1 Correction to: J Stat Phys (2017) 167:123–134 https://doi.org/10.1007/s10955-017-1742-x

The original version of this article unfortunately contained a mistake.

Equation (4) should read \(h^{\mu \nu }=\eta ^{\mu \nu }-\frac{1}{c^{2}}\mathscr {U}^{\mu }\mathscr {U}^{\nu }\), the sign being minus due to the \(+--\) signature employed. The expression used in the rest of the manuscript is the correct one, and only Eq. (4) needs to be modified. Also, the right hand side of Eq. (24) should be multiplied by \(-1\) in order to be consistent with Eq. (16). This leads to the relation \(b_{1}=-a_{1}/g\left( z\right) \) (from Eqs. (23) and (24)) and thus to a missing minus sign in Eq. (38). The coefficient \(L_{n}\) is then given by

$$\begin{aligned} L_{n}=-\frac{30mc^{3}}{d}\frac{z^{5}\left( 2+3z\left( 2+z\right) \right) ^{2}}{\left( 3z^{2}+3z+1\right) \left( 1+z\right) }\frac{\exp \left( -\frac{2}{z}\right) }{I\left( z\right) } \end{aligned}$$

and the value plotted in Fig. 1 corresponds to its absolute value \(\left| L_{n}\right| /L_{T_{NR}}\).