Measurements of Deep Inelastic Scattering (DIS) provide a powerful tool to probe the fundamental structure of protons and other nuclei. The DIS cross sections can be expressed in terms of structure functions which are conventionally expressed in terms of parton distribution functions (PDFs) that obey the DGLAP evolution equations. However, it is also possible to formulate the DGLAP evolution directly in terms of measurable DIS structure functions entirely sidestepping the need for introducing PDFs. We call this as the physical-basis approach. In a global analysis one would thereby directly parametrize the (observable) structure functions -- not the (unobservable) PDFs. Ideally, with data constraints at fixed $Q^2$, the initial condition for the evolution would be the same at each perturbative order (unlike for PDFs) and the approach thus provides a more clean test of the QCD dynamics. We first study a physical basis consisting of the structure functions $F_2$ and $F_{\rm L}$ in the fixed-flavour number scheme to the leading non-zero order in $\alpha_s$. We show how to express the quark singlet and gluon PDFs in terms of $F_2$ and $F_{\rm L}$ directly in momentum space which then leads to the DGLAP evolution of the structure functions $F_2$ and $F_{\rm L}$. In the second step we expand the physical basis to include six independent structure functions, which allows for a consistent global analysis. The steps towards NLO accuracy and the variable-flavour-number scheme are outlined. At NLO accuracy (when the scheme dependence of PDFs starts to play a part), we can take advatage of the physical basis and express e.g. the Drell-Yan cross sections at the LHC directly in terms of measurable DIS structure functions and thus without the scheme dependence.
Comment: 6 pages, 1 figure. Talk by M.T at DIS2023