This figure shows the $\Upsilon$(1S) $v_2$ coefficient as a function of transverse momentum in the 5--60\% centrality interval. The central (0--5\%) and peripheral (60--100\%) collisions are not considered as the eccentricity of the initial collision geometry is small for the former and the signal yield is low in the latter. The $p_{\rm T}$ intervals are 0--3, 3--6, and 6--15 GeV/$c$ and the points are located at the average transverse momentum of the reconstructed $\Upsilon$(1S) uncorrected for detector acceptance and efficiency. The results are compatible with zero and with the small positive values predicted by the available theoretical models within uncertainties. The BBJS (former KSU) model calculations consider only the path-length dependent dissociation of initially-created bottomonia inside the QGP medium~\cite{Bhaduri:2018iwr}. The Texas A\&M University (TAMU) model incorporates in addition a regeneration component originating from the recombination of (partially) thermalized bottom quarks~\cite{Du:2017qkv}. Given that the regeneration component gives practically negligible contribution to the total $\Upsilon$(1S) $v_2$, the differences between the two models are marginal.