Flavor Asymmetry of the Polarized Light Sea: Models vs. Data

The flavor asymmetry of the polarized light sea, ∆ū −∆d̄, discriminates between different model calculations of helicity densities. We show that the chiral chromodielectric model, differently from models based on a 1/Nc expansion, predicts a small value for this asymmetry, what seems in agreement with preliminary HERMES data. 1. Following the discovery that the quark contribution to the spin of the nucleon is surprisingly small [1], a considerable experimental effort was made to elucidate the details of the helicity densities of valence, sea and glue (for reviews on longitudinal spin physics, see [2]). On the other theoretical side, many models have been studied and most of them reproduce the gross features of the spin content of the nucleon, namely the g1 structure function and the singlet distribution ∆Σ = ∑ f(∆f+∆f̄). In order to discriminate between the models, one has to look at the quark and antiquark helicity densities for each separate flavor. This has been a lacking piece of information until last year, when the HERMES Collaboration at DESY, measuring semi-inclusive deep inelastic scattering, succeeded in extracting the polarizations of u, ū, d, d̄, s+ s̄ [3]. Thus, HERMES experiment opened for the first time the possibility to test model results against data. To this purpose, non-singlet distributions are especially interesting because their evolution does not involve the polarized gluon density: as a consequence, they can be predicted in a more reliable way, without any extra assumption on the constituent polarized glue. In what follows, our attention will be directed to the isoscalar and isovector combinations of antiquark densities, f̄+(x) = ū(x) + d̄(x) , f̄−(x) = ū(x)− d̄(x) , (1) ∆f̄+(x) = ∆ū(x) + ∆d̄(x) , ∆f̄−(x) = ∆ū(x)−∆d̄(x) . (2) The two classes of models most widely used for computing quark distributions, i.e. the chiral quark soliton model (CQSM, based on a 1/Nc expansion) and the bag-like confinement models – including the chiral chromodielectric model (CDM) –, make very different predictions for the relative weight of ∆f̄− and ∆f̄+, and of f̄+ and f̄−. In particular, in the 1/Nc expansion, ∆f̄− is a leading quantity compared to ∆f̄+ and f̄−, hence it is expected to be large (in absolute value) and to satisfy the inequalities (which may be, as a matter of fact, strong inequalities) |∆ū−∆d̄| > |ū− d̄| , (3) |∆ū−∆d̄| > |∆ū+∆d̄| . (4)