Normalized additional velocity distribution: testing the radial profile of dark matter halos and MOND

We propose a complementary and fast approach to study galaxy rotation curves directly from the sample data, instead of individual fits. With this approach, some relevant tests can be done analytically. It is based on a dimensionless difference between the observational rotation curve and the expected one from the baryonic matter ($\delta V^2$) as a function of the normalized radius $r_n$ (i.e., for all galaxies, $0 < r_n < 1$). Using 153 galaxies from the SPARC galaxy sample, we find the observational distribution of $\delta V^2$. Considering radii with $0.2 < r_n < 0.9$, most of the SPARC data are close to the curve $\delta V^2 = r_n^{0.42}$, and about $95\%$ of the SPARC data is between the curves $\delta V^2 = r_n^{2.2}$ and $\delta V^2 = 2 r_n^{0.38} - r_n^{1.9}$. We consider three well known dark matter halo models (NFW, Burkert and DC14), a simple dark matter rotation curve profile for the purpose of model comparison (Arctan$_\alpha$) and one modified gravity model without dark matter (MOND). By comparing the observational data distribution with the model-inferred data, we confirm that the NFW halo lacks the necessary diversity to reproduce several observed rotation curves, while Burkert and DC14 models have better concordance with observational data. The lowest $\delta V^2$ curves that can be found from NFW are linear on the normalized radius (i.e., $\delta V^2_{NFW} = r_n$), while for Burkert $\delta V^2_{Bur} = r_n^2$ (this result is independent of the halo density parameter, i.e., $\rho_{c}$ or $\rho_{s}$). MOND only covers the very central region of the observed distribution, hence it also lacks the necessary diversity, which in turn is related to larger $\chi^2$ values. In a second paper, the method will be extended to consider other classes of modified gravity models.