Anomalous dipion invariant mass distribution of the $\Upsilon(4S)$ decays into $\Upsilon(1S) \pi^{+} \pi^{-}$ and $\Upsilon(2S) \pi^{+} \pi^{-}$

To solve the discrepancy between the experimental data on the partial widths and lineshapes of the dipion emission of $\Upsilon(4S)$ and the theoretical predictions, we suggest that there is an additional contribution which was not taken into account in previous calculations. Noticing that the mass of $\Upsilon(4S)$ is above the production threshold of $B\bar B$, the contribution of the sequential process $\Upsilon(4S)\to B\bar B\to \Upsilon(nS)+S\to \Upsilon(nS)+\pi^+\pi^-$ ($n=1,2$) may be sizable, and its interference with that from the direct production would be important. The goal of this work is to investigate if a sum of the two contributions with a relative phase indeed reproduces the data. Our numerical results on the partial widths and the lineshapes $d\Gamma(\Upsilon(4S)\to \Upsilon(2S,1S)\pi^+\pi^-)/d(m_{\pi^+\pi^-})$ are satisfactorily consistent with the measurements, thus the role of this mechanism is confirmed. Moreover, with the parameters obtained by fitting the data of the Belle and Babar collaborations, we predict the distributions $d\Gamma(\Upsilon(4S)\to \Upsilon(2S,1S)\pi^+\pi^-)/d\cos\theta$ which have not been measured yet.