Leopold Infeld Colloquium

The finite order perturbative predictions of the quantum chromodynamics depend on the choice of the renormalization scheme. Formally this dependence is a higher order effect, but at lower energies, where the effective coupling parameter is not so small, it becomes numerically significant. To improve the reliability of QCD predictions it is proposed to use a modified perturbation expansion, based on a modified couplant, which is free from Landau singularity and is much less scheme dependent itself. The modified couplant is obtained by integrating the renormalization group equation with an appropriately constructed nonpolynomial generator. The predictions obtained with the modified expansion appear to be much more stable with respect to the change of the renormalization scheme parameters than the predictions obtained in the conventional approach. As a concrete example we discuss the QCD correction to the sum rule in the deep inelastic scattering. It is also found that the modified predictions display somewhat weaker energy dependence than the predictions obtained with the conventional expansion, which may be interesting from the point of view of comparing the low energy and high energy determinations of the strong coupling parameter.