Comment on the frozen QCD coupling

The frozen QCD coupling is a parameter often used as an effective fixed coupling. It is supposed to mimic both the running coupling effects and the lack of knowledge of α s in the infrared region. Usually the value of the frozen coupling is fixed from the analysis of the experimental data. We present a novel way to define such coupling(s) independently of the experiments. We argue that there are different frozen couplings which are used in double-logarithmic (DL) and single-logarithmic (SL) approximations. We introduce four kinds of frozen couplings: the coupling used in DLA with a time-like argument (i.e. the coupling present in the non-singlet scattering amplitudes and DIS structure functions), which we find to be 0.24 approximately; the DLA coupling with a space-like argument (e + e --annihilation, in DY processes and in any scattering amplitude in the hard or backward kinematics) which is a factor two larger, namely 0.48. We also show that the frozen coupling in the SL evolution equations like BFKL can be defined with much less accuracy compared to DLA, and our estimate for this coupling is 0.1. Our estimates for the singlet and non-singlet intercepts are also in a good agreement with the results available in the literature.