-
We describe the production amplitudes of
$ e^{+}e^{-} $ annihilation into$ \Omega^{-} \bar{\Omega}^{+} $ hyperons using the helicity formalism introduced by Jacob and Wick in their seminal work [31]. In this formalism, the helicity angle$ \theta_{\Omega^{-}} $ corresponds to the polar angle of the$ \Omega^{-} $ particle in the center-of-mass (c.m.) frame of the$ e^{+}e^{-} $ collision and serves as the variable of interest. The helicity system for the$ \Omega^{-} \bar{\Omega}^{+} $ production process is defined and illustrated in Fig. 1.Figure 1. (color online) In the helicity formalism, the helicity angle
$ \theta_{\Omega^{-}} $ is defined as the polar angle of the$ \Omega^{-} $ particle in the center-of-mass (c.m.) system of the$ e^+e^- $ collision. The$ \hat{z} $ -axis in the$ e^+e^- $ c.m. system corresponds to the direction of the incoming positron.In the helicity framework, the production process of
$ e^+e^-\rightarrow \gamma^*\rightarrow \psi(3686)\rightarrow \Omega^-\bar{\Omega}^+ $ can be described as [32]:$ \begin{equation} \begin{aligned} \rho^{\lambda_{1},\lambda_{2};\lambda_{1}^{'},\lambda_{2}^{'}}\propto A_{\lambda_{1},\lambda_{2}}A^{*}_{\lambda_{1}^{'},\lambda_{2}^{'}} \rho^{\lambda_{1}-\lambda_{2},\lambda_{1}^{'}-\lambda_{2}^{'}}_{1}(\theta_{\Omega^-}),\\ \rho^{i,j}_{1}:=\sum_{\kappa=\pm1}D^{1*}_{\kappa,i}(0,\theta_{\Omega^-},0)D^{1}_{\kappa,j}(0,\theta_{\Omega^-},0), \end{aligned} \end{equation} $
(1) where D represents the Wigner transformation matrix,
$ \theta_{\Omega^{-}} $ is the helicity angle of the$ \Omega^{-} $ particle,$ A_{\lambda_1,\lambda_2} $ ($ A_{\lambda_1',\lambda_2'} $ ) denotes the transition amplitude with helicities$ \lambda_1 $ ($ \lambda_1' $ ) and$ \lambda_2 $ ($ \lambda_2' $ ) of the$ \Omega^{-} $ and$ \bar{\Omega}^{+} $ particles, respectively, and κ represents the helicity difference between the initial$ e^{+} $ and$ e^{-} $ states, which can only take values of$ \pm 1 $ . Since the$ e^{+}e^{-} $ beam is non-polarized, it is necessary to sum over the helicity difference κ. If only the$ \Omega^{-} $ particle is reconstructed, without considering the decays of the recoil side, the density function for a single$ \Omega^{-} $ particle can be obtained by requiring$ \lambda_2 = \lambda_{2}^{\prime} $ and summing over the helicity states:$ \begin{equation} \begin{aligned} \rho_{\Omega^-}\propto \sum_{\lambda_{2}}A_{\lambda_{1},\lambda_{2}}A^{*}_{\lambda_{1}^{'},\lambda_{2}}\rho_{1}^{\lambda_{1}-\lambda_{2},\lambda_{1}^{'}-\lambda_{2}}(\theta_{\Omega^-}). \end{aligned} \end{equation} $
(2) Due to the conservation of parity and charge conjugate invariance, only four transition amplitudes are obtained:
$ \begin{equation} \begin{aligned} H_1 :=& A_{\frac{1}{2},\frac{1}{2}}=A_{-\frac{1}{2},-\frac{1}{2}},\\ H_2 :=& A_{\frac{1}{2},-\frac{1}{2}}=A_{-\frac{1}{2},\frac{1}{2}},\\ H_3 :=& A_{\frac{3}{2},\frac{1}{2}}=A_{-\frac{3}{2},-\frac{1}{2}}\\=&A_{\frac{1}{2},\frac{3}{2}}=A_{-\frac{1}{2},-\frac{3}{2}},\\ H_4 :=& A_{\frac{3}{2},\frac{3}{2}}=A_{-\frac{3}{2},-\frac{3}{2}}.\\ \end{aligned} \end{equation} $
(3) The transition amplitude matrix is then given by
$ \begin{equation} \left( \begin{array}{cccc} H_4 & H_3 & 0 &0\\ H_3 & H_1 & H_2 &0\\ 0 & H_2 & H_1 & H_3\\ 0 & 0 & H_3 & H_4\\ \end{array} \right). \end{equation} $
(4) In the experimental data analysis, by restricting oneself to the azimuthal dependence of the reaction cross-section and polarization, the ratios of the amplitudes can be considered without the need to consider their absolute magnitudes. These ratios are defined as follows:
$H_1/H_2= h_1 {\rm e}^{{\rm i}\phi_1}$ ,$H_3/H_2=h_3{\rm e}^{{\rm i}\phi_3}$ ,$H_4/H_2=h_4{\rm e}^{{\rm i}\phi_4}$ , where$ h_i $ and$ \phi_i $ (i = 1, 3, 4) are real variables obtained through experimental fitting. Based on the aforementioned theoretical framework, BESIII has conducted an analysis of the angular distribution of the single-tag process$\psi(3686)\to \Omega^{-}\bar{\Omega}^{+}$ ($ \Omega^-\to K^-\Lambda $ ,$ \Lambda\to p\pi^- $ ) using a data set of$ 448\times 10^6 \psi(3686) $ decays collected by the BESIII detector at the BEPCII electron-positron collider [30]. The analysis results are presented in Table 1.parameter solution I solution II $ h_1 $ 0.30±0.11±0.04 0.31±0.10±0.04 $ \phi_1 $ 0.69±0.41±0.13 2.38±0.37±0.13 $ h_3 $ 0.26±0.05±0.02 0.27±0.05±0.01 $ \phi_3 $ 2.60±0.16±0.08 2.57±0.16±0.04 $ h_4 $ 0.51±0.03±0.01 0.51±0.03±0.01 $ \phi_4 $ 0.34±0.80±0.31 1.37±0.68±0.16 Table 1. Two sets of fit values of the helicity parameters of
$ \psi(3686)\to \Omega^{-}\bar{\Omega}^{+} $ decays in BESIII measurements [30]. The first uncertainties are statistical, and the second ones systematic. -
The description of particles with spin can be achieved using a spin density matrix ρ in the rest frame of the particle, as discussed in literature [36]. The spin density matrix enables the characterization of the spin state of a system by introducing irreducible spin components up to rank
$ 2J $ . For example, the most general spin density matrix for a spin-1/2 particle is given by the following form, as described in [36]:$ \begin{align} \rho_{1/2} = \frac{1}{2}(\mathit{\boldsymbol{1}}+ S^i\mathbf{\sigma}^i), \end{align} $
(5) where
$ \mathbf{\sigma}^i $ represents the set of Pauli matrices. The (rank-one) spin vector$ S^i $ corresponds to the transverse ($ S_{x} $ ,$ S_{y} $ ) and longitudinal ($ S_{L} $ ) vector polarizations of the particle.To parametrize the density matrix of a spin-3/2 particle, we can utilize a Cartesian basis comprising 4 × 4 matrices. This basis includes the identity matrix and three spin matrices denoted as
$ \Sigma^i $ , which serve as a generalization of the Pauli matrices to the four-dimensional case:$ \begin{aligned}[b] &\Sigma^x=\frac{1}{2}\left( \begin{array}{cccc} 0 & \sqrt{3} & 0 & 0 \\ \sqrt{3} & 0 & 2 & 0 \\ 0 & 2 & 0 & \sqrt{3} \\ 0 & 0 & \sqrt{3} & 0 \\ \end{array} \right), \\ & \Sigma^y=\frac{\rm i}{2}\left( \begin{array}{cccc} 0 & -\sqrt{3} & 0 & 0 \\ \sqrt{3} & 0 & -2 & 0 \\ 0 & 2 & 0 & -\sqrt{3} \\ 0 & 0 & \sqrt{3} & 0 \\ \end{array} \right), \\&\Sigma^z=\frac{1}{2}\left( \begin{array}{cccc} 3 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & -1 & 0 \\ 0 & 0 & 0 & -3 \\ \end{array} \right). \end{aligned} $
(6) In addition to these, there are five extra rank-2 matrices, denoted
$ \Sigma^{ij} $ , and seven extra rank-3 matrices, denoted$ \Sigma^{ijk} $ [28, 34]. These matrices provide a comprehensive representation for the density matrix in the$ s_z $ basis and can be constructed from the spin matrices of$ (\Sigma^{i})s $ . The (rank-2) spin tensor basis ($ \Sigma^{ij} $ )s are given by$ \begin{equation} \begin{aligned} &\Sigma^{ij}=\frac{1}{2}(\Sigma^i\Sigma^j+\Sigma^j\Sigma^i)-\frac{5}{4} \delta^{ij} {\bf{1}}. \end{aligned} \end{equation} $
(7) Five independent spin tensor basis
$ \Sigma^{zx} $ ,$ \Sigma^{zy} $ ,$ \Sigma^{zz} $ ,$ \Sigma^{yx} $ and ($ \Sigma^{xx}-\Sigma^{yy} $ ) are selected. Similarly, we construct the (rank-3) spin tensor basis ($ \Sigma^{ijk} $ )s as:$ \begin{equation} \begin{aligned} \Sigma^{ijk}=\frac{1}{6}\Sigma^{\{i}\Sigma^j\Sigma^{k\}}-\frac{41}{60}(\delta^{ij}\Sigma^k+\delta^{jk}\Sigma^i +\delta^{ki}\Sigma^j). \end{aligned} \end{equation} $
(8) where the symbol
$ \{\cdots\} $ represents the symmetrization of the indices, indicating the sum over all permutations. Finally, seven independent spin tensor basis$ \Sigma^{xzz} $ ,$ \Sigma^{yzz} $ ,$ \Sigma^{zzz} $ ,$ \Sigma^{xyz} $ , ($ \Sigma^{xxz}-\Sigma^{yyz} $ ), ($ \Sigma^{xxx}-3\Sigma^{xyy} $ ), and (3$ \Sigma^{xxy}-\Sigma^{yyy} $ ) are selected.With these preliminaries, the spin density matrix can be expressed as follows [34]:
$ \rho_{3/2}=\frac{1}{4}\left(\mathit{\boldsymbol{1}} + \frac{4}{5}S^i\Sigma^i + \frac{2}{3}T^{ij}\Sigma^{ij} + \frac{8}{9}R^{ijk}\Sigma^{ijk}\right), $
(9) where we introduce the symmetric traceless (rank-2) spin tensor
$ T^{ij} $ and the (rank-3) spin tensor$ R^{ijk} $ . The spin density matrix consists of 15 spin components:$ \begin{equation} \begin{aligned} S^{i}&: S_{L}, S^{x}_{T}, S^{y}_{T},\\ T^{ij}&: S_{LL}, S^{x}_{LT}, S^{y}_{LT}, S_{TT}^{xx}, S_{TT}^{xy},\\ R^{ijk}&: S_{LLL}, S^{x}_{LLT}, S^{y}_{LLT},S_{LTT}^{xx}, S_{LTT}^{xy}, S_{TTT}^{xxx}, S_{TTT}^{yxx}.\\ \end{aligned} \end{equation} $
(10) These spin components can be described as the probabilities of finding specific eigenstates from the spin density matrix
$ \rho_{3/2} $ , as detailed in Appendix A. The domains of the spin components can be determined through probabilistic interpretations:$ \begin{equation} \begin{aligned} &S_{L}, S^{x}_{T}, S^{y}_{T}\in[-\frac{3}{2},\frac{3}{2}],\\ &S_{LL}\in[-1,1],\quad S^{x}_{LT}, S^{y}_{LT}, S_{TT}^{xy}, S_{TT}^{xx}\in[-\sqrt{3},\sqrt{3}],\\ &S_{LLL}\in[-\frac{9}{10},\frac{9}{10}],\quad S^{x}_{LLT}, S^{y}_{LLT}\in[-\frac{3+\sqrt{21}}{10},\frac{3+\sqrt{21}}{10}],\\ &S_{LTT}^{xx},S_{LTT}^{xy}\in[-\sqrt{3},\sqrt{3}],\quad S_{TTT}^{xxx}, S_{TTT}^{yxx}\in[-3,3].\\ \end{aligned} \end{equation} $
(11) To simplify the notation, we adopt a shorthand notation where a single index is used to represent the spin matrices:
$ \begin{aligned}[b] \Sigma_0 &=\frac{1}{4} {\bf{1}}, \\ \Sigma_1 &=\frac{1}{5} \Sigma^z,\quad \Sigma_2=\frac{1}{5} \Sigma^x,\quad \Sigma_3=\frac{1}{5} \Sigma^y ,\\ \Sigma_4 &=\frac{1}{4} \Sigma^{z z},\quad \Sigma_5=\frac{1}{6} \Sigma^{x z},\quad \Sigma_6=\frac{1}{6} \Sigma^{y z},\\ \Sigma_7 &=\frac{1}{12}\left(\Sigma^{x x}-\Sigma^{y y}\right),\quad \Sigma_8=\frac{1}{6} \Sigma^{x y}, \\ \Sigma_9 &=\frac{5}{9} \Sigma^{z z z},\quad \Sigma_{10}=\frac{5}{6} \Sigma^{x z z},\quad \Sigma_{11}=\frac{5}{6} \Sigma^{y z z},\\ \Sigma_{12}&=\frac{1}{6}\left(\Sigma^{x x z}-\Sigma^{y y z}\right), \\ \Sigma_{13} &=\frac{1}{3} \Sigma^{x y z}, \quad\Sigma_{14}=\frac{1}{18}\left(\Sigma^{x x x}-3 \Sigma^{x y y}\right), \\\Sigma_{15}&=\frac{1}{18}\left(3 \Sigma^{x x y}-\Sigma^{y y y}\right), \end{aligned} $
(12) where
${\bf{1}}$ is the 4$ \times $ 4 identity matrix. The explicit expression for these basis matrices is given in Appendix B. The corresponding spin density matrix of a spin-3/2 particle can then be expressed as follows using the shorthand notation:$ \begin{equation} \begin{aligned} \rho_{3/2}=\Sigma_{\mu=0}^{15}S_{\mu}\Sigma_{\mu}, \end{aligned} \end{equation} $
(13) where
$ S_{0} $ represents the cross-section term and$ S_{\mu} $ are 15 real numbers that correspond to the 15 spin components. By using the chosen matrix basis, we have established a direct mapping between the decomposition of spin components in the helicity formalism and the decomposition of spin components in terms of the density matrix [34]. The relationship between the polarization components$ S_{L} $ ,$ S^{x}_{T} \cdots $ and the expansion coefficients$ S_{\mu} $ is presented in Table 2.$1$ $S_{L}$ $S^{x}_{T}$ $S^{y}_{T}$ $S_{LL}$ $S^{x}_{LT}$ $S^{y}_{LT}$ $S^{xx}_{TT}$ $S^{xy}_{TT}$ $S_{LLL}$ $S^{x}_{LLT}$ $S^{y}_{LLT}$ $S^{xx}_{LTT}$ $S^{xy}_{LTT}$ $S^{xxx}_{TTT}$ $S^{yxx}_{TTT}$ $S_{0}$ $S_{1}$ $S_{2}$ $S_{3}$ $S_{4}$ $S_{5}$ $S_{6}$ $S_{7}$ $S_{8}$ $S_{9}$ $S_{10}$ $S_{11}$ $S_{12}$ $S_{13}$ $S_{14}$ $S_{15}$ Table 2. The correspondences of the expansion coefficients
$ S_{\mu} $ and the spin components.The density matrix of
$ \Omega^{-} $ can also be expressed in terms of the helicity amplitudes given in Eq. (2). Due to the parity-conserving nature of the process, it is found that only seven parameters are non-zero, namely$ S_{0} $ ,$ S_{3} $ ,$ S_{4} $ ,$ S_{5} $ ,$ S_{7} $ ,$ S_{11} $ , and$ S_{13} $ :$ \begin{equation} \begin{aligned} S_0 &=2 \sin ^2 \theta_{\Omega^-}\left(|H_1|^2 + |H_4|^2\right) + \left(1 + \cos ^2 \theta_{\Omega^-}\right)\left(|H_2|^2 + 2|H_3|^2\right) ,\\ S_3 &=\frac{1}{\sqrt{2}} \sin 2 \theta_{\Omega^-}\left(2 \operatorname{Im}\left[H_2 H_1^*\right]+\sqrt{3} \operatorname{Im}\left[H_3\left(H_1^*+H_4^*\right)\right]\right) ,\\ S_4 &=2 \sin ^2 \theta_{\Omega^-}\left(|H_4|^2-|H_1|^2\right)-\left(1+\cos ^2 \theta_{\Omega^-}\right)|H_2|^2 ,\\ S_5 &=\sqrt{6} \sin 2 \theta_{\Omega^-} \operatorname{Re}\left[\left(H_4-H_1\right) H_3^*\right], \\ S_7 &=2 \sqrt{3} \sin ^2 \theta_{\Omega^-} \operatorname{Re}\left[H_2 H_3^*\right] ,\\ S_{11} &=\frac{\sqrt{2}}{5} \sin \left(2 \theta_{\Omega^-}\right)\left(3 \operatorname{Im}\left[H_1 H_2^*\right]+\sqrt{3} \operatorname{Im}\left[H_3\left(H_1^*+H_4^*\right)\right]\right), \\ S_{13} &=2 \sqrt{3} \sin ^2 \theta_{\Omega^-} \operatorname{Im}\left[H_2 H_3^*\right]. \end{aligned} \end{equation} $
(14) These parameters characterize the spin properties of the
$ \Omega^{-} $ particle in the considered process.With the measured helicity amplitudes provided in Table 1, we investigated the polarization dependence of the reaction cross-section as a function of the azimuthal angle
$ \cos\theta_{\Omega^{-}} $ , as shown in Fig. 2. The uncertainties, which include both statistical and systematic contributions, were estimated using the experimentally determined covariance matrix of the fitted$ h_i $ and$ \phi_i $ parameters [30]. In the$ e^{+}e^{-}\to \psi(3686)\to \Omega^{-}\bar{\Omega}^{+} $ process, the$ \Omega^{-} $ particles exhibit not only transverse vector polarization ($ S_3 $ ) but also contributions from rank-2 tensor polarization ($ S_4 $ ,$ S_5 $ ,$ S_7 $ ) and rank-3 tensor polarization ($ S_{11} $ ,$ S_{13} $ ). All of these spin components were computed using the helicity amplitudes listed in Table 1. Indeed, it is worth noting that both sets of solutions yield the same overall effect. For the sake of clarity and simplicity, we choose to present only one set of solutions in our analysis and figures. -
In Section III, we provided the Cartesian form of the spin density matrix for
$ \Omega^{-} $ , which is parameterized by vector polarizations and tensor polarizations represented by the coefficients$ S_{\mu} $ . Each component of the spin state corresponds to a combination of probabilities, indicating the likelihood of finding the system in a specific spin state defined in the particle's rest frame. In this section, we focus on analyzing the polarization dependence of the cross-section along the coordinate axes and provide a physical interpretation for the observed behaviors. -
The spin operator can be defined in any direction; for our analysis, we choose to express it in terms of the eigenstates of the spin vector operator in a specific direction. This allows us to write the spin vector operator in terms of polar and azimuthal angles [20].
$ \begin{align} \Sigma^i\hat n^i=\Sigma^x \sin\theta\cos\phi +\Sigma^y \sin\theta\sin\phi +\Sigma^z \cos\theta, \end{align} $
(15) As per the common convention, we denote the eigenstate with eigenvalue m along the direction
$ (\theta,\phi) $ as$ |m_{(\theta,\phi)}\rangle $ . The eigenstates of$ \Sigma^z $ ,$ \Sigma^x $ , and$ \Sigma^y $ are given by$ \begin{aligned}[b] &\left|\frac{3}{2}\right\rangle_z=\left(\begin{array}{c} 1 \\ 0 \\ 0 \\ 0 \end{array}\right), \,\left|\frac{1}{2}\right\rangle_z=\left(\begin{array}{l} 0 \\ 1 \\ 0 \\ 0 \end{array}\right),\\& \,\left|-\frac{1}{2}\right\rangle_z=\left(\begin{array}{l} 0 \\ 0 \\ 1 \\ 0 \end{array}\right), \,\left|-\frac{3}{2}\right\rangle_z=\left(\begin{array}{c} 0 \\ 0 \\ 0 \\ 1 \end{array}\right), \\ &\left|\frac{3}{2}\right\rangle_x=\frac{\sqrt{2}}{4}\left(\begin{array}{c} 1 \\ \sqrt{3} \\ \sqrt{3} \\ 1 \end{array}\right), \,\left|\frac{1}{2}\right\rangle_x=\frac{\sqrt{2}}{4}\left(\begin{array}{c} \sqrt{3} \\ 1 \\ -1 \\ -\sqrt{3} \end{array}\right),\\& \,\left|-\frac{1}{2}\right\rangle_x=\frac{\sqrt{2}}{4}\left(\begin{array}{c} \sqrt{3} \\ -1 \\ -1 \\ \sqrt{3} \end{array}\right), \,\left|-\frac{3}{2}\right\rangle_x=\frac{\sqrt{2}}{4}\left(\begin{array}{c} 1 \\ -\sqrt{3} \\ \sqrt{3} \\ -1 \end{array}\right), \\ &\left|\frac{3}{2}\right\rangle_y=\frac{\sqrt{2}}{4}\left(\begin{array}{c} 1 \\ {\rm i} \sqrt{3} \\ -\sqrt{3} \\ -{\rm i} \end{array}\right), \,\left|\frac{1}{2}\right\rangle_y=\frac{\sqrt{2}}{4}\left(\begin{array}{c} \sqrt{3} \\ {\rm i} \\ 1 \\ {\rm i} \sqrt{3} \end{array}\right),\\& \,\left|-\frac{1}{2}\right\rangle_y=\frac{\sqrt{2}}{4}\left(\begin{array}{c} \sqrt{3} \\ -{\rm i} \\ 1 \\ -{\rm i} \sqrt{3} \end{array}\right), \,\left|-\frac{3}{2}\right\rangle_y=\frac{\sqrt{2}}{4}\left(\begin{array}{c} 1 \\ -{\rm i} \sqrt{3} \\ -\sqrt{3} \\ {\rm i} \end{array}\right), \end{aligned} $
(16) where
$ \Sigma^{i}|m\rangle_i = m |m\rangle_i $ , such as$\Sigma^{z}\left|\dfrac{3}{2}\right\rangle_z = \dfrac{3}{2} \left|\dfrac{3}{2}\right\rangle_z$ . We have used a more intuitive notation (x, y and z) to denote the direction angles; the relation between the two notations is:$ \begin{equation} |m\rangle_z=|m_{(0,0)}\rangle, \quad|m\rangle_x=|m_{\left(\frac{\pi}{2}, 0\right)}\rangle, \quad|m\rangle_y=|m_{\left(\frac{\pi}{2}, \frac{\pi}{2}\right)}\rangle. \end{equation} $
(17) The probability of observing one of these eigenstates is defined as follows:
$ \begin{equation} P_i(m)\equiv {_i}\langle m|\rho_{3/2} |m\rangle_{i}. \end{equation} $
(18) Inserting Eqs. (13) and (16) into (18), one readily obtains:
$ \begin{aligned}[b] P_z\left(\frac{3}{2}\right) =&\frac{1}{4}+\frac{3}{10} S_L+\frac{1}{4} S_{L L}+\frac{1}{6} S_{L L L}=\frac{1}{4}+\frac{S_4}{4},\\ P_z\left(\frac{1}{2}\right) =&\frac{1}{4}+\frac{1}{10} S_L-\frac{1}{4} S_{L L}-\frac{1}{2} S_{L L L}=\frac{1}{4}-\frac{S_4}{4},\\ P_z\left(-\frac{1}{2}\right) =&\frac{1}{4}-\frac{1}{10} S_L-\frac{1}{4} S_{L L}+\frac{1}{2} S_{L L L}=\frac{1}{4}-\frac{S_4}{4},\\ P_z\left(-\frac{3}{2}\right) =&\frac{1}{4}-\frac{3}{10} S_L+\frac{1}{4} S_{L L}-\frac{1}{6} S_{L L L}=\frac{1}{4}+\frac{S_4}{4},\\ P_x\left(\frac{3}{2}\right) =&\frac{1}{4}+\frac{3}{10} S_T^x-\frac{1}{8} S_{L L}+\frac{1}{8} S_{T T}^{x x}-\frac{1}{8} S_{L L T}^x\\&+\frac{1}{24} S_{T T T}^{x x x}=\frac{1}{4}-\frac{S_4}{8} +\frac{S_7}{8} ,\\ P_x\left(\frac{1}{2}\right) =&\frac{1}{4}+\frac{1}{10} S_T^x+\frac{1}{8} S_{L L}-\frac{1}{8} S_{T T}^{x x}+\frac{3}{8} S_{L L T}^x\\&-\frac{1}{8} S_{T T T}^{x x x}=\frac{1}{4}+\frac{S_4}{8} -\frac{S_7}{8} ,\\ P_x\left(-\frac{1}{2}\right) =&\frac{1}{4}-\frac{1}{10} S_T^x+\frac{1}{8} S_{L L}-\frac{1}{8} S_{T T}^{x x}-\frac{3}{8} S_{L L T}^x\\&+\frac{1}{8} S_{T T T}^{x x x}=\frac{1}{4}+\frac{S_4}{8} -\frac{S_7}{8} ,\\ P_x\left(-\frac{3}{2}\right) =&\frac{1}{4}-\frac{3}{10} S_T^x-\frac{1}{8} S_{L L}+\frac{1}{8} S_{T T}^{x x}+\frac{1}{8} S_{L L T}^x\\&-\frac{1}{24} S_{T T T}^{x x x}=\frac{1}{4}-\frac{S_4}{8} +\frac{S_7}{8} ,\\ P_y\left(\frac{3}{2}\right) =&\frac{1}{4}+\frac{3}{10} S_T^y-\frac{1}{8} S_{L L}-\frac{1}{8} S_{T T}^{x x}-\frac{1}{8} S_{L L T}^y\\&-\frac{1}{24} S_{T T T}^{y x x}=\frac{1}{4}+\frac{3 S_3}{10} -\frac{S_4}{8} -\frac{S_7}{8} -\frac{S_{11}}{8} ,\\ P_y\left(\frac{1}{2}\right) =&\frac{1}{4}+\frac{1}{10} S_T^y+\frac{1}{8} S_{L L}+\frac{1}{8} S_{T T}^{x x}+\frac{3}{8} S_{L L T}^y\\&+\frac{1}{8} S_{T T T}^{y x x}=\frac{1}{4}+\frac{S_3}{10} +\frac{S_4}{8} +\frac{S_7}{8} +\frac{3S_{11}}{8} ,\\ P_y\left(-\frac{1}{2}\right) =&\frac{1}{4}-\frac{1}{10} S_T^y+\frac{1}{8} S_{L L}+\frac{1}{8} S_{T T}^{x x}-\frac{3}{8} S_{L L T}^y\\&-\frac{1}{8} S_{T T T}^{y x x}=\frac{1}{4}-\frac{S_3}{10} +\frac{S_4}{8} +\frac{S_7}{8} -\frac{3S_{11}}{8} ,\\ P_y\left(-\frac{3}{2}\right) =&\frac{1}{4}-\frac{3}{10} S_T^y-\frac{1}{8} S_{L L}-\frac{1}{8} S_{T T}^{x x}+\frac{1}{8} S_{L L T}^y\\&+\frac{1}{24} S_{T T T}^{y x x}=\frac{1}{4}-\frac{3S_3}{10} -\frac{S_4}{8} -\frac{S_7}{8} +\frac{S_{11}}{8} . \end{aligned} $
(19) In the intermediate expressions of Eq. (19), we have included the complete results, which are valid for calculations in any process. However, in the
$e^{+}e^{-}\to \psi(3686)\to \Omega^{-}\bar{\Omega}^{+}$ process, certain polarization components, such as$ S_L $ and$ S_{LLL} $ , are found to be zero. Therefore, in the final expression on the right-hand side, we have omitted these zero polarization components. The values of these probabilities are then determined exclusively by the non-zero parameters in Eq. (14). The ranges of the aforementioned spin components, as dictated by their corresponding physical interpretations in Appendix A, are presented in Eq. (11).According to Eq. (19), it is evident that the influence of
$ S_{LL} $ extends beyond the z-axis and encompasses the polarization along the x- and y-axes as well. This peculiar polarization dependence observed in spin-3/2 particles contradicts the straightforward behavior exhibited by spin-1/2 particles. In essence, when the$ \Omega^- $ particle possesses a non-zero$ S_{LL} $ component, it not only induces disparities in polarization probabilities along the z-axis ($ P_z(\pm\frac{3}{2})\neq P_z(\pm\frac{1}{2}) $ ) but also gives rise to deviations in polarization probabilities along the x and y-axes ($ P_{x/y}(\pm\frac{3}{2})\neq P_{x/y}(\pm\frac{1}{2}) $ ). The complexity arises from the intrinsic nature of spin-3/2, where the observation of spin in any direction can yield four distinct states. Investigating alternative forms of spin representation may prove worthwhile in mitigating this complexity. Nevertheless, the physical implications conveyed by Eq. (19) underscore the substantial advantages of projecting the spin of$ \Omega^- $ particles along the coordinate axes, as this approach simplifies the intricate polarization dependence of the$ \Omega^- $ particle, facilitating a more intuitive understanding of its spin structure.Employing the computed multipolar polarization operators as described in Eq. (14), we present a graph (Fig. 3) that illustrates the polarization dependence of the cross-section along the x, y, and z-axes with respect to
$ \cos\theta_{\Omega^{-}} $ . All calculations were performed utilizing the helicity amplitudes presented in Table 1. From Fig. 3, it is evident that the$ \Omega^{-} $ particle exhibits significant tensor polarization along the z-axis in the$ \psi(3686)\rightarrow \Omega^{-}\bar{\Omega}^{+} $ process. This is due to a substantial difference in the probability of occupying eigenstates with eigenvalues of$\pm {3}/{2}$ and$\pm {1}/{2}$ along the z-axis. Additionally, it demonstrates tensor polarization along the x-axis, which gradually increases in the direction of the electron beam and reaches its maximum at$ \theta_{\Omega^{-}}=0 $ or π. Furthermore, notable tensor polarization is observed along the y-axis, accompanied by weak transverse vector polarization and rank-3 tensor polarization. This is attributed to a significant difference in the probability of occupying eigenstates with eigenvalues of$\pm {3}/{2}$ and$\pm {1}/{2}$ , while only a minor difference is observed in the probability of occupying eigenstates with eigenvalues of$ {3}/{2} $ and$ - {3}/{2} $ or$ {1}/{2} $ and$ - {1}/{2} $ .Figure 3. (color online) The cross-section displays the polarization dependence along the z-axis (left), x-axis (middle), and y-axis (right) as a function of the azimuthal angle
$ \cos\theta_{\Omega^{-}} $ . The solid lines represent the central values, and the shaded areas represent$ \pm $ one standard deviation -
In order to provide a theoretical interpretation for the experimental measurements, we calculate the polarization dependence of the cross-section in the process of
$ e^+e^- $ annihilation into$ \Omega^{-}\bar{\Omega}^{+} $ using a point-like vertex, as depicted in Fig. 4. Since both the photon and$ \psi(3686) $ are vector particles, we neglect the influence of the vector boson$ \psi(3686) $ on the cross-section. The calculations are performed in the center-of-mass frame of the leptons, as shown in Fig. 1. Neglecting the electronic mass, the momentum components of$ e^+e^- $ and$ \Omega^{-}\bar{\Omega}^{+} $ can be expressed as follows:Figure 4. Graph describing the reaction
$ e^{+}e^{-}\rightarrow \Omega^{-}\bar{\Omega}^{+} $ in the one-photon approximation.$ \begin{align} k_{1} & = \left\{ E,0,0,-E\right\} , \end{align} $
(20) $ \begin{align} k_{2} & = \left\{ E,0,0,E\right\} , \end{align} $
(21) $ \begin{align} p_{1} & = \left\{ E,|\vec{p}|\sin\theta_{\Omega},0,|\vec{p}|\cos\theta_{\Omega}\right\} , \end{align} $
(22) $ \begin{align} p_{2} & = \left\{ E,-|\vec{p}|\sin\theta_{\Omega},0,-|\vec{p}|\cos\theta_{\Omega}\right\} . \end{align} $
(23) In the
$ e^+e^- \rightarrow \Omega^- \bar{\Omega}^+ $ process, the cross-section can be expressed in terms of the scattering amplitude. When considering the polarization dependence of the cross-section along the helicity direction, we project the polarizations of the initial and final state particles onto their respective helicity directions. The expression for the cross-section is given by$ \begin{align} \frac{{\rm d}\sigma^{\lambda_{1}}}{{\rm d}\Omega} = \frac{1}{64\pi^{2}s}\frac{|\vec{p}|}{E}\frac{1}{4}\sum_{\lambda=\pm}\sum_{\lambda_{2}=\pm\frac{1}{2},\pm\frac{3}{2}}|\mathcal{M}_{-\lambda,\lambda,\lambda_{1},\lambda_{2}}|^{2}, \end{align} $
(24) where λ,
$ \lambda_{1} $ , and$ \lambda_{2} $ represent the helicities of the positron,$ \Omega^{-} $ , and$ \bar{\Omega}^{+} $ , respectively. The scattering amplitude$ \mathcal{M} $ can be expressed as follows:$ \begin{aligned}[b] \mathcal{M}_{-\lambda,\lambda,\lambda_{1},\lambda_{2}} = & \frac{g^{\mu\nu}g^{\rho\sigma}}{q^{2}}\left[\bar{u}^{\rho}\left(p_{1},\lambda_{1}\right)\left(-{\rm i}e\gamma^{\mu}\right)v^{\sigma}\left(p_{2},\lambda_{2}\right)\right]\\&\times\left[\bar{v}\left(k_{2},\lambda\right)\left(-{\rm i}e\gamma^{\nu}\right)u\left(k_{1},-\lambda\right)\right], \end{aligned} $
(25) where
$ \bar{u}^{\rho}(v^{\sigma}) $ and$ u(\bar{v}) $ represent the polarization vectors for$ \Omega^{-} $ ($ \bar{\Omega}^{+} $ ) and$ e^{-} $ ($ e^{+} $ ), respectively, following the conventions described in Appendix C. When investigating the dependence of the cross-section on the transverse polarization of$ \Omega^- $ , it is essential to decompose its spin within its own helicity frame, specifically along the direction$ \hat{x}_1 $ or$ \hat{y}_1 $ , as illustrated in Fig. 1. However, the polarization states of the other particles are not of interest, and thus we sum over their spins. It is important to note that any spin decomposition for these particles along different directions does not affect the final result. Therefore, we continue to employ their spin decomposition along their own helicity direction. We can now express the dependence of the cross-section on the transverse polarization of$ \Omega^- $ as follows:$ \begin{align} \frac{{\rm d}\sigma^{s_{1}\uparrow_{x/y}}}{{\rm d}\Omega} = \frac{1}{64\pi^{2}s}\frac{|\vec{p}|}{E}\frac{1}{4}\sum_{\lambda=\pm}\sum_{\lambda_{2}=\pm\frac{1}{2},\pm\frac{3}{2}}|\mathcal{M}_{-\lambda,\lambda,s_{1}\uparrow_{x/y},\lambda_{2}}|^{2}, \end{align} $
(26) where we introduce the notation
$ s_{1}\uparrow_{x/y} $ to represent the transverse polarization state of the$ \Omega^{-} $ particle. The expressions for all the scattering amplitudes are provided in Appendix D. Based on the aforementioned considerations, we derive the polarization dependence of the cross-section as follows:$ \begin{aligned}[b] \frac{{\rm d}\sigma^{\pm\frac{3}{2}}}{{\rm d}\Omega} = & \frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\Bigg\{ \left(a^{2}+\frac{m^{2}}{E^{2}}\right)\\&+\left(a^{2}-\frac{m^{2}}{E^{2}}\right)\cos^{2}\theta_{\Omega}\Bigg\} , \end{aligned} $
(27) $ \begin{aligned}[b]\\[-3pt] \frac{{\rm d}\sigma^{\pm\frac{1}{2}}}{{\rm d}\Omega} = & \frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\left\{ \left[a^{2}+b^{4}\left(\frac{2E^{2}}{m^{2}}-1\right)^{2}+\left(\frac{2E}{m}b^{2}-\frac{m}{E}\left(a^{2}+b^{2}\right)\right)^{2}\right]\right.\\ & \left.+\left[a^{2}+b^{4}\left(\frac{2E^{2}}{m^{2}}-1\right)^{2}-\left(\frac{2E}{m}b^{2}-\frac{m}{E}\left(a^{2}+b^{2}\right)\right)^{2}\right]\cos^{2}\theta_{\Omega}\right\} , \end{aligned} $ (28) $ \begin{aligned}[b] \frac{{\rm d}\sigma^{\pm\frac{3}{2}\uparrow_{x}}}{{\rm d}\Omega} = &\frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\left\{ \left[a^{2}\left(1-3b^{2}\right)+\frac{\sqrt{3}}{2}ab^{2}-\frac{9}{4}b^{4}\right.\right. \left.+\frac{3E^{4}}{m^{4}}b^{4}-\frac{\sqrt{3}E^{2}}{m^{2}}ab^{2}+\frac{m^{2}}{4E^{2}}\left(1+3\left(a^{2}+b^{2}\right)^{2}\right)\right]\\ & +\left[a^{2}\left(1+3b^{2}\right)-\frac{\sqrt{3}}{2}ab^{2}+\frac{15}{4}b^{4}+\frac{3E^{4}}{m^{4}}b^{4}\right.\left.\left.+\frac{E^{2}}{m^{2}}b^{2}\left(\sqrt{3}a-6b^{2}\right)-\frac{m^{2}}{4E^{2}}\left(1+3\left(a^{2}+b^{2}\right)^{2}\right)\right]\cos^{2}\theta_{\Omega}\right\} , \end{aligned} $
(29) $ \begin{aligned}[b] \frac{{\rm d}\sigma^{\pm\frac{1}{2}\uparrow_{x}}}{{\rm d}\Omega} = &\frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\left\{ \left[a^{2}\left(1-b^{2}\right)-\frac{\sqrt{3}}{2}ab^{2}-\frac{3}{4}b^{4}\right.\right.\left.+\frac{E^{4}}{m^{4}}b^{4}+\frac{\sqrt{3}E^{2}}{m^{2}}ab^{2}+\frac{m^{2}}{4E^{2}}\left(3+\left(a^{2}+b^{2}\right)^{2}\right)\right]\\ & +\left[a^{2}\left(1+b^{2}\right)+\frac{\sqrt{3}}{2}ab^{2}+\frac{5}{4}b^{4}+\frac{E^{4}}{m^{4}}b^{4}\right.\left.\left.-\frac{E^{2}}{m^{2}}b^{2}\left(\sqrt{3}a+2b^{2}\right)-\frac{m^{2}}{4E^{2}}\left(3+\left(a^{2}+b^{2}\right)^{2}\right)\right]\cos^{2}\theta_{\Omega}\right\} , \end{aligned} $
(30) $ \begin{aligned}[b] \frac{{\rm d}\sigma^{\pm\frac{3}{2}\uparrow_{y}}}{{\rm d}\Omega} & =\frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\left\{ \left[a^{2}\left(1-3b^{2}\right)-\frac{\sqrt{3}}{2}ab^{2}-\frac{9}{4}b^{4}\right.\right.\left.+\frac{3E^{4}}{m^{4}}b^{4}+\frac{\sqrt{3}E^{2}}{m^{2}}ab^{2}+\frac{m^{2}}{4E^{2}}\left(1+3\left(a^{2}+b^{2}\right)^{2}\right)\right]\\ & +\left[a^{2}\left(1+3b^{2}\right)+\frac{\sqrt{3}}{2}ab^{2}+\frac{15}{4}b^{4}+\frac{3E^{4}}{m^{4}}b^{4}\right.\left.\left.-\frac{E^{2}}{m^{2}}b^{2}\left(\sqrt{3}a+6b^{2}\right)-\frac{m^{2}}{4E^{2}}\left(1+3\left(a^{2}+b^{2}\right)^{2}\right)\right]\cos^{2}\theta_{\Omega}\right\} , \end{aligned} $
(31) $ \begin{aligned}[b] \frac{{\rm d}\sigma^{\pm\frac{1}{2}\uparrow_{y}}}{{\rm d}\Omega} = & \frac{\alpha^{2}}{32E^{2}}\sqrt{1-\frac{m^{2}}{E^{2}}}\left\{ \left[a^{2}\left(1-b^{2}\right)+\frac{\sqrt{3}}{2}ab^{2}-\frac{3}{4}b^{4}\right.\right.\left.+\frac{E^{4}}{m^{4}}b^{4}-\frac{\sqrt{3}E^{2}}{m^{2}}ab^{2}+\frac{m^{2}}{4E^{2}}\left(3+\left(a^{2}+b^{2}\right)^{2}\right)\right]\\ & +\left[a^{2}\left(1+b^{2}\right)-\frac{\sqrt{3}}{2}ab^{2}+\frac{5}{4}b^{4}+\frac{E^{4}}{m^{4}}b^{4}\right.\left.\left.+\frac{E^{2}}{m^{2}}b^{2}\left(\sqrt{3}a-2b^{2}\right)-\frac{m^{2}}{4E^{2}}\left(3+\left(a^{2}+b^{2}\right)^{2}\right)\right]\cos^{2}\theta_{\Omega}\right\} , \end{aligned} $
(32) where m denotes the mass of the
$ \Omega^- $ particle, and a and b are normalization coefficients satisfying$ a^2 + b^2 = 1 $ . In the SU(6) quark model, one can determine$ a = \sqrt{1/3} $ and$ b = \sqrt{2/3} $ , as presented in Appendix C. We retain a and b as coefficients and demonstrate the dependence of the cross-section on a and b. It is important to note that altering the values of a and b introduces certain challenges elaborated upon in the subsequent paragraph.In Fig. 5, we present the theoretical predictions for the polarization dependence of the cross-section in the
$ e^{+}e^{-}\rightarrow \Omega^{-}\bar{\Omega}^{+} $ process. The corresponding experimental measurements are also included for comparison. Based on the graphical representation, it is apparent that the azimuth-angle dependence of the polarization cross-section agrees with the experimental measurements when$ a=\sqrt{1/3} $ and$ b=\sqrt{2/3} $ . Nevertheless, when we allow the coefficients a and b to vary, it leads to a corresponding variation in the polarization dependence of the cross-section. Remarkably, when$ a=\sqrt{1/6} $ and$ b=\sqrt{5/6} $ are chosen, our calculations exhibit a notably improved agreement with the experimental measurements reported in Ref. [30]. This observation suggests that the field equation of$ \Omega^{-} $ may deviate from the predictions of the SU(6) quark model. Additionally, the coefficients a and b play a significant role in eliminating the spin-1/2 component from the field equation, indicating that$ \Omega^{-} $ may not exclusively exist in a pure spin-3/2 state. This intriguing finding offers novel insights for future investigations and extensive discussions regarding the nature and potential dynamics of$ \Omega^{-} $ . It is important to acknowledge that our calculations did not consider the influence of intermediate resonance states ($ \psi(3686) $ ) and form factors of$ \Omega^{-} $ on the cross-section. The incorporation of these factors presents intriguing prospects for future research and additional exploration.Figure 5. (color online) The dashed lines represent the polarization dependence for
$ e^{+}e^{-}\rightarrow \Omega^{-}\bar{\Omega}^{+} $ at$ \sqrt{s}=3.686 $ GeV, calculated theoretically using Eqs. (27)–(32). The solid lines correspond to the results obtained from the BESIII experiment [30]. -
To present the explicit expression of the probabilistic interpretations of spin polarization using Eq. (7) and Eq. (8), we first decompose the spin operators for spin-3/2 particles into the following forms:
$ \begin{aligned}[b]\\[-7pt] &\Sigma^{xz}=\frac{1}{4}(\Sigma^{x}+\Sigma^{z})^{2}-\frac{1}{4}(\Sigma^{x}-\Sigma^{z})^{2},\\ &\Sigma^{yz}=\frac{1}{4}(\Sigma^{y}+\Sigma^{z})^{2}-\frac{1}{4}(\Sigma^{y}-\Sigma^{z})^{2},\\ &\Sigma^{zz}=(\Sigma^{z})^2- \frac{5}{4},\\ &\Sigma^{xy}=\frac{1}{4}(\Sigma^{x}+\Sigma^{y})^{2}-\frac{1}{4}(\Sigma^{x}-\Sigma^{y})^{2},\\ &\Sigma^{xx}-\Sigma^{yy}=(\Sigma^{x})^{2}-(\Sigma^{y})^{2},\\ &\Sigma^{xzz}=\frac{1}{3}[\frac{1}{2}(\Sigma^{x}+\Sigma^{z})^{3}+\frac{1}{2}(\Sigma^{x}-\Sigma^{z})^{3}-(\Sigma^{x})^{3}]-\frac{41}{60}\Sigma^{x},\\ &\Sigma^{yzz}=\frac{1}{3}[\frac{1}{2}(\Sigma^{y}+\Sigma^{z})^{3}+\frac{1}{2}(\Sigma^{y}-\Sigma^{z})^{3}-(\Sigma^{y})^{3}]-\frac{41}{60}\Sigma^{y},\\ &\Sigma^{zzz}=(\Sigma^{z})^3- \frac{41}{20}\Sigma^{z},\\ &\Sigma^{xyz}=\frac{1}{6}[(\Sigma^{x}+\Sigma^{y}+\Sigma^{z})^{3}-(\Sigma^{x}+\Sigma^{y})^{3}-(\Sigma^{x}+\Sigma^{z})^{3}\\&\quad\quad-(\Sigma^{y}+\Sigma^{z})^{3}+(\Sigma^{x})^{3}+(\Sigma^{y})^{3}+(\Sigma^{z})^{3}],\\ &\Sigma^{xxz}-\Sigma^{yyz}=\frac{1}{6}[(\Sigma^{x} + \Sigma^{z})^{3} - (\Sigma^{x} - \Sigma^{z})^{3}-(\Sigma^{y}+\Sigma^{z})^{3} + (\Sigma^{y}-\Sigma^{z})^{3}],\\ &\Sigma^{xxx}-3\Sigma^{xyy}=2(\Sigma^{x})^{3}-\frac{1}{2}(\Sigma^{x}+\Sigma^{y})^{3}-\frac{1}{2}(\Sigma^{x}-\Sigma^{y})^{3},\\ &3\Sigma^{yxx}-\Sigma^{yyy}=-2(\Sigma^{y})^{3}+\frac{1}{2}(\Sigma^{y}+\Sigma^{x})^{3}+\frac{1}{2}(\Sigma^{y}-\Sigma^{x})^{3}.\\ \end{aligned} \tag{A1} $
Within this framework, we define the eigenstates
$ |m_{(\theta,\phi)}\rangle $ associated with the spin projection operator along the direction$ (\theta,\phi) $ . Here, ϕ represents the azimuthal angle and θ the polar angle. These eigenstates have corresponding eigenvalues denoted by m. In addition, for the sake of clarity and convenience, we introduce the following notation:$ \begin{aligned}[b] |m\rangle_{x+y}&=|m_{(\frac{\pi}{2},\frac{\pi}{4})}\rangle, \quad|m\rangle_{x+z}=|m_{\left(\frac{\pi}{4}, 0\right)}\rangle, \quad|m\rangle_{y+z}=|m_{\left(\frac{\pi}{4}, \frac{\pi}{2}\right)}\rangle,\\ |m\rangle_{x-y}&=|m_{(\frac{\pi}{2},-\frac{\pi}{4})}\rangle, \quad|m\rangle_{x-z}=|m_{\left(-\frac{\pi}{4}, 0\right)}\rangle, \quad|m\rangle_{y-z}=|m_{\left(-\frac{\pi}{4}, \frac{\pi}{2}\right)}\rangle,\\ |m\rangle_{x+y+z}&=|m_{(\theta_{xyz},\frac{\pi}{4})}\rangle,\\ \end{aligned} \tag{A2} $
where
$ \theta_{xyz} $ = arctan($ \sqrt{2} $ ).The probability of the system being in a particular eigenstate can be determined by evaluating the spin density matrix using the formula given by Eq. (18). Subsequently, all the spin components can be expressed in terms of the probabilities
$ P(m_{(\theta,\phi)}) $ s. The three spin components of the spin vector$ S^{i} $ are given by [34]:$ S_{L}=\frac{3}{2}[P_{z}(\frac{3}{2})-P_{z}(-\frac{3}{2})]+\frac{1}{2}[P_{z}(\frac{1}{2})-P_{z}(-\frac{1}{2})], \tag{A3} $
$ S_{T}^{x}= \frac{3}{2}[P_{x}(\frac{3}{2})-P_{x}(-\frac{3}{2})]+ \frac{1}{2}[P_{x}(\frac{1}{2})-P_{x}(-\frac{3}{2})], \tag{A4} $
$ S_{T}^{y}= \frac{3}{2}[P_{y}(\frac{3}{2})-P_{y}(-\frac{3}{2})]+ \frac{1}{2}[P_{y}(\frac{1}{2})-P_{y}(-\frac{1}{2})]. \tag{A5} $
The five spin components of the rank-2 spin tensor
$ T^{ij} $ are given by [34]:$ S_{LL} = [P_{z}(\frac{3}{2})+P_{z}(-\frac{3}{2})]-[P_{z}(\frac{1}{2})+P_{z}(-\frac{1}{2})], \tag{A6} $
$ S^{x}_{LT}= 2[P_{x+z}(\frac{3}{2})+P_{x+z}(-\frac{3}{2})]- 2[P_{x-z}(\frac{3}{2})+P_{x-z}(-\frac{3}{2})], \tag{A7} $
$ S^{y}_{LT} = 2[P_{y+z}(\frac{3}{2})+P_{y+z}(-\frac{3}{2})]- 2[P_{y-z}(\frac{3}{2})+P_{y-z}(-\frac{3}{2})], \tag{A8} $
$ S_{TT}^{xx} = 2[P_{x}(\frac{3}{2})+P_{x}(-\frac{3}{2})]- 2[P_{y}(\frac{3}{2})+P_{y}(-\frac{3}{2})],\tag{A9} $
$ S_{TT}^{xy} = 2[P_{x+y}(\frac{3}{2})+P_{x+y}(-\frac{3}{2})] -2[P_{x-y}(\frac{3}{2})+P_{x-y}(-\frac{3}{2})]. \tag{A10} $
The seven spin components of the rank-3 spin tensor
$ R^{ijk} $ are given by [34]:$ S_{LLL} = \frac{3}{10}[P_{z}(\frac{3}{2})-P_{z}(-\frac{3}{2})]-\frac{9}{10}[P_{z}(\frac{1}{2})-P_{z}(-\frac{1}{2})], \tag{A11} $
$ \begin{aligned}[b]\\[-10pt] S_{LLT}^{x} &= \frac{\sqrt{2}}{24}\{27[P_{x+z}(\frac{3}{2})-P_{x+z}(-\frac{3}{2})]+ [P_{x+z}(\frac{1}{2})-P_{x+z}(-\frac{1}{2})]\}+\frac{\sqrt{2}}{24}\{27[P_{x-z}(\frac{3}{2})-P_{x-z}(-\frac{3}{2})] +[P_{x-z}(\frac{1}{2})-P_{x-z}(-\frac{1}{2})]\}\\ &-\frac{1}{60}\{129[P_{x}(\frac{3}{2})-P_{x}(-\frac{3}{2})] +23[P_{x}(\frac{1}{2})-P_{x}(-\frac{1}{2})]\}, \end{aligned} \tag{A12} $ $ \begin{aligned}[b] S_{LLT}^{y} =& \frac{\sqrt{2}}{24}\{27[P_{y+z}(\frac{3}{2})-P_{y+z}(-\frac{3}{2})]+ [P_{y+z}(\frac{1}{2})-P_{y+z}(-\frac{1}{2})]\}+\frac{\sqrt{2}}{24}\{27[P_{y-z}(\frac{3}{2})-P_{y-z}(-\frac{3}{2})] +[P_{y-z}(\frac{1}{2})-P_{y-z}(-\frac{1}{2})]\}\\ &-\frac{1}{60}\{129[P_{y}(\frac{3}{2})-P_{y}(-\frac{3}{2})] +23[P_{y}(\frac{1}{2})-P_{y}(-\frac{1}{2})]\}, \end{aligned} \tag{A13} $
$ \begin{aligned}[b] S_{LTT}^{xy} =& \frac{1}{12}\{27[P_{z}(\frac{3}{2})-P_{z}(-\frac{3}{2})]+[P_{z}(\frac{1}{2})-P_{z}(-\frac{1}{2})]\}+\frac{1}{12}\{27[P_{x}(\frac{3}{2})-P_{x}(-\frac{3}{2})] +[P_{x}(\frac{1}{2})-P_{x}(-\frac{1}{2})]\}\\ &+\frac{1}{12}\{27[P_{y}(\frac{3}{2})-P_{y}(-\frac{3}{2})] +[P_{y}(\frac{1}{2})-P_{y}(-\frac{1}{2})]\}-\frac{\sqrt{2}}{6}\{27[P_{x+z}(\frac{3}{2})-P_{x+z}(-\frac{3}{2})] +[P_{x+z}(\frac{1}{2})-P_{x+z}(-\frac{1}{2})]\}\\ &-\frac{\sqrt{2}}{6}\{27[P_{y+z}(\frac{3}{2})-P_{y+z}(-\frac{3}{2})] +[P_{y+z}(\frac{1}{2})-P_{y+z}(-\frac{1}{2})]\}-\frac{\sqrt{2}}{6}\{27[P_{x+y}(\frac{3}{2})-P_{x+y}(-\frac{3}{2})] +[P_{x+y}(\frac{1}{2})-P_{x+y}(-\frac{1}{2})]\}\\ &+\frac{\sqrt{3}}{4}\{27[P_{x+y+z}(\frac{3}{2})-P_{x+y+z}(-\frac{3}{2})] +[P_{x+y+z}(\frac{1}{2})-P_{x+y+z}(-\frac{1}{2})]\}. \end{aligned}\tag{A14} $
$ \begin{aligned}[b] S_{LTT}^{xx} =& \frac{\sqrt{2}}{12}\{27[P_{x+z}(\frac{3}{2})-P_{x+z}(-\frac{3}{2})] +[P_{x+z}(\frac{1}{2})-P_{x+z}(-\frac{1}{2})]\}-\frac{\sqrt{2}}{12}\{27[P_{x-z}(\frac{3}{2})-P_{x-z}(-\frac{3}{2})] +[P_{x-z}(\frac{1}{2})-P_{x-z}(-\frac{1}{2})]\}\\ &-\frac{\sqrt{2}}{12}\{27[P_{y+z}(\frac{3}{2})-P_{y+z}(-\frac{3}{2})] +[P_{y+z}(\frac{1}{2})-P_{y+z}(-\frac{1}{2})]\}+\frac{\sqrt{2}}{12}\{27[P_{y-z}(\frac{3}{2})-P_{y-z}(-\frac{3}{2})] +[P_{y-z}(\frac{1}{2})-P_{y-z}(-\frac{1}{2})]\}, \end{aligned} \tag{A15} $
$ \begin{aligned}[b] S_{TTT}^{xxx} =& \frac{1}{4}\{27[P_{x}(\frac{3}{2})-P_{x}(-\frac{3}{2})] +[P_{x}(\frac{1}{2})-P_{x}(-\frac{1}{2})]\}-\frac{\sqrt{2}}{8}\{27[P_{x+y}(\frac{3}{2})-P_{x+y}(-\frac{3}{2})] +[P_{x+y}(\frac{1}{2})-P_{x+y}(-\frac{1}{2})]\}\\ &-\frac{\sqrt{2}}{8}\{27[P_{x-y}(\frac{3}{2})-P_{x-y}(-\frac{3}{2})] +[P_{x-y}(\frac{1}{2})-P_{x-y}(-\frac{1}{2})]\}, \end{aligned}\tag{A16} $
$ \begin{aligned}[b] S_{TTT}^{yxx} &= -\frac{1}{4}\{27[P_{y}(\frac{3}{2})-P_{y}(-\frac{3}{2})] +[P_{y}(\frac{1}{2})-P_{y}(-\frac{1}{2})]\}+\frac{\sqrt{2}}{8}\{27[P_{y+x}(\frac{3}{2})-P_{y+x}(-\frac{3}{2})] +[P_{y+x}(\frac{1}{2})-P_{y+x}(-\frac{1}{2})]\}\\ &+\frac{\sqrt{2}}{8}\{27[P_{y-x}(\frac{3}{2})-P_{y-x}(-\frac{3}{2})] +[P_{y-x}(\frac{1}{2})-P_{y-x}(-\frac{1}{2})]\}. \end{aligned} \tag{A17}$
-
In order to describe the density matrix of a spin-3/2 particle, we employ a set of
$ \Sigma_{\mu} $ matrices, where μ takes values from 0 to 15. This set consists of 16 4$ \times $ 4 matrices. Here, we provide the explicit expressions for the$ \Sigma_{\mu} $ matrices:$ \begin{aligned}[b]\\[-4pt] &\Sigma_{0}=\frac{1}{4}\left( \begin{array}{cccc} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1 \\ \end{array} \right), \Sigma_{1}=\frac{1}{10}\left( \begin{array}{cccc} 3 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & -1 & 0 \\ 0 & 0 & 0 & -3 \\ \end{array} \right), \quad \Sigma_{2}=\frac{1}{10}\left( \begin{array}{cccc} 0 & \sqrt{3} & 0 & 0 \\ \sqrt{3} & 0 & 2 & 0 \\ 0 & 2 & 0 & \sqrt{3} \\ 0 & 0 & \sqrt{3} & 0 \\ \end{array} \right),\\ & \Sigma_{3}=\frac{{\rm i}}{10}\left( \begin{array}{cccc} 0 & -\sqrt{3} & 0 & 0 \\ \sqrt{3} & 0 & -2 & 0 \\ 0 & 2 & 0 & -\sqrt{3} \\ 0 & 0 & \sqrt{3} & 0 \\ \end{array} \right), \quad \Sigma_{4}=\frac{1}{4}\left( \begin{array}{cccc} 1 & 0 & 0 & 0 \\ 0 & -1 & 0 & 0 \\ 0 & 0 & -1 & 0 \\ 0 & 0 & 0 & 1 \\ \end{array} \right), \quad \Sigma_{5}=\frac{\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & 1 & 0 & 0 \\ 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & -1 \\ 0 & 0 & -1 & 0 \\ \end{array} \right),\\ & \Sigma_{6}=\frac{{\rm i}\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & -1 & 0 & 0 \\ 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 \\ 0 & 0 & -1 & 0 \\ \end{array} \right), \quad \Sigma_{7}=\frac{\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ \end{array} \right), \quad \Sigma_{8}=\frac{{\rm i}\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & 0 & -1 & 0 \\ 0 & 0 & 0 & -1 \\ 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ \end{array} \right),\\ & \Sigma_{9}=\frac{1}{6}\left( \begin{array}{cccc} 1 & 0 & 0 & 0 \\ 0 & -3 & 0 & 0 \\ 0 & 0 & 3 & 0 \\ 0 & 0 & 0 & -1 \\ \end{array} \right), \quad \Sigma_{10}=\frac{\sqrt{3}}{6}\left( \begin{array}{cccc} 0 & 1 & 0 & 0 \\ 1 & 0 & -\sqrt{3} & 0 \\ 0 & -\sqrt{3} & 0 & 1 \\ 0 & 0 & 1 & 0 \\ \end{array} \right), \quad \Sigma_{11}=\frac{{\rm i}\sqrt{3}}{6}\left( \begin{array}{cccc} 0 & -1 & 0 & 0 \\ 1 & 0 & \sqrt{3} & 0 \\ 0 & -\sqrt{3} & 0 & -1 \\ 0 & 0 & 1 & 0 \\ \end{array} \right) \end{aligned} $
$ \begin{aligned}[b] & \Sigma_{12}=\frac{\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & -1 \\ 1 & 0 & 0 & 0 \\ 0 & -1 & 0 & 0 \\ \end{array} \right), \quad \Sigma_{13}=\frac{{\rm i}\sqrt{3}}{12}\left( \begin{array}{cccc} 0 & 0 & -1 & 0 \\ 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \\ 0 & -1 & 0 & 0 \\ \end{array} \right), \quad \Sigma_{14}=\frac{1}{6}\left( \begin{array}{cccc} 0 & 0 & 0 & 1 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 1 & 0 & 0 & 0 \\ \end{array} \right)\\ &\Sigma_{15}= \frac{{\rm i}}{6}\left( \begin{array}{cccc} 0 & 0 & 0 & -1 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 1 & 0 & 0 & 0 \\ \end{array} \right).\\ \end{aligned} \tag{B1}$
-
In the following section, we investigate the polarization state of a particle (u) or an antiparticle (v), which is characterized by its polarization vector. We consider a particle or antiparticle with a mass M and a four-momentum
$ p^{\mu} = \left(E,|\vec{p}|\sin\theta\cos\phi,|\vec{p}|\sin\theta\sin\phi,|\vec{p}|\cos\theta\right). \tag{C1} $
For spin-1/2, spin-1, and spin-3/2 particles and antiparticles, we adopt the Pauli-Dirac representation.
The spin-1/2 field is derived by solving the Dirac equation, as detailed in Ref. [37]. The polarization vectors for spin-1/2 particles with helicity
$ \lambda=\pm $ are given by$ u\left(\vec{p},\pm\right) = \left(\begin{array}{c} \sqrt{E+m}\chi_{\pm}\\ \pm\sqrt{E-m}\chi_{\pm} \end{array}\right), \tag{C2} $
$ v\left(\vec{p},\pm\right) = \left(\begin{array}{c} \sqrt{E-m}\chi_{\mp}\\ \mp\sqrt{E+m}\chi_{\mp} \end{array}\right), \tag{C3}$
where
$ \chi_{\pm} $ represent the two-component spinors:$ \chi_+ = \left(\begin{array}{c} \cos\dfrac{\theta}{2}\\ \sin\dfrac{\theta}{2}{\rm e}^{{\rm i}\phi} \end{array}\right),\quad \chi_- = \left(\begin{array}{c} -\sin\dfrac{\theta}{2}\\ \cos\dfrac{\theta}{2}{\rm e}^{{\rm i}\phi} \end{array}\right). \tag{C4}$
The spin-1 field is derived by solving the Proca equation, as outlined in Ref. [38]. The polarization vectors for spin-1 particles with helicity
$ \lambda=\pm,0 $ are given by$ \varepsilon^{\mu}\left(\vec{p},\pm\right) = \frac{1}{\sqrt{2}}\left(\begin{array}{c} 0\\ \mp\cos\theta\cos\phi+{\rm i}\sin\phi\\ \mp\cos\theta\sin\phi-{\rm i}\cos\phi\\ \pm\sin\theta \end{array}\right), \tag{C5}$
$ \varepsilon^{\mu}\left(\vec{p},0\right) = \frac{1}{m}\left(\begin{array}{c} |\vec{p}|\\ E\sin\theta\cos\phi\\ E\sin\theta\sin\phi\\ E\cos\theta \end{array}\right). \tag{C6} $
The spin-3/2 field is derived by solving the Rarita-Schwinger equation, as outlined in Ref. [39]. The polarization vectors for spin-3/2 particles with helicity
$ \lambda=\pm {3}/{2},~\pm {1}/{2} $ are given by$ u^{\mu}\left(p_{z},\pm\frac{3}{2}\right) = \varepsilon_{\pm}^{\mu}\left(p_{z}\right)u_{\pm}\left(p_{z}\right), \tag{C7} $
$ u^{\mu}\left(p_{z},\pm\frac{1}{2}\right) = a\varepsilon_{\pm}^{\mu}\left(p_{z}\right)u_{\mp}\left(p_{z}\right)+b\varepsilon_{0}^{\mu}\left(p_{z}\right)u_{\pm}\left(p_{z}\right), \tag{C8} $
$ v^{\mu}\left(p_{z},\pm\frac{3}{2}\right) = \varepsilon_{\pm}^{\mu*}\left(p_{z}\right)v_{\pm}\left(p_{z}\right), \tag{C9} $
$ v^{\mu}\left(p_{z},\pm\frac{1}{2}\right) = a\varepsilon_{\pm}^{\mu*}\left(p_{z}\right)v_{\mp}\left(p_{z}\right)+b\varepsilon_{0}^{\mu*}\left(p_{z}\right)v_{\pm}\left(p_{z}\right), \tag{C10} $
where a and b are normalization coefficients that satisfy
$ a^{2}+b^{2}=1 $ . In the context of the SU(6) quark model, the values$ a=\sqrt{1/3} $ and$ b=\sqrt{2/3} $ are determined to remove the spin-1/2 fermions. -
When investigating the longitudinal polarization dependence of the reaction cross-section, we compute the helicity scattering amplitude. Considering the principles of parity conservation and charge conjugation invariance, we derive the following independent terms:
$ |\mathcal{M}_{+,-,\pm\frac{3}{2},\pm\frac{3}{2}}|^{2} =|\mathcal{M}_{-,+,\pm\frac{3}{2},\pm\frac{3}{2}}|^{2}=e^{4}\frac{m^{2}}{E^{2}}\sin^{2}\theta, \tag{D1}$ $ |\mathcal{M}_{+,-,\pm\frac{1}{2},\pm\frac{1}{2}}|^{2} =|\mathcal{M}_{-,+,\pm\frac{1}{2},\pm\frac{1}{2}}|^{2}=e^{4}\left[\frac{2b^{2}E^{2}-\left(a^{2}+b^{2}\right)m^{2}}{mE}\right]^{2}\sin^{2}\theta, \tag{D2} $
$ |\mathcal{M}_{+,-,\pm\frac{1}{2},\mp\frac{1}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{1}{2},\pm\frac{1}{2}}|^{2}=e^{4}b^{4}\left(\frac{2E^{2}-m^{2}}{m^{2}}\right)^{2}\left(1\mp\cos\theta\right)^{2},\tag{D3} $
$ |\mathcal{M}_{+,-,\pm\frac{3}{2},\pm\frac{1}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{3}{2},\mp\frac{1}{2}}|^{2}=|\mathcal{M}_{-,+,\pm\frac{1}{2},\pm\frac{3}{2}}|^{2}=|\mathcal{M}_{+,-,\mp\frac{1}{2},\mp\frac{3}{2}}|^{2}=e^{4}a^{2}\left(1\mp\cos\theta\right)^{2}. \tag{D4} $
When investigating the dependence of the cross-section on the transverse polarization of the
$ \Omega^{-} $ particle, we decompose its polarization in the transverse direction. For the other particles, we sum over their polarizations, allowing for longitudinal polarization decomposition. By utilizing Eq. (16) and considering parity conservation, we derive the following independent terms:$ |\mathcal{M}_{+,-,\pm\frac{3}{2}\uparrow_{x},+\frac{3}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{3}{2}\uparrow_{x},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[\sqrt{3}aE\left(1+\cos\theta\right)\pm m\sin\theta\right]^{2}, \tag{D5} $
$ |\mathcal{M}_{-,+,\pm\frac{3}{2}\uparrow_{x},+\frac{3}{2}}|^{2} =|\mathcal{M}_{+,-,\mp\frac{3}{2}\uparrow_{x},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[\sqrt{3}aE\left(1-\cos\theta\right)\mp m\sin\theta\right]^{2}, \tag{D6} $
$ |\mathcal{M}_{+,-,\pm\frac{3}{2}\uparrow_{x},+\frac{1}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{3}{2}\uparrow_{x},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ am^{2}\left[E\left(1-\cos\theta\right)\pm\sqrt{3}am\sin\theta\right]\right. \left.-\sqrt{3}b^{2}\left(2E^{2}-m^{2}\right)\left[E\left(1+\cos\theta\right)\pm m\sin\theta\right]\right\} ^{2}, \tag{D7} $
$ |\mathcal{M}_{-,+,\pm\frac{3}{2}\uparrow_{x},+\frac{1}{2}}|^{2} =|\mathcal{M}_{+,-,\mp\frac{3}{2}\uparrow_{x},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ am^{2}\left[E\left(1+\cos\theta\right)\mp\sqrt{3}am\sin\theta\right]\right.\left.-\sqrt{3}b^{2}\left(2E^{2}-m^{2}\right)\left[E\left(1-\cos\theta\right)\mp m\sin\theta\right]\right\} ^{2}, \tag{D8} $
$ |\mathcal{M}_{+,-,\pm\frac{1}{2}\uparrow_{x},+\frac{3}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{1}{2}\uparrow_{x},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[aE\left(1+\cos\theta\right)\pm\sqrt{3}m\sin\theta\right]^{2}, \tag{D9}$
$ |\mathcal{M}_{-,+,\pm\frac{1}{2}\uparrow_{x},+\frac{3}{2}}|^{2} =|\mathcal{M}_{+,-,\mp\frac{1}{2}\uparrow_{x},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[aE\left(1-\cos\theta\right)\mp\sqrt{3}m\sin\theta\right]^{2}, \tag{D10} $
$ |\mathcal{M}_{+,-,\pm\frac{1}{2}\uparrow_{x},+\frac{1}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{1}{2}\uparrow_{x},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ am^{2}\left[\sqrt{3}E\left(1-\cos\theta\right)\pm am\sin\theta\right]\right.\left.+b^{2}\left(2E^{2}-m^{2}\right)\left[E\left(1+\cos\theta\right)\mp m\sin\theta\right]\right\} ^{2},\tag{D11} $
$ |\mathcal{M}_{-,+,\pm\frac{1}{2}\uparrow_{x},+\frac{1}{2}}|^{2} =|\mathcal{M}_{+,-,\mp\frac{1}{2}\uparrow_{x},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ am^{2}\left[\sqrt{3}E\left(1+\cos\theta\right)\mp am\sin\theta\right]\right.\left.+b^{2}\left(2E^{2}-m^{2}\right)\left[E\left(1-\cos\theta\right)\pm m\sin\theta\right]\right\} ^{2},\tag{D12} $
$ |\mathcal{M}_{+,-,\pm\frac{3}{2}\uparrow_{y},+\frac{3}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{3}{2}\uparrow_{y},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[3a^{2}E^{2}\left(1+\cos\theta\right)^{2}+m^{2}\sin^{2}\theta\right] \tag{D13} $
$ |\mathcal{M}_{-,+,\pm\frac{3}{2}\uparrow_{y},+\frac{3}{2}}|^{2} =|\mathcal{M}_{+,-,,\mp\frac{3}{2}\uparrow_{y},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[3a^{2}E^{2}\left(1-\cos\theta\right)^{2}+m^{2}\sin^{2}\theta\right] \tag{D14} $
$ \begin{aligned}[b] |\mathcal{M}_{+,-,\pm\frac{3}{2}\uparrow_{y},+\frac{1}{2}}|^{2} =&|\mathcal{M}_{-,+,\mp\frac{3}{2}\uparrow_{y},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ 3m^{2}\left[m^{2}\left(a^{2}+b^{2}\right)-2b^{2}E^{2}\right]^{2}\sin^{2}\theta\right.\\ & \left.+\left[am^{2}E\left(1-\cos\theta\right)+\sqrt{3}b^{2}E\left(2E^{2}-m^{2}\right)\left(1+\cos\theta\right)\right]^{2}\right\} \end{aligned}\tag{D15} $
$ \begin{aligned}[b] |\mathcal{M}_{-,+,\pm\frac{3}{2}\uparrow_{y},+\frac{1}{2}}|^{2} =&|\mathcal{M}_{+,-,\mp\frac{3}{2}\uparrow_{y},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ 3m^{2}\left[m^{2}\left(a^{2}+b^{2}\right)-2b^{2}E^{2}\right]^{2}\sin^{2}\theta\right.\\ & \left.+\left[am^{2}E\left(1+\cos\theta\right)+\sqrt{3}b^{2}E\left(2E^{2}-m^{2}\right)\left(1-\cos\theta\right)\right]^{2}\right\} \end{aligned}\tag{D16} $
$ |\mathcal{M}_{+,-,\pm\frac{1}{2}\uparrow_{y},+\frac{3}{2}}|^{2} =|\mathcal{M}_{-,+,\mp\frac{1}{2}\uparrow_{y},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[a^{2}E^{2}\left(1+\cos\theta\right)^{2}+3m^{2}\sin^{2}\theta\right], \tag{D17} $
$ |\mathcal{M}_{-,+,\pm\frac{1}{2}\uparrow_{y},+\frac{3}{2}}|^{2} =|\mathcal{M}_{+,-,\mp\frac{1}{2}\uparrow_{y},-\frac{3}{2}}|^{2}=\frac{e^{4}}{8E^{2}}\left[a^{2}E^{2}\left(1-\cos\theta\right)^{2}+3m^{2}\sin^{2}\theta\right], \tag{D18} $
$ \begin{aligned}[b] |\mathcal{M}_{+,-,\pm\frac{1}{2}\uparrow_{y},+\frac{1}{2}}|^{2} =&|\mathcal{M}_{-,+,\mp\frac{1}{2}\uparrow_{y},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ m^{2}\left[m^{2}\left(a^{2}+b^{2}\right)-2b^{2}E^{2}\right]^{2}\sin^{2}\theta\right.\\ & \left.+\left[\sqrt{3}am^{2}E\left(1-\cos\theta\right)-b^{2}E\left(2E^{2}-m^{2}\right)\left(1+\cos\theta\right)\right]^{2}\right\} , \end{aligned}\tag{D19} $
$ \begin{aligned}[b] |\mathcal{M}_{-,+,\pm\frac{1}{2}\uparrow_{y},+\frac{1}{2}}|^{2} =&|\mathcal{M}_{+,-,\mp\frac{1}{2}\uparrow_{y},-\frac{1}{2}}|^{2}=\frac{e^{4}}{8E^{2}m^{4}}\left\{ m^{2}\left[m^{2}\left(a^{2}+b^{2}\right)-2b^{2}E^{2}\right]^{2}\sin^{2}\theta\right.\\ & \left.+\left[\sqrt{3}am^{2}E\left(1+\cos\theta\right)-b^{2}E\left(2E^{2}-m^{2}\right)\left(1-\cos\theta\right)\right]^{2}\right\}. \end{aligned}\tag{D20} $
Spin density matrix for ${ \bf\Omega^{-}} $ and its polarization alignment in ${\boldsymbol\psi\bf(3686)\rightarrow \Omega^{-}\bar{\Omega}^{+}}$
- Received Date: 2023-02-12
- Available Online: 2023-09-15
Abstract: We investigate the spin density matrix of