Fig. 4: Recovery of the expected Stokes-Einstein exponent.

a, \({\xi }^{C{M}^{\perp }}\) versus \({r}\). \({\xi }^{C{M}^{\perp }}\to -1.01\pm 0.02\) for \({r} \, > \, 8{\sigma}\); the ideal expected value of \({\xi}\) is shown by black dashed line. Standard error from power-law fittings between \({D}^{C{M}^{\perp }}\) and \({\tau }_{\alpha }^{C{M}^{\perp }}\) are used in \({\xi }^{C{M}^{\perp }}\) versus \({r}\) plot; systematic errors, obtained by extraction of \({D}^{C{M}^{\perp }}\) from different time-windows, are found to be larger than standard error and are used when quoting the value of \({\xi }^{C{M}^{\perp }}\) in the main text and figure captions. Inset: schematic to visualize the direction in which the displacements of particles in the pair are least correlated, i.e., direction perpendicular to the centre-of-mass displacement direction. b, \(\frac{dH}{dr}\) versus \({r}\) at \(\phi=0.61\); \(H=H_L+H_T\). Inset: comparison of \({\xi_L}\) with \({\xi }^{C{M}^{\perp }}\) at discrete values of \({r}\) corresponding to the extrema of \(\frac{dH}{dr}\).