. In thatcase,the 5 -stringwill breakinone step [9] F.KnechtliandR.Sommer,Phys.Lett.B440(1998)345; { } Nucl. Phys. B590 (2000) 309. by the creation of four gluons, without any intermediate [10] O. Philipsen and H. Wittig, Phys. Rev. Lett. 81 (1999) string decay. 2684. Finally, it would be interesting to investigate the im- [11] F. Gliozzi and A. Rago, Nucl. Phys. B714 (2005) 91. portance of the center for the phenomenon of string de- [12] G. I. Poulis and H.D. Trottier, Phys. Lett. B400 (1997) cay. TheexceptionalgroupG(2)isthesimplestLiegroup 358. with a trivial center. Still, G(2) Yang-Mills theory con- [13] P. W. Stephenson,Nucl. Phys.B550 (1999) 427. fines color (although without an asymptotic string ten- [14] O.PhilipsenandH.Wittig,Phys.Lett.B451(1999)146. [15] S. Kratochvila and P. de Forcrand, Nucl. Phys. B671 sion)[26]. Furthermore,ithasafirstorderdeconfinement (2003) 103. phase transition [27, 28]. In fact, as we have discussed [16] P.deForcrandandO.Philipsen,Phys.Lett.B475(2000) inthe contextofSp(N)Yang-Millstheories,the orderof 280. the deconfinement phase transition is controlled by the [17] K. Kallio and H. D. Trottier, Phys. Rev. D66 (2002) size of the gauge group and not by the center [29]. In 034503. G(2) Yang-Mills theory, even a charge in the fundamen- [18] J. Greensite et al., Phys. Rev.D75 (2007) 034501. tal 7 representation can be screened by gluons in the [19] G. Bali et al., Phys. Rev.D71 (2005) 114513. { } [20] P. Olesen, Nucl.Phys. B200 (1982) 381. adjoint 14 representation. As a result, there are no { } [21] J.Ambjorn,P.OlesenandC.Peterson,Nucl.Phys.B240 unbreakable strings. Since in G(2) (1984) 189. [22] A.ArmoniandM.Shifman,Nucl.Phys.B671(2003)67. 7 14 = 7 27 64 , (6) { }⊗{ } { }⊕{ }⊕{ } [23] F. Gliozzi, JHEP 0508 (2005) 063. [24] L. Del Debbio, H. Panagopoulos and E. Vicari, JHEP a single gluon can screen a charge 7 only to a 27 or 0309 (2003) 034. { } { } a 64 . In G(2) Yang-Mills theory,approximateCasimir [25] H. Meyer and M. Teper, Nucl. Phys. B668 (2003) 111. { } scaling has been verified for unstable strings including [26] K. Holland, P. Minkowski, M. Pepe, and U.-J. Wiese, the 27 - and the 64 -string [30]. As a consequence, Nucl. Phys. B668 (2003) 207. the f{und}amental 7{-st}ring is stable against decay due [27] M. Pepe, PoS LAT2005 (2006) 017, Nucl. Phys. Proc. { } Suppl.153 (2006) 207. to the creation of a single pair of gluons. The same is [28] M. Pepe and U.-J. Wiese, Nucl.Phys. B768 (2007) 21. true even for processes involving four gluons. Based on [29] K.Holland,M.Pepe,andU.-J.Wiese,Nucl.Phys.B694 the group theory of G(2), we expect the fundamental (2004) 35. 7 -string to break due to the simultaneous creation of [30] L.LiptakandS.Olejnik,Phys.Rev.D78(2008)074501. { } six gluons, without any intermediate string decay.