Intersection between Microscopic and Macroscopic Abelian Dominance in the Connement Physics of QCD Hiroko Ichie Department of Physics,

Tokyo Institute of Technology, Tokyo, 152-8551, Japan Hideo Suganuma Research Center for Nuclear Physics, Osaka University, Osaka, 567-0047, Japan

arXiv:hep-lat/9809006v1 2 Sep 1998

ABSTRACT We study abelian dominance for connement in terms of the local gluon properties in the maximally abelian (MA) gauge, where the diagonal component of the gluon is maximized by the gauge transformation. We nd microscopic abelian dominance on the link-variable for the whole region of in the lattice QCD in the MA gauge. The o-diagonal angle variable, which is not constrained by the MA-gauge xing condition, tends to be random besides the residual gauge degrees of freedom. Within the random-variable approximation for the o-diagonal angle variable, we analytically prove that o-diagonal gluon contribution W o to the Wilson loop obeys the perimeter law in the MA gauge. The perimeterlaw behavior of W o is also conrmed using the lattice QCD simulation. This indicates macroscopic abelian dominance for the string tension.

1.

Introduction

In the low-energy region of QCD, there appear interesting phenomena such as color connement and chiral symmetry breaking reecting the strong gauge-coupling. However, because of nonperturbative and nonabelian nature, these phenomena are dicult to treat analytically, and it is desired to extract the relevant degrees of freedom for description of infrared phenomena. In 1974, Nambu proposed an idea that quark connement can be interpreted using the dual version of the superconductivity.1 In the superconductor, Cooper-pair condensation leads to the Meissner eect, and the magnetic ux is squeezed like a quasi-one-dimensional tube as the Abrikosov vortex. In this dual-superconductor picture for the QCD vacuum, the squeezing of the color-electric ux between quarks is realized by the dual Meissner eect as the result of condensation of color-magnetic monopoles. However, there are two following large gaps between QCD and the dual superconductor picture; 1) This picture is based on the abelian gauge theory, while QCD is a nonabelian gauge theory. 2) The dual-superconductor scenario requires condensation of magnetic monopoles as the key concept, while QCD does not have such a

E-mail : ichie@th.phys.titech.ac.jp

monopole as the elementary degrees of freedom. As the connection between QCD and the dual superconductor scenario, t Hooft proposed the concept of the abelian gauge xing2 with assumption of abelian dominance for the infrared QCD. The abelian gauge xing is dened so as to diagonalize a suitable gauge-dependent variable [A (x)] and reduces QCD into an abelian gauge theory, where the o-diagonal element of the gluon eld behaves as a charged matter eld. Moreover, in the abelian gauge, color-magnetic monopoles appear as topological objects corresponding to the nontrivial homotopy c 1 group 2 (SU(Nc )/U(1)Nc 1 ) = ZN . If monopole condenses, the scenario of color connement by the dual Meissner eect would be realized in QCD. In this paper, with the help of the lattice QCD simulation, we study intersection between abelian dominance of the gluon eld (microscopic variable) and connement force (macroscopic variable) as the theoretical basis of dual superconductor picture. 2. Microscopic Abelian Dominance in the Maximally Abelian Gauge

Abelian dominance on the connement force have been investigated using the lattice QCD simulation in the maximally abelian (MA) gauge.4 In terms of the link 0 a variable U (s) U (s) + i a U (s), the MA gauge xing is dened by maximizing 0 3 1 2 (s))2 +(U (s))2 (U (s))2 (U (s))2 } through R s, tr{U (s)3 U (s)3 } = s,{(U 1 2 the gauge transformation. In the MA gauge, the o-diagonal components, U and U , are forced to be small, and therefore the QCD system seems describable only by U(1)like variables approximately. The MA gauge is a sort of the abelian gauge, because the MA gauge xing diagonalizes (s) , U (s)3 U (s) with U (s) U (s ). In this section, we study abelian dominance on the link variable U (s). In the lattice formalism, the SU(2) link variable U (s) is factorized as U (s) =
cos (s) sin (s)ei (s) sin (s)ei (s) cos (s) ei (s) 0 3 (s) i 0 e
3

M (s)u (s).

Here, the U(1)3 link-variable u (s) corresponds to the diagonal gluon part and behaves as the abelian gauge led in the MA gauge, while M (s) corresponds to the o-diagonal gluon part. In order to investigate abelian dominance on the link variable in the MA gauge, we dene abelian projection rate4 as RAbel = cos (s) [0, 1] with 0 . For instance, the SU(2) link variable becomes completely diagonal if cos = 1, 2 while it becomes o-diagonal if cos = 0. In Fig.1, we show local abelian projection rate RAbel expressed by the arrow (sin , cos ) in a typical conguration of the lattice QCD. In the MA gauge, most of all SU(2) link variables become U(1)-like. For the quantitative argument, we show in Fig.2 the probability distribution P (RAbel ) of the abelian projection rate RAbel . Without gauge xing, one nds the average RAbel = 2 . 3 In the MA gauge, the o-diagonal component of the SU(2) link variable is forced to be reduced, and RAbel approaches to unity; one obtains RAbel MA 0.93 on 164 lattice with = 2.4. Thus, we nd microscopic abelian dominance on the link variable.

15

P( RAbel)

10

0 0.0

0.5

1.0

Fig.1 Local abelian projection rate RAbel cos 4 (0 2 ) at = 2.4 on 16 lattice without gauge xing (left) and in MA gauge xing (right). The arrow expresses (sin , cos ).

RAbel(s, )

Fig.2 The probability distribution P (RAbel ) of abelian projection rate RAbel at = 2.4 on 164 lattice in the MA gauge (solid curve) and without gauge xing (dashed curve).

3.

Semi-analytical Proof of Abelian Dominance for Connement

In the MA gauge, the diagonal element cos (s) in M (s) is maximized by the gauge transformation. Then, the o-diagonal element ei (s) sin (s) is forced to take a small value in the MA gauge, and therefore the approximate treatment on the odiagonal element would be allowed in the MA gauge. Moreover, the angle variable (s) is not constrained by the MA gauge-xing condition at all, and tends to take a random value4 besides the residual U(1)3 gauge degrees of freedom. Hence, (s) in the MA gauge can be regarded as a random angle variable in a good approximation. In this section, we rst investigate properties of (s) using the lattice QCD simulation, and then study the origin of abelian dominance on connement. We examine the randomness of (s) using the lattice QCD simulation in the MA gauge with U(1)3 Landau-gauge xing.4 We show in Fig.3 the probability distribution P () of the correlation (s) mod | (s) (s + )| [0, ], which is the dierence between two neighboring angle variables, at =0, 1.0, 2.4, 3.0. In the strongregion as 1, (s) behaves as a random variable, and there is no correlation between neighboring . On the other hand, in the weak-coupling region, the smallness of sin makes o-diagonal components more irrelevant in the MA gauge, which permits the approximate treatment on (s). Thus, we can take the random-variable approximation for (s) as a good approximation in the whole region of in the MA gauge. Next, let us consider the Wilson loop WC [U (s)] trC U (s) in the MA gauge. In calculating WC [U (s)] , the expectation value of ei (s) in M (s) vanishes as ei (s) 02 d (s) exp{i (s)} = 0 within the random-variable approximation on (s). Then, the o-diagonal factor M (s) appearing in the Wilson loop WC [U (s)] becomes a diagonal matrix, U (s) M (s)u (s) cos (s)u (s). Then, for the I J rectangular C , the Wilson loop WC [U (s)] trL i=1 Ui (si ) in the MA gauge is estimated as
WC [U (s)] trL i=1 cos i (si )ui (si )
MA L = L i=1 cos i (si ) trj =1 uj (sj ) MA , MA

exp{L ln(cos (s))

MA } WC [u (s)]

(1)

where L 2(I + J ) denotes the perimeter length and WC [u (s)] trL i=1 ui (si ) L the abelian Wilson loop. Here, we have replaced i=1 ln{cos(i (si )} by its average L ln{cos (s)} MA in a statistical sense. In this way, we derive a simple estimation as
o WC WC [U (s)] / WC [u (s)] MA

exp{L ln(cos (s))

MA }

(2)

for the contribution of the o-diagonal gluon element to the Wilson loop. From this o analysis, WC is expected to obey the perimeter law in the MA gauge for large loops, where the statistical treatment would be accurate. o In the lattice QCD, we nd that WC seems to obey the perimeter law for the Wilson loop with I, J 2 in the MA gauge (Fig.4). We nd also that the lattice data o of WC as the function of L are well reproduced by the above analytical estimation with microscopic information on cos (s) as ln{cos (s)} MA 0.082 for = 2.4. o Thus, the o-diagonal contribution WC to the Wilson loop obeys the perimeter 1 law in the MA gauge, and therefore the SU(2) string-tension SU(2) limI,J IJ ln WI J [U (s)] coincides with to the abelian string-tension Abel ,
SU(2) = 2 ln{cos (s)}
MA

I +J + Abel IJ

I,J

Abel .

(3)

Thus, abelian dominance for the string tension, SU(2) = Abel , can be proved in the MA gauge by approximating the o-diagonal angle variable (s) as a random variable.
0.6

1
8 7 6 5 4

0.4

P( )

W off
0.2 0.0 0.0

3 2

0.1
8 7 6 5

/2

10

20

30

L = 2(I+J)

Fig.3 The probability distribution P () of the correlation mod (| (s) (s + )|) in the MA gauge with U(1)3 Landau-gauge xing at = 0 (thin line), 1.0 (dotted curve), 2.4 (solid curve), 3.0 (dashed curve).

Fig.4 The comparison between the analytical estimation (straight line) and the lattice data o () of the o-diagonal gluon contribution WC for the Wilson loop as the function of L 2(I + J ) in the MA gauge at = 2.4.

References 1. Y. Nambu, Phys. Rev. D10 (1974) 4262. 2. G. t Hooft, Nucl. Phys. B190 (1981) 455. 3. H. Ichie and H. Suganuma, Proc. of Int. Workshop on Future Directions in Quark Nuclear Physics, 1998, (World Scientic): hep-lat/9807006. 4. H. Ichie and H. Suganuma, preprint, hep-lat/9807025. 5. H. Suganuma, H. Ichie, A. Tanaka and K. Amemiya, Prog. Theor. Phys. Suppl. 131 (1998): hep-lat/9804027, and references.