A convergent reformulation of perturbative QCD
We present and explore a new formulation of perturbative QCD based not on the renormalised coupling but on the dimensional transmutation parameter of the theory and the property of asymptotic scaling. The approach yields a continued function, the iterated function being that involved in the solution of the two-loop β-function equation. In the so-called large-b limit the continued function reduces to a continued fraction and the successive approximants are diagonal Padé approximants. We investigate numerically the convergence of successive approximants using the leading-b approximation, motivated by renormalons, to model the all-orders result. We consider the Adler D-function of vacuum polarisation, the Polarised Bjorken and Gross-Llewellyn Smith sum rules, the (unpolarised) Bjorken sum rule, and the Minkowskian quantities R(_r) and the R-ratio of e(^+)e(^-) annihilation. In contrast to diagonal Fade approximants the truncated continued function method gives remarkably stable large-order approximants in cases where infra-red renormalon effects are important. We also use the new approach to determine the QCD fundamental parameters from the R(_r) and the R-ratio measurements, where we find Ā(^(3))(_MS)=516±48 MeV (which yields a(_s)(µ=m(_r))=0.360(^+0.021)(_=0.020)), and Ā(^(5))(_MS)=299(^+6)(_-7) MeV (which yields a(_s)(µ=m(_zo)=0.1218±0.0004), respectively. The evolution of the former value to the m(_zo) energy results in a(_s)(µ= m(_zo)) = 0.123 ± 0.002. These values are in line with other determinations available in the literature. We implement the Complete Renormalisation Group Improvement (CORGI) scheme throughout all the calculations. We report on how the mathematical concept of Stieltjes series can be used to assess the convergence of Padé approximants of perturbative series. We find that the combinations of UV renormalons which occur in perturbative QCD may or may not be Stieltjes series depending on the renormalisation scheme used.