Distribution of “hot” molecular gas. Left panel: Same as the left panel of Fig. 2 but showing the change with L_X of the concentration of “hot” molecular gas measured for the central regions of the galaxies in our sample derived from the ratio (in log units) of the average H₂ surface densities estimated from the 2.1 μm line emission at two spatial scales: r ≤ 50 pc (${\mathrm{\Sigma }}_{50}^{\mathrm{hot}}$) and r ≤ 200 pc (${\mathrm{\Sigma }}_{200}^{\mathrm{hot}}$). The small (red) squares represent the individual galaxy measurements for AGN in the sample. Concentration indices for the three nonAGN targets are displayed as (gray) square symbols at log(L_X) = 37 erg s⁻¹. The large (green) squares show the median value of the concentration index estimated for the five L_X bins that cover the range of AGN luminosities spanned by our sample, as defined in Sect. 6.1. Symbols for error bars are as in Fig. 2. Right panel: Same as the left panel but showing the two-branch linear solution found by the MARS algorithm to fit the distribution of concentration indices of “hot” molecular gas as a function of L_X for the AGN build-up phase (blue straight line) and the AGN feedback phase (red straight line).