Concerning the quark condensate K. Langfeld,1 H. Markum,2 R. Pullirsch,3 C.D. Roberts,4,5 and S.M. Schmidt1,6 1Institut fu¨r Theoretische Physik, Universit¨at Tu¨bingen, Auf der Morgenstelle 14, D-72076 Tu¨bingen, Germany 2Atominstitut, Technische Universit¨at Wien, A-1040 Vienna, Austria 3Institut fu¨r Theoretische Physik, Universit¨at Regensburg, D-93040 Regensburg, Germany 4Physics Division, Bldg 203, Argonne National Laboratory, Argonne Illinois 60439-4843 5Fachbereich Physik, Universit¨at Rostock, D-18051 Rostock, Germany 6Helmholtz-Gemeinschaft, Ahrstrasse 45, D-53175 Bonn, Germany A continuum expression for the trace of the massive dressed-quark propagator is used to ex- plicate a connection between the infrared limit of the QCD Dirac operator’s spectrum and the quark condensate appearing in the operator product expansion, and the connection is verified via comparison with a lattice-QCD simulation. The pseudoscalar vacuum polarisation provides agood 3 approximation to thecondensate overa larger range of current-quarkmasses. 0 0 PACSnumbers: 12.38.Aw,11.30.Qc,11.30.Rd,24.85.+p 2 n I. INTRODUCTION they occur in pairs: {u (x),γ u (x)}, with eigenvalues a n 5 n of opposite sign. It follows that in an external gauge J 8 Dynamicalchiralsymmetrybreaking(DCSB)isacor- field, A, one can write the Green function for a massive nerstone of hadron physics. This phenomenon whereby, propagating quark in the form 1 even in the absence of a current-quark mass, self- u (x)u†(y) v interactions generate a momentum-dependent running S(x,y;A)=hq(x)q¯(y)i = n n , (3) 4 quarkmass,M(p2),thatislargeintheinfrared: M(0)∼ A iλn+m 2 Xn 0.5GeV, but power-law suppressed in the ultraviolet [1]: 0 where m is the current-quark mass and, naturally, the 1 eigenvaluesdependonA. (NB.Theexpectationvaluede- 30 M(p2)larg=e−p2 2π2γm −hq¯qi0 , (1) notes aGrassmannianfunctionalintegralevaluatedwith 0 3 p2 1l(cid:0)n p2 (cid:1) 1−γm a fixed gauge field configuration.) Assuming, e.g., a lat- / 2 Λ2QCD tice regularisation,it follows that h (cid:16) h i(cid:17) t is impossible in weakly interacting theories. In Eq. (1), 1 2m 1 ucl- γwmith=N1f2/th(3e3n−um2Nbefr)oifsltighhet-mquasasrkanfloamvoaulorsu,sadnidmheq¯nqsii0onis, V ZV d4xhq¯(x)q(x)iA =− V λXn>0 λ2n+m2 , (4) the renormalisation-group-invariant vacuum quark con- n where V is the lattice volume [8]. One may now define densate [2], to which we shallhereafter refer as the OPE : v condensate. WhileEq.(1)isexpressedinLandaugauge, a quark condensate as the infinite volume limit of the Xi hq¯qi0 is gauge parameter independent. In the chiral average in Eq. (4) over all gauge field configurations: limit the OPE condensate plays a role analogous to that r 1 a played by the renormalisation-group-invariant current- h0|q¯q|0i:= lim d4x hhq¯(x)q(x)iAi . (5) quark mass in the massive theory: it sets the scale of V→∞V ZV the mass function in the ultraviolet. In the infinite volume limit the operator spectrum be- The evolution of the dressed-quark mass-function in comes dense and Eqs. (4), (5) become Eq.(1)toalargeandfiniteconstituent-quark-likemassin the infrared, M(0) ∼ 0.5GeV, is a longstanding predic- ∞ ρ(λ) −h0|q¯q|0i=2m dλ , (6) tion of Dyson-Schwingerequation(DSE) studies [3] that λ2+m2 Z0 has recently been confirmed in simulations of quenched with ρ(λ) the spectral density. This equation expresses lattice-QCD[4]. AdeterminationoftheOPEcondensate an assumption that in QCD the full two-point massive- directly fromlattice-QCD simulations must awaitan ac- quark Schwinger function, when viewed as a function of curate chiral extrapolation [5] but DSE models tuned to reproducemodernlatticedatagive[6](−hq¯qi0)∼Λ3 . the current-quark mass, has a spectral representation. QCD It follows formally from Eq. (6) that Another view of DCSB is obtained by considering the eigenvaluesandeigenfunctions ofthe masslessEuclidean ∞ ρ(λ) 1 Dirac operator [7]: lim m dλ = πρ(0), (7) m→0 Z0 λ2+m2 2 γ·Du (x)=iλ u (x). (2) n n n and hence one arrives at the chiral limit result The operatoris anti-Hermitianand hence the eigenfunc- tions form a complete set, and except for zero modes −h0|q¯q|0i=πρ(0). (8) 2 This is the so-called Banks-Casher relation [10, 11]. It and is obtained subject to the condition that at some has long been advocated as a means by which a quark large, spacelike ζ2 condensatemaybe measuredinlattice-QCDsimulations S (p)−1 =iγ·p+m (ζ), (11) [11] and has been used in analysing chiral symmetry f p2=ζ2 f restorationatnonzerotemperature[12]andchemicalpo- where m (ζ) is the r(cid:12)enormalised current-quark mass: tential [13], and to explore the connection between mag- f (cid:12) netic monopoles and chiral symmetry breaking in U(1) Z (ζ,Λ)m (ζ)=Z (ζ,Λ)mbm(Λ), (12) 4 f 2 f gauge theory [14]. Much has been learnt [15, 16] by ex- ploitingthefactthatqualitativefeaturesofthebehaviour with Z4 the renormalisation constant for the scalar part of ρ(λ) for λ∼0 canbe understood using chiralrandom ofthequarkself-energy. SinceQCDisanasymptotically matrix theory; i.e., from considerations based solely on free theory, the chiral limit is defined by QCD’s global symmetries. Z (ζ2,Λ2)mbm(Λ)≡0, Λ≫ζ (13) Our main goal is to explicate a correspondence be- 2 f tween the condensate in Eq. (1) and that in Eq. (8). andin this case the scalarprojectionofEq. (9) does not In Sec. II we discuss the OPE condensate and its con- exhibit an ultraviolet divergence [2, 17]. nection with QCD’s gap equation, and emphasise that Important in describing chiral symmetry is the axial- the residue of the lowest-mass pole-contribution to the vector Ward-Takahashi identity: flavour-nonsingletpseudoscalar vacuum polarisation is a TH TH direct measure of the OPE condensate [17]. A natu- P ΓH(k;P) = S−1(k )iγ +iγ S−1(k ) ral ability to express DCSB through the formation of µ 5µ + 5 2 5 2 − a nonzero OPE condensate is fundamental to the suc- −M(ζ)iΓH5 (k;P)−iΓH5 (k;P)M(ζ), cess of DSE models of hadron phenomena [18]. In Sec. (14) III we carefully define the trace of the massive dressed- quark propagatorand use that to illustrate a connection where P is the total momentum entering the vertex. between ρ(0) and the OPE condensate, which we verify In Eq. (14): M(ζ) = diag[mu(ζ),md(ζ),ms(ζ)], S = via comparison with a lattice simulation. Section IV is diag[Su,Sd,Ss] and {TH} are flavour matrices, e.g., an epilogue. Tπ+ = 1 λ1+iλ2 (we consider SU (3) because chiral 2 f symmetry is unimportant for heavier quarks); ΓH(k;P) (cid:0) (cid:1) 5µ istherenormalisedaxial-vectorvertex,whichisobtained II. OPE CONDENSATE from the inhomogeneous Bethe-Salpeter equation TH A. Gap and Bethe-Salpeter equations ΓH(k;P) =Z γ γ 5µ tu 2 5 µ 2 (cid:20) (cid:21)tu (cid:2) Λ (cid:3) DynamicalchiralsymmetrybreakinginQCDisreadily + [χH(q;P)] Krs(q,k;P), (15) explored using the DSE for the quark self-energy: 5µ sr tu Zq S(p)−1 =Z2(iγ·p+mbm) where χH5µ := S(q+)ΓH5µ(q;P)S(q−), q± = q±P/2, and K(q,k;P)isthefullyrenormalisedquark-antiquarkscat- Λ λa +Z g2D (p−q) γ S(q)Γa(q,p), (9) tering kernel; and ΓH is the pseudoscalar vertex, 1 µν 2 µ ν 5 Zq TH ΓH(k;P) =Z γ awghaetroeri;n:ΓaνD(µqν;p(k))isistthheerreennoorrmmaalliisseeddddrreesssseedd--qguluaorkn-gplruoopn- (cid:2) 5 Λ (cid:3)tu 4 (cid:20) 5 2 (cid:21)tu vertex;mbm istheΛ-dependentcurrent-quarkbaremass + χH(q;P) Krs(q,k;P), (16) thatappearsinthe Lagrangian;and Λ := Λd4q/(2π)4 Zq 5 sr tu q (cid:2) (cid:3) represents a translationally-invariant regularisation of with χH := S(q )ΓH(q;P)S(q ). Multiplicative renor- R R 5 + 5 − the integral, with Λ the regularisation mass-scale which malisability ensures that no new renormalisation con- is removed to infinity as the completion of all calcula- stants appear in Eqs. (15) and (16). tions. The quark-gluon-vertex and quark wave function Flavour-octetpseudoscalarbound statesappearasco- renormalisation constants, Z1(ζ2,Λ2) and Z2(ζ2,Λ2) re- incident pole solutions of Eqs. (15), (16), namely, spectively,depend onthe renormalisationpoint, the reg- 1 ularisation mass-scale and the gauge parameter. ΓH(k;P)∝ΓH(k;P)∝ Γ (k;P), (17) If the current-quark mass changes with flavour, then 5µ 5 P2+m2H H the solution of Eq. (9) is flavour dependent: where Γ is the bound state’s Bethe-Salpeter amplitude H and m , its mass. (Regular terms are overwhelmed at S (p)−1 = iγ·pA (p2,ζ2)+B (p2,ζ2) H f f f the pole.) Consequently, Eq. (14) entails [2, 17] 1 = Zf(p2,ζ2) iγ·p+Mf(p2,ζ2) , (10) fHm2H =rH(ζ)M(Hζ), (18) (cid:2) (cid:3) 3 where: M(ζ) = tr [M {TH,(TH)t] is the sum of calculated in this way: H flavour (ζ) the constituents’ current-quark masses (“t” denotes ma- trix transpose); and α p2 B (p2)=m 1− ln +... . (24) f f π m2 f P = Z Λ 1tr TH tγ γ χ (q;P) , (19) " f# ! H µ 2 5 µ H irH(ζ) = Z4ZqΛ 122trh(cid:0)TH(cid:1)tγ5 χH(q;P) i, (20) Eqduevnaesrrakytemteiarsmssimainpnodtshsheiebnsleecreiineaspniesorntpuzrreobrpoaotvriotainlounteaholefotrotyh.tehOe PcuErrceonnt-- Zq h(cid:0) (cid:1) i where χ := S(q )Γ (q;P)S(q ) and the expressions H + H − are evaluated at P2+m2 =0. B. Pseudoscalar Vacuum Polarisation H Equation (19) is the pseudovector projection of the meson’s Bethe-Salpeter wave function evaluated at the ConsiderthecoloursingletSchwingerfunctiondescrib- origin in configuration space. It is the precise expres- ing the pseudoscalar vacuum polarisation sionfortheleptonicdecayconstant. Therenormalisation 1 1 constant, Z2(ζ,Λ), ensures that the r.h.s. is independent ∆5(x)=hq¯(x) λfγ5q(x)q¯(0) λgγ5q(0)i, (25) of: the regularisation scale, Λ, which may therefore be 2 2 removed to infinity; the renormalisation point; and the which can be estimated, e.g., in lattice simulations. Its gauge parameter. Hence it is truly an observable. renormalised form can completely be expressed in mo- Equation (20) is the pseudoscalar analogue. Therein mentum space using quantities introduced already: the renormalisation constant Z (ζ,Λ) entails that the 4 r.h.s. is independent of the regularisation scale, Λ, and Λ1 the gauge parameter. It also ensures that the ζ- ω5fg(P)=Z22tr 2λfγ5S(q+)Γg5(q;P)S(q−). (26) dependenceofr(ζ) ispreciselythatrequiredtoguarantee Zq H the r.h.s. of Eq. (18) is independent of the renormalisa- Equation (16) can be rewritten in terms of the fully- tion point. (NB. r(ζ) is finite, and Eq. (18) valid, for amputated quark-antiquark scattering amplitude: M = H arbitrary values of the current-quark masses [19, 20].) K + K(SS)K + ..., and in the neighbourhood of the In the chiral limit the existence of a solution of Eq. lowest mass pole (9) with B (p2) 6= 0; i.e., DCSB, necessarily entails [17] 0 1 that Eqs. (15), (16) exhibit a massless pole solution: the M(q,k;P)=Γ (q;P) Γ¯(k;−P)+R(q,k;P), Goldstone mode, which is described by H P2+m2H (27) Γg(k;P) = λgγ iE (k;P)+γ·PF (k;P) (21) where R is regular in this neighbourhood. 0 5 0 0 Assuming SU (3) flavour symmetry, substituting Eq. f (cid:20) (27) into Eq. (26) gives + γ·kk·PG (k;P)+σ k P H (k;P) , 0 µν µ ν 0 1 (cid:21) ωfg(P)=δfg Z−2r2 +... (28) wherein fπ0E0(k;0) = B0(k2). (The index “0” indicates 5 P2+m2H m H a quantity calculated in the chiral limit.) It follows im- (the ellipsis denotes terms regular in the pole’s neigh- mediately [17] that bourhood). It follows that the large-x2 behaviour of Λ f0r(ζ) =−hq¯qi0 = lim Z (ζ,Λ)N tr S (q), (22) m2 ∆ (x) (29) π 0 ζ 4 c D 0 bm 5 Λ→∞ Zq is a measure of the renormalisation-group-invariant where the trace is only over Dirac indices. This result and multiplicative renormalisability entail 2 (ζ) m(ζ)r (30) H hq¯qi0 ζ =Z (ζ,ζ′)Z−1(ζ,ζ′)=Z (ζ,ζ′), (23) h i hq¯qi0 4 2 m that appears in Eq. (18). Hence the correlator in Eq. ζ′ (25) providesadirectmeans ofestimating the OPEcon- whereZmisthemass-renormalisationconstant. Itisthus densate in lattice simulations [21], one whose ultraviolet apparentthatthechirallimitbehaviourofr(ζ) yieldsthe behaviour ensures a well-defined and calculable evolu- H OPE condensate evolved to a renormalisationpoint ζ. tionundertherenormalisationgroupforanyvalueofthe It is important to recall that the DSEs reproduce ev- current-quark mass. (NB. f can similarly be extracted π ery diagram in perturbation theory. Therefore a weak from the axial-vector correlator analogous to Eq. (25).) coupling expansion of Eq. (9) yields the perturbative se- ThemodelofRef.[22]yieldsamesonmasstrajectoryvia riesforthe dressed-quarkpropagator. Thismaybe illus- Eq.(18) thatprovidesaqualitativeandquantitativeun- tratedbytheresultforthescalarpieceofthepropagator derstanding of recent quenched lattice simulations [20]. 4 III. BANKS-CASHER RELATION evaluated at a fixed value of the regularisationscale: σ(m;ζ) := − limhq¯(x)q(0)i A. Continuum analysis x→0 Λ = Z (ζ,Λ)N tr S (p;ζ), (37) It is readily apparent that Eq. (6) is meaningless as 4 c D m Zp written: dimensionalcountingrevealsther.h.s.hasmass- dimensionthreeandsinceλwillatsomepointbegreater where the argument remains m = mbm(Λ), which thananyrelevantinternalscale,theintegralmustdiverge is permitted because mbm(Λ) is proportional to the as Λ2, where Λ is the regularising mass-scale. renormalisation-point-independent current-quark mass. The renormalisation constant Z vanishes logarithmi- Tolearnmore,considerthetraceoftheunrenormalised 4 cally with increasing Λ and hence one still has σ(m) ∼ massive dressed-quark propagator: Λ2mbm(Λ). However, using Eq. (22) it is clear that for Λ anyfinitebutlargevalueofΛandtoleranceδ,itisalways σ˜(m):=NctrD S˜m(p), (31) possible to find mbm(Λ) such that δ Zp σ(m;ζ)+hq¯qi0 <δ, ∀mbm <mbm. (38) ζ δ evaluated at a fixed value of the regularisation scale, Λ. This Schwinger function can be identified with the l.h.s. This is true in QCD. It can be illustrated using the ofEq.(6). Furthermore,assumethatσ˜(m)hasaspectral DSE model of Ref. [2], which preserves the one-loop representation, since this is the essence of the Banks- renormalisation group properties of QCD. In Fig. 1 we Casher relation: plot σ(m,ζ), evaluated using a hard cutoff, Λ, on the integralinEq.(37),calculatedwiththe massivedressed- Λ ρ˜(λ) quark propagatorsobtained by solving the gap equation σ˜(m):=2m dλ , (32) λ2+m2 asdescribedintheappendix. SinceEq.(38)specifiesthe Z0 domain on which the value of σ(m;ζ) is determined by where m=mbm(Λ). Equation (32) entails nonperturbative effects, one anticipates mbm(Λ)≈−hq¯qi0/Λ2 ∼10−9 (39) 1 δ ρ˜(λ)= lim [σ˜(iλ+η)−σ˜(iλ−η)] . (33) 2π η→0+ for Λ = 2.0TeV in QCD where |hq¯qi0| ∼ Λ3 , an esti- QCD mate confirmed in Fig. 1. Thecontentandmeaningofthissequenceofequations The dotted line in Fig. 1 is is well illustrated by inserting the free quark propagator 2 Λ in Eq. (31). The integral thus obtained is readily evalu- σ(m,ζ)=−hq¯qi0 arctan ated using dimensional regularisation: ζ π m N + Z (ζ,Λ) c m Λ2−m2ln 1+Λ2/m2 .(40) N m2 1 4 4π2 σ˜ (m)= c m3 ln + +γ−ln4π . (34) free 4π2 ζ2 ε (NB. We used the on(cid:2)e-loop form(cid:0)ula: Z (ζ(cid:1),(cid:3)Λ) = (cid:20) (cid:21) 4 [α(Λ)/α(ζ)]γm, for the numerical comparison.) The With Eq.(33)the regularisationdependent terms cancel difference between Eq. (40) and the curve is of and one obtains O(α(Λ)m(Λ)Λ2) because the DSE model incorporates QCD’sone-loopbehaviour. InFig.1wealsoplotσ(m,ζ) N ρ˜(λ)= c λ3. (35) obtainedintheabsenceofconfinement,inwhichcase[24] 4π2 hq¯qi0 ≡0, as is apparent. The one-loop contribution to ρ˜(λ) has been evaluated The discussion establishes that σ(m,ζ) has a regular using the same procedure [23]. It is also proportional chiral limit in QCD and is a monotonically increasing to λ3 and arises from the m3lnm2/ζ2 terms in σ˜(m). convex-upfunction. Itfollowsthatσ(m,ζ)hasaspectral In fact, every term obtainable in perturbation theory is representation: proportional to λ3, for precisely the same reason that Λ ρ(λ) each term in the perturbative expression for the scalar σ(m,ζ)=2m . (41) λ2+m2 part of the quark propagator is proportional to m, see Z0 Eq. (24). Hence, at every order in perturbation theory, This lays the vital plank in a veracious connection be- tween the condensates in Eqs. (1) and (8). On the do- ρ˜(λ=0)=0 (36) main specified by Eq. (39), the behaviour of σ(m,ζ) in Eq.(37)is givenby Eq.(40), whichyields, viaEq.(33), and h0|q¯q|0i = 0. A nonzero value of ρ(0) is plainly an N essentially nonperturbative effect. πρ(λ)=−hq¯qi0+Z (ζ,Λ) c λ3+... (42) A precise analysis requires that attention be paid to ζ 4 4π renormalisation. Consider then the gauge-parameter- where the ellipsis denotes contributions from the higher- independent trace of the renormalised quark propagator order terms implicit in Eq. (40). 5 100 0.5 β=5.4 80 0.4 V) Ge 0.3 60 1/3m) ( λ( ) σ( 0.2 ρ 40 β=5.6 β=5.8 0.1 20 0.0 10−10 10−9 10−8 10−7 10−6 m (Λ) (GeV) bm 0 0 0.1 0.2 FIG. 1: Circles/solid-line: σ(m)1/3 in Eq. (37) as afunction λ of the current-quark bare-mass, evaluated using the dressed- quark propagator obtained in the model of Ref. [2]; dashed FIG. 2: Spectral density of the staggered Dirac operator line: themodel’s value of (−hq¯qi0ζ=1GeV)=(0.24GeV)3; and in quenched SU(3) gauge theory calculated on a 44-lattice. dottedline: Eq.(40). Diamonds: σ(m)1/3evaluatedinanon- (Measuredinunitsofthelatticespacing.) Thedeconfinement chqo¯qnifi0n≡ing0.versionofthemodel;dot-dashedline: Eq.(40)with transition takes place at β >∼5.6. justenteredthedeconfineddomainandclosetothetran- B. Comparison with a lattice-QCD simulation sitionboundarynonperturbativeeffectsarestillmaterial, as seen, e.g., in the heavy-quark potential and equation In Fig. 2 we plot the spectral density of the staggered ofstate[25]. It isa modernchallengeto determine those Dirac operator in quenched SU(3) gauge theory calcu- gaugecouplingsandlatticeparametersforwhichthedata lated with 3000 configurations obtained on a V = 44- are quantitatively consistent with Eq. (42). lattice,inthevicinityofthedeconfiningphasetransition at β >∼ 5.6. Details of the simulation are given in Ref. [12]. Dimensioned quantities are measured in units of IV. EPILOGUE 1/a,where a is the lattice spacing,and it is ρ(λ)/V that shouldbecomparedwiththecontinuumspectraldensity. We verified that the gauge-invarianttrace of the mas- While the effect of finite lattice volume is apparent in sivedressed-quarkpropagatorpossessesaspectralrepre- Fig. 2 for λa>∼0.1, the behaviour at smallλa is qualita- sentation when considered as a function of the current- tively in agreement with Eq. (42) and Fig. 1: a nonzero quark mass. This is key to establishing that the OPE OPE condensate dominates the Dirac spectrum in the condensate, which sets the ultraviolet scale for the confined domain; and it vanishes in the deconfined do- momentum-dependenceofthe traceofthe dressed-quark main whereupon ρ(0) = 0 and the perturbative evolu- propagator, does indeed measure the density of far- tion, Eq. (35), is manifest. infraredeigenvaluesofthegauge-averagedmasslessDirac To be more quantitative, we note that at β = 5.4, operator, a` la the Banks-Casher relation. This relation ρ(0)a3 ≈70, so that is intuitively appealing because a measurable accumu- lation of eigenvalues of the massless Dirac operator at πρ(0)a3/V ≈(0.95)3. (43) zero-virtuality expresses a mass gap in its spectrum. In practice, there are three main parameters in a sim- ThevalueofthelatticespacingwasnotmeasuredinRef. ulation of lattice-QCD: the lattice volume, characterised [12] but one can nevertheless assess the scale of Eq. (43) by a length L; the lattice spacing, a; and the current- by supposing a ∼ 0.3fm ∼ 0.3/Λ , a value typical of QCD quark mass, m. So long as the lattice size is large small couplings, β, wherewith the r.h.s. is ∼ (3Λ )3. QCD comparedwiththecurrent-quark’sComptonwavelength; Thisistoolargebutnotunreasonablegiventheparame- viz.,L≫1/m,thendynamicalchiralsymmetrybreaking tersofthesimulation,itserrorsandthesystematicuncer- canbeexpressedinthesimulation. Supposingthattobe tainties in our estimate. One can alsofit the lattice data the case then, as we have explicated, so long as the lat- at β = 5.8, whereby one finds ρ(λ) ∝ λ3 on λ < 0.1 but tice spacing is small compared with the current-quark’s with a proportionality constant larger than that antici- Compton wavelength; i.e., a≪1/m≪L, patedfromperturbationtheory;viz.Eq.(42). Somemis- match is to be expected because at β =5.8 one has only πρ(0)≈−hq¯qi0 , (44) 1/a 6 where the r.h.s. is the scale-dependent OPE condensate The ultraviolet (Q2 >∼ 1GeV2) behaviour of G(Q2) in [ζ =Λ=1/a in Eq. (42)]. Eq. (A1) is fixed by the known behaviour of the quark- In our continuum analysis we found that one requires antiquark scattering kernel [2]. The form of that kernel am<∼(aΛQCD)3 ifρ(λ=0)istoprovideaveraciousesti- ontheinfrareddomainiscurrentlyunknownandamodel mate of the OPE condensate. The residue at the lowest- is employed to complete the specification of the kernel. masspole in the flavour-nonsingletpseudoscalarvacuum An efficacious form is [2] polarisation provides a measure of the OPE condensate that is accurate for larger current-quark masses. G(Q2) =8π4Dδ4(k)+ 4π2DQ2e−Q2/ω2 Q2 ω6 γ π +4π m F(Q2), (A2) Acknowledgments 2 1ln τ + 1+Q2/Λ2 2 QCD We benefited from interactions with M.A. Pichowsky (cid:20) (cid:16) (cid:17) (cid:21) and P.C. Tandy. This work was supported by: Deutsche where: F(Q2) = [1 − exp(−Q2/[4m2])]/Q2, m = t t Forschungsgemeinschaft, under contract no. Ro 1146/3- 0.5GeV; τ = e2−1; γ = 12/(33−2N ), N = 4; and m f f 1; the Department of Energy, Nuclear Physics Division, Λ = Λ(4) = 0.234GeV. The true parameters in Eq. QCD under contract no. W-31-109-ENG-38; the National Sci- MS (A2) are D and ω, however, they are not independent: ence Foundationunder grantno.INT-0129236;andben- in fitting, a change in one is compensated by altering efitedfromtheresourcesoftheNationalEnergyResearch the other, with fitted observables changing little along a Scientific Computing Center. trajectory ωD =(0.6GeV)3. Herein we used D =(0.884GeV)2, ω =0.3GeV. (A3) APPENDIX A: MODEL GAP EQUATION A non-confining model is obtained with D =0. Thegapequation’skernelisbuiltfromaproductofthe Equation (A1), is readily solved for the dressed-quark dressed-gluon propagator and dressed-quark-gluon ver- propagator,with the renormalisationconstants fixed via tex. Itcanbecalculatedinperturbationtheorybutthat Eq.(11). Thatsolutionprovidestheelementsusedinthe is inadequate for the study of intrinsically nonperturba- illustration of Sec. III. tivephenomena. Tomakemodel-independentstatements about DCSB one must employ an alternative systematic and chiral symmetry preserving truncation scheme. The leading order term in one such scheme [26] is the renormalisation-group-improved rainbow truncation of the gap equation (Q=p−q): S(p)−1 =Z (iγ·p+mbm) 2 Λ λa λa + G(Q2)Dfree(Q) γ S(q) γ . (A1) µν 2 µ 2 ν Zq [1] K.D.Lane,Phys.Rev.D10,2605(1974); H.D.Politzer, which is justified underbroad conditions [9]. Nucl.Phys. B 117, 397 (1976). [9] H. Leutwyler and A. Smilga, Phys. Rev. D 46, 5607 [2] P.MarisandC.D.Roberts,Phys.Rev.C56,3369(1997). (1992). [3] C.D.RobertsandA.G.Williams,Prog.Part.Nucl.Phys. [10] T.BanksandA.Casher,Nucl.Phys.B169,103(1980). 33, 477(1994). [11] E. Marinari, G. Parisi and C. Rebbi, Phys. Rev. Lett. [4] P.O. Bowman, U.M. Heller and A.G. Williams, Phys. 47, 1795 (1981). Rev.D 66, 014505 (2002). [12] E. Bittner, H. Markum and R. Pullirsch, Nucl. Phys. [5] J.B.Zhang,F.D.R.Bonnet,P.O.Bowman,D.B.Leinwe- Proc. Suppl.96, 189 (2001). ber and A.G. Williams, “Towards the Continuum Limit [13] E.Bittner,M.P.Lombardo,H.MarkumandR.Pullirsch, of the Overlap Quark Propagator in Landau Gauge,” Nucl. Phys. Proc. Suppl.94, 445 (2001). hep-lat/0208037. [14] T. Bielefeld, S. Hands, J.D. Stack and R.J. Wensley, [6] P. Maris, A. Raya, C.D. Roberts and S.M. Schmidt, Phys. Lett. B 416, 150 (1998). “Confinementanddynamicalchiralsymmetrybreaking,” [15] J.J.Verbaarschot,Phys.Rev.Lett.72,2531(1994); J.J. to appear in the proceedings of Quark Nuclear Physics Verbaarschot and T. Wettig, Ann. Rev. Nucl. Part. Sci. 2002, nucl-th/0208071. 50, 343 (2000). [7] In our Euclidean metric: {γ ,γ } =2δ ; γ† =γ ; and [16] M.E. Berbenni-Bitsch, S. Meyer, A. Sch¨afer, J.J. Ver- µ ν µν µ µ a·b= 4 a b . baarschot and T. Wettig, Phys. Rev. Lett. 80, 1146 i=1 i i [8] In deriving Eq. (4), zero modes have been neglected, (1998); M. G¨ockeler, H. Hehl, P.E. Rakow, A. Sch¨afer P 7 andT.Wettig,Phys.Rev.D59,094503(1999);R.G.Ed- JHEP 0107, 018 (2001); A. Duncan, S. Pernice and wards,U.M.Heller,J.E.KiskisandR.Narayanan,Phys. J. Yoo, Phys.Rev. D 65, 094509 (2002). Rev. Lett. 82, 4188 (1999); B.A. Berg, H. Markum, R. [22] P. Maris and P.C. Tandy, Phys. Rev. C 60, 055214 PullirschandT.Wettig,Phys.Rev.D63,014504(2001); (1999). P.H. Damgaard, U.M. Heller, R. Niclasen and K. Rum- [23] K. Zyablyuk,JHEP 0006, 025 (2000). mukainen, Phys. Lett. B 495, 263 (2000); B.A. Berg, [24] F.T.Hawes,C.D.RobertsandA.G.Williams,Phys.Rev. U.M. Heller, H. Markum, R. Pullirsch and W. Sakuler, D 49, 4683 (1994); A. Bender, D. Blaschke, Yu.L. Kali- Phys.Lett. B 514, 97 (2001). novsky and C.D. Roberts, Phys. Rev. Lett. 77, 3724 [17] P. Maris, C.D. Roberts and P.C. Tandy, Phys. Lett. B (1996); F.T. Hawes, P. Maris and C.D. Roberts, Phys. 420, 267 (1998). Lett. B 440, 353 (1998). [18] C.D.RobertsandS.M.Schmidt,Prog.Part.Nucl.Phys. [25] E. Laermann, Fiz. E´lem. Chastits At. Yadra 30 (1999) 45, S1 (2000); R. Alkofer and L.v. Smekal, Phys. Rept. 720 (Phys. Part. Nucl. 30 (1999) 304). 353, 281 (2001). [26] A. Bender, C.D. Roberts and L. v. Smekal, Phys. Lett. [19] M.A.Ivanov,Yu.L.KalinovskyandC.D.Roberts,Phys. B 380,7(1996);A.Bender,W.Detmold,A.W.Thomas Rev.D 60, 034018 (1999). and C.D. Roberts, Phys. Rev.C 65, 065203 (2002). [20] C.D. Roberts, Nucl. Phys.Proc. Suppl. 108, 227 (2002) [21] P. Hernandez, K. Jansen, L. Lellouch and H. Wittig,