Hyperon spin correlation in high-energy heavy-ion collisions Xin-Li Sheng, Xiang-Yu Wu, Dirk H. Rischke, Xin-Nian Wang Abstract Recent experimental data show an unexpectedly large spin alignment of ϕ \phi mesons in high-energy heavy-ion collisions, which can be explained by short-distance fluctuations of strong force fields (vector ϕ \phi fields) within the constituent-quark model. We calculate the hyperon spin correlations within the same model, taking into account hydrodynamic effects and a ϕ \phi field fluctuating in space-time according to a Gaussian distribution. The Λ ​ Λ ¯ \Lambda\bar{\Lambda} spin correlation induced by the ϕ \phi field is shown to be negative as opposed to that of Λ ​ Λ \Lambda\Lambda or Λ ¯ ​ Λ ¯ \bar{\Lambda}\bar{\Lambda} . We thus propose a new net spin correlation as a sensitive probe to separate strong force effects from hydrodynamic ones. With the strength of the field fluctuations extracted from the observed ϕ \phi spin alignment, we predict the collision energy dependence of the hyperon spin correlations and also investigate the dependence of net spin correlation on azimuthal-angle and rapidity difference. Introduction. The hot and dense matter produced in non-central heavy-ion collisions, known as the quark-gluon plasma (QGP), carries a large amount of orbital angular momentum (OAM) in the form of a fluid velocity field with nonvanishing vorticity. Quarks in the QGP are shown to be globally polarized through the spin-orbital coupling. It was predicted that this leads to a global polarization of the final-state hyperons Liang and Wang 2005a ; Voloshin 2004 ; Becattini et al. 2008 ; Becattini et al. 2013a . Such a phenomenon has been confirmed by the STAR experiment at RHIC Adamczyk et al. 2017 ; Adam et al. 2018 ; Adam et al. 2021 ; Abdallah et al. 2021 as well as by the HADES experiment at lower energies Abou Yassine et al. 2022 , while the global polarization at the LHC energies is close to zero Acharya et al. 2020 as expected. Recently, an interesting azimuthal and rapidity dependence of the hyperon polarization has also been observed Adam et al. 2019 ; Acharya et al. 2022 , which is related to the local structure of the vorticity or the fluid shear in the QGP Becattini et al. 2013b ; Csernai et al. 2013 ; Becattini et al. 2015 ; Jiang et al. 2016 ; Pang et al. 2016 ; Becattini et al. 2021a ; Becattini et al. 2021b ; Liu and Yin 2021 ; Fu et al. 2021a . The OAM in heavy-ion collisions was also predicted to lead to the spin alignment of vector mesons Liang and Wang 2005b , which was indeed observed recently for \phi mesons Abdallah et al. 2023 . However, theoretical predictions for the spin alignment induced by conventional mechanisms such as the fluid vorticity or the magnetic field Yang et al. 2018 ; Sheng et al. 2020a ; Sheng et al. 2024a ; Wei and Huang 2023 are almost two orders of magnitude smaller than what was observed in the experiments. To account for the disparity, several new mechanisms have been proposed Sheng et al. 2020a ; Xia et al. 2021 ; Müller and Yang 2022 ; Sheng et al. 2023a ; Sheng et al. 2024b ; Li and Liu 2022 ; Wagner et al. 2023 ; Kumar et al. 2023 ; Sheng et al. 2023b ; Dong et al. 2024 ; Sheng et al. 2024c ; Yang 2025 ; Chen et al. 2023 ; Chen et al. 2025 . Among them is the local strong force field (SFF) Sheng et al. 2020a ; Sheng et al. 2023a ; Sheng et al. 2024b ; Sheng et al. 2023b , which can polarize quarks through its coupling to the quark spin. This field is dominated by fluctuations and thus its mean value over a large space and time scale should be zero. As a consequence, it does not contribute to the global polarization of quarks and final-state hyperons. However, since the vector meson’s spin alignment is related to the spin correlation between the constituent quark and anti-quark Liang and Wang 2005b ; Yang et al. 2018 , the fluctuating SFF can cause the spin alignment if its field correlation is nonzero over the size of the vector meson. This model is shown to describe the transverse-momentum dependence of the observed \phi meson spin alignment and predicts the azimuthal-angle or rapidity dependence which can be tested in future experiments Sheng et al. 2023a ; Sheng et al. 2023b . Within the constituent-quark model Fayyazuddin and Riazuddin 2011 ; Yang et al. 2018 , the spin of \Lambda ( \bar{\Lambda} ) is carried by the strange (anti-strange) quark. In analogy to the vector-meson spin alignment, one can also expect a spin correlation of hyperons and anti-hyperons that are produced in close proximity on the freeze-out hypersurface due to the fluctuating SFF, if the correlation length of the SFF extends over the size of a hadron. In this Letter, we compute spin correlations among \Lambda ’s and \bar{\Lambda} ’s at the freeze-out hypersurface in a (3+1)-D viscous hydrodynamical model, taking into account both the SFF and various hydrodynamic effects such as thermal vorticity, thermal shear, and the spin Hall effect. We demonstrate that the difference between “same-sign” pairs ( \Lambda\Lambda or \bar{\Lambda}\bar{\Lambda} ) and “opposite-sign” pairs ( \Lambda\bar{\Lambda} ) is a sensitive observable to distinguish the SFF-induced polarization from the conventional hydrodynamic contributions. Our predicted spin correlations at RHIC-BES energies are significant and their measurement will shed light on the mechanisms of spin polarization and the nature of the SFF at high temperature. Spin correlation. For a relativistic spin-1/2 particle system, the spin state can be described by a polarization vector P^{\mu}(x,p) , which is a four-component function of phase-space coordinates. For a two-particle system composed of particle 1 and 2 , a straightforward definition of the spin correlation tensor is C_{12}^{\mu\nu}(p_{1},p_{2})\equiv\left\langle P_{1}^{\mu}(x_{1},p_{1})P_{2}^{\nu}(x_{2},p_{2})\right\rangle\,, (1) where P_{i}^{\mu} denotes the polarization vector for particle i with momentum p_{i}^{\mu} at space-time point x_{i}^{\mu} . Here, \left\langle\cdots\right\rangle includes the event-by-event average as well as the average over space-time points x_{1}^{\mu} and x_{2}^{\mu} weighted by particle densities. Since P_{i}^{\mu} is orthogonal to p_{i}^{\mu} , it can be expressed in terms of the 3-components of the spin polarization in the particle’s rest frame (\mathcal{P}_{i,x}^{\text{rest}},\mathcal{P}_{i,y}^{\text{rest}},\mathcal{P}_{i,z}^{\text{rest}}) . One then finds that C_{12}^{\mu\nu} contains only 9 independent degrees of freedom, related to the following 3\times 3 matrix c_{12}^{ab}(p_{1},p_{2})\equiv n_{1,a}^{\mu}(p_{1})n_{2,b}^{\nu}(p_{2})C_{\mu\nu}^{12}(p_{1},p_{2}),\ \ a,b=x,y,z, (2) where n_{i,a}^{\mu}(p_{i})\equiv\Lambda_{\ \nu}^{\mu}(p_{i})\left(0,\,{\bf e}_{a}\right)^{\nu} with {\bf e}_{a},\ a=x,y,z denoting the unit 3-vector in the direction of the a -axis and \Lambda_{\ \nu}^{\mu}(p_{i}) being the transformation matrix for the boost from the rest frame of the i -th particle to the laboratory frame. It is straightforward to show that c_{12}^{ab}=\left\langle\mathcal{P}_{1,a}^{\text{rest}}\mathcal{P}_{2,b}^{\text{rest}}\right\rangle is the correlation between the polarizations in the respective rest frames of particles 1 and 2. In experiments, c_{12}^{ab} can be extracted from the polar-angle distributions of the decay products if particles 1 and 2 undergo weak decays Shen et al. 2024 , e.g., \Lambda hyperon decays. In this work, we concentrate on \Lambda\Lambda , \Lambda\bar{\Lambda} , and \bar{\Lambda}\bar{\Lambda} pairs, since the multiplicity of \Lambda ( \bar{\Lambda} ) is much larger than other hyperon species. Recently the STAR collaboration has observed a significant \Lambda\bar{\Lambda} spin correlation in proton-proton collisions ( STAR 2025 ) . This correlation is dominated by the hadronization of entangled s\bar{s} pairs produced via virtual gluon splitting, i.e., g\rightarrow s+\bar{s}\rightarrow\Lambda+\bar{\Lambda}+X . Due to helicity conservation, the spin orientations of s and \bar{s} are enforced to align in the same direction. This maximizes the spin correlation of s\bar{s} , which is in turn transferred to the final-state \Lambda\bar{\Lambda} spin correlation ( STAR 2025 ; Ellis et al. 1996 ; Ellis and Hwang 2012 ; Wei and Huang 2023 ; Lin et al. 2025 ) . By contrast, in nucleus-nucleus collisions, the \Lambda\bar{\Lambda} pairs are from the hadronization of QGP, while their constituent quarks originate from different processes and thus are unlikely to be entangled. The spin correlation arises only if both hyperons are polarized by the same external field in the surrounding hot medium. Such a kind of correlation corresponds to the “induced spin correlation” in Refs. Lv et al. 2024 ; Zhang et al. 2024 . Quark and hyperon spin polarization. According to the non-relativistic coalescence model Fayyazuddin and Riazuddin 2011 ; Yang et al. 2018 ; Sheng et al. 2020b , the polarization of a \Lambda baryon is fully carried by its constituent s quark. We generalize such a relation to the relativistic case by setting P_{\Lambda}^{\mu}(x,p)\approx P_{s}^{\mu}(x,R_{s}p)\,, (3) where P_{\Lambda}^{\mu} and P_{s}^{\mu} are polarizations for a \Lambda hyperon and an s quark, respectively, and R_{s}\equiv m_{s}/m_{\Lambda} is the ratio between their invariant masses. Here we have neglected the size of the \Lambda and the motion of the s quark relative to the \Lambda hyperon, indicating that the \Lambda hyperon and the constituent s quark have the same position and velocity. The polarization of \bar{\Lambda} is related to that of \bar{s} in the same way. The relation Eq. ( 3 ) has been previously used in Ref. Fu et al. 2021a as the “strange-memory scenario” to study the local polarization of \Lambda . For a system in local thermodynamical equilibrium, the spin polarization of the s ( \bar{s} ) quark is in general given by Yi et al. 2021 P^{\mu}_{s/\bar{s}}(x,p)=P^{\mu}_{\omega}+P^{\mu}_{\text{shear}}+P^{\mu}_{\text{SHE}}+P^{\mu}_{\phi}\,, (4) where the first three terms on the right-hand-side denote polarizations induced by the thermal vorticity, the thermal shear tensor, and the spin Hall effect, respectively. The last term in Eq. ( 4 ) is the contribution of the SFF, P^{\mu}_{\phi}=\pm\frac{g_{\phi}}{4mE_{p}T}\epsilon^{\mu\nu\rho\sigma}F^{\phi}_{\rho\sigma}p_{\nu}(1-n_{\text{FD}}^{\pm})\,, (5) where F^{\phi}_{\mu\nu} is the \phi field tensor, g_{\phi} is the coupling between \phi field and s quark, and n_{\text{FD}}^{\pm}\equiv 1/[1+\exp(\beta\cdot p\mp\mu_{s}/T)] is the Fermi-Dirac distribution with \mu_{s} being the s quark chemical potential and \beta^{\mu}\equiv u^{\mu}/T being the thermal velocity. Classical electromagnetic fields can also polarize quarks, which are neglected considering that they are sufficiently small at the hadronization stage McLerran and Skokov 2014 ; Tuchin 2013 . The quantities u^{\mu} , \mu_{s} , T , and F^{\phi}_{\mu\nu} depend on the space-time position x^{\mu} and thus the total polarization in Eq. ( 4 ) is locally defined. Assuming that the QGP in heavy-ion collisions undergoes a hydrodynamic evolution, the space-time-averaged spin polarization is calculated via the Cooper-Frye formula Cooper and Frye 1974 ; Becattini et al. 2013a ; Becattini et al. 2024 . For \Lambda ( \bar{\Lambda} ) hyperons with four-momentum p^{\mu} , the latter reads \left\langle P^{\mu}_{\Lambda/\bar{\Lambda}}\right\rangle(p)=\sum_{\text{events}}\frac{\int d\Sigma(x)\cdot p\,P^{\mu}_{\Lambda/\bar{\Lambda}}(x,p)f_{\Lambda/\bar{\Lambda}}(x,p)}{N_{\text{event}}\int d\Sigma(x)\cdot p\,f_{\Lambda/\bar{\Lambda}}(x,p)}\,, (6) where d\Sigma_{\mu}(x) is the differential volume vector at point x on the freezeout hypersurface and P^{\mu}_{\Lambda/\bar{\Lambda}} is related to the quark polarization [Eq. ( 4 )] through Eq. ( 3 ). If we also implement the local-equilibrium condition for \Lambda ( \bar{\Lambda} ) hyperons, the distribution f_{\Lambda/\bar{\Lambda}} should be the Fermi-Dirac distribution f_{\Lambda/\bar{\Lambda}}(x,p)=\{1+\exp[\beta(x)\cdot p\mp\mu_{B}(x)/T(x)]\}^{-1}\,, (7) where \mu_{B} is the baryon chemical potential. Analogously, the spin correlation in Eq. ( 1 ) is calculated via \displaystyle C^{\mu\nu}_{12}(p_{1},p_{2}) \displaystyle= \displaystyle\frac{1}{N_{\text{event}}}\sum_{\text{event}} (8) \displaystyle\hskip-22.76228pt\times\frac{\int d\Sigma(x_{1})\cdot p_{1}\int d\Sigma(x_{2})\cdot p_{2}\,P^{\mu}_{1}\,P^{\nu}_{2}\,f_{1}\,f_{2}\,}{\left[\int d\Sigma(x_{1})\cdot p_{1}\,f_{1}\right]\left[\int d\Sigma(x_{2})\cdot p_{2}\,f_{2}\right]}\,, where P_{i}^{\mu},\,f_{i},\,i=1,2 are functions of x_{i}^{\mu} and p_{i}^{\mu} , and particles 1 , 2 denote \Lambda or \bar{\Lambda} hyperons. Numerical setup. In this work, we employ the CLVisc hydrodynamic model Pang et al. 2018 ; Wu et al. 2022a ; Wu et al. 2022b to generate the hadronization hypersurface and the corresponding hydrodynamic fields. This hypersurface is determined by the local energy density reaching the freeze-out condition, \varepsilon_{\text{frz}}=0.4\text{ GeV}/\text{fm}^{3} . Here, we use an event-averaged smooth initial profile, which is generated by averaging over 5000 fluctuating SMASH initial events in a given centrality bin. This initial condition can capture the main features of the \Lambda polarization, as shown in Refs. Wu et al. 2022b ; Fu et al. 2021a ; Fu et al. 2021b ; Becattini et al. 2021b . High-statistics and more detailed event-by-event simulations are left for future studies. To calculate the spin correlation induced by hydrodynamic effects, i.e., P^{\mu}_{\omega} , P^{\mu}_{\text{shear}} , and P^{\mu}_{\text{SHE}} , we first extract space-time-dependent values of \varpi_{\mu\nu} , \xi_{\mu\nu} , T , and \mu_{s} from numerical simulations to obtain the spin polarization of s ( \bar{s} ). Then, the spin correlation is evaluated using Eqs. ( 1 ), ( 2 ), ( 3 ), and ( 8 ). Furthermore, we assume that fluctuations of the vector \phi field have very weak correlations to hydrodynamic quantities. This implies that we can neglect the correlation between the polarization induced by hydrodynamic effects and that induced by the \phi field. We further assume that the \phi field tensor F^{\phi}_{\mu\nu} fluctuates in space-time while the field-field correlation is parameterized as \displaystyle g_{\phi}^{2}F_{ij}^{\phi}(x_{1})F_{ij}^{\phi}(x_{2})/[T(x_{1})T(x_{2})] \displaystyle=\begin{cases}F_{T}^{2}G(x_{1}-x_{2}),&(i,j)=(0,1),\,(0,2),\,(2,3),\,(3,1)\,,\\ F_{z}^{2}G(x_{1}-x_{2}),&(i,j)=(0,3),\,(1,2)\,,\end{cases}\, (9) while other unlisted components are zero. The long-distance correlation is suppressed by a Gaussian function G(x_{1}-x_{2})\equiv\text{exp}\left[-\frac{(t_{1}-t_{2})^{2}}{\sigma_{t}^{2}}-\frac{({\bf x}_{1}-{\bf x}_{2})^{2}}{\sigma_{x}^{2}}\right]\,, (10) where \sigma_{t} and \sigma_{x} are typical temporal and spatial scales of the field correlation. In our calculation we take \sigma_{t}=\sigma_{x}=0.6 fm, which is comparable to the radius of light mesons. The parameters F_{T}^{2} and F_{z}^{2} are local fluctuations of transverse and longitudinal fields, which are extracted in Ref. Sheng et al. 2023a by fitting the \phi meson spin-alignment data from the STAR experiment. They can be effectively parametrized by the following analytical form Sheng et al. 2023a as a function of the colliding energy \sqrt{s_{\text{NN}}} , \ln(F_{T,z}^{2}/m_{\pi}^{2})=a_{T,z}-b_{T,z}\ln(\sqrt{s_{\text{NN}}}/\text{GeV})\,, (11) with m_{\pi}=0.138 GeV, a_{T}=3.90\pm 1.11 , b_{T}=0.924\pm 0.234 , a_{z}=3.33\pm 0.917 , and b_{z}=0.760\pm 0.189 . Here the intrinsic anisotropy of the QGP induced by a faster longitudinal expansion relative to the transverse expansion is taken into account, while the difference between the in-plane direction and the out-of-plane direction is neglected. Since the \phi meson spin alignment is more strongly influenced by the relative motion of the meson to the background than by the transverse anisotropy of the QGP, as pointed out in Ref. Sheng et al. 2023b , this setup should capture the dominant behavior of the \Lambda spin correlation. Figure 1: The collision energy dependence of the \Lambda\Lambda spin correlation, compared to experimental data Abdallah et al. 2023 (red stars with error bars) for the \phi meson global spin alignment \delta\rho_{00}\equiv\rho_{00}-1/3 . The xx , yy , and zz components of c^{ab}_{\Lambda\Lambda} are denoted by the orange long-dashed line, the red short-dashed line, and the cyan solid line, respectively. The global spin alignment is rescaled by a factor \alpha=0.01 such that it has the same order of magnitude as the spin correlation. Numerical results. We present in Fig. 1 the diagonal components of the \Lambda\Lambda spin correlation ( 2 ) as functions of the colliding energy. The calculations are carried out for Au+Au collisions at \sqrt{s_{\text{NN}}}=7.7 , 14.5 , 27 , 39 , and 62.4 GeV, within the centrality class of 20-50\% , and are averaged over hyperons at central rapidity |Y|<1 with transverse momentum p_{T}\in(0.5,\,3) GeV/ c . The spin correlation is stronger at lower energies, showing the same behavior as the \phi meson spin alignment Abdallah et al. 2023 . For the xx and zz components, the correlation is governed by SFF fluctuations, which are parameterized as decreasing functions of the collision energy in order to fit the experimental data for \rho_{00} of the \phi meson. Furthermore, the SFF has nearly the same contribution to c_{\Lambda\Lambda}^{xx} and c_{\Lambda\Lambda}^{yy} . The splitting between them in Fig. 1 is mainly a consequence of the global spin polarization of quark and anti-quark induced by the vorticity field. The zz component of the spin correlation appears to be smaller than the transverse components, reflecting the asymmetry between the fireball’s longitudinal and transverse expansion. Figure 2: The spin correlation component c_{12}^{yy} as a function of the azimuthal-angle difference between particles 1 and 2 . Correlations in the presence of SFF for \Lambda\Lambda , \Lambda\bar{\Lambda} , and \bar{\Lambda}\bar{\Lambda} are denoted by orange long-dashed, red short-dashed, and cyan solid lines, respectively. The contribution from hydrodynamic fields to c_{\Lambda\bar{\Lambda}}^{yy} is shown by the green dash-dotted line, while c_{\Lambda\Lambda}^{yy} and c_{\bar{\Lambda}\bar{\Lambda}}^{yy} in absence of SFF are nearly identical with c_{\Lambda\bar{\Lambda}}^{yy} and thus are not shown in this figure. Figure 2 displays the yy component of the spin correlation for \Lambda\Lambda , \Lambda\bar{\Lambda} , and \bar{\Lambda}\bar{\Lambda} . The absolute values of the spin correlations reach their maximum at \delta\phi=0 and decrease towards zero as \delta\phi\rightarrow\pi . This trend suggests that two hyperons are more (less) correlated if they move in the same (opposite) direction. A slight difference is observed between the spin correlation of \Lambda\Lambda and that of \bar{\Lambda}\bar{\Lambda} as a consequence of different space-time distributions of \Lambda and \bar{\Lambda} when the baryon chemical potential is nonzero. Even in the absence of SFF fluctuations (shown as the green dash-dotted line), there still exist nonzero spin correlations, which are mainly attributed to the global rotation of the QGP. Differences between the green dash-dotted line and the other lines arise due to the SFF fluctuation. When \delta\phi=0 , the contribution from hydrodynamic fields is much smaller than that from the SFF fluctuation, indicating that spin correlations between hyperons are also dominated by SFF, much like the vector-meson spin alignment. When \delta\phi=\pi , the hydrodynamic fields, especially the vorticity field, become the dominant source for the spin correlation, while the magnitude is approximately equal to the product of \Lambda ’s and \bar{\Lambda} ’s global polarizations, P_{\Lambda}P_{\bar{\Lambda}}\sim\mathcal{O}(10^{-4}) . A key unique feature of the hyperon spin correlation caused by the SFF is that the spin correlation of \Lambda\Lambda and \bar{\Lambda}\bar{\Lambda} has the opposite sign as the \Lambda\bar{\Lambda} spin correlation, as shown in Fig. 2 . In order to separate the SFF contribution from others, we introduce the following net spin correlation, c_{\text{net}}^{ab}\equiv\frac{1}{2}(c^{ab}_{\Lambda\Lambda}+c^{ab}_{\bar{\Lambda}\bar{\Lambda}})-c^{ab}_{\Lambda\bar{\Lambda}}\,, (12) which denotes the difference between the “same-sign” pairs ( \Lambda\Lambda or \bar{\Lambda}\bar{\Lambda} ) and the “opposite-sign” pairs ( \Lambda\bar{\Lambda} ). We show in Figs. 3 and 4 its dependence on the azimuthal-angle difference and the rapidity difference between the two hyperons. When the SFF is taken into account, the xx , yy , and zz components of c_{\text{net}} peak at \delta\phi=0 or \delta Y=0 and approach zero when \delta\phi or \delta Y increase. In the absence of SFF, c_{\text{net}} becomes 2-3 orders of magnitude smaller than in the case with SFF, shown in the insets in Figs. 3 and 4 . This behavior arises because the SFF has an opposite effect on the polarizations of \Lambda and \bar{\Lambda} , as described by Eq. ( 5 ), leading to opposite contributions to the spin correlations of \Lambda\Lambda and \Lambda\bar{\Lambda} . In contrast, the polarizations induced by the thermal vorticity and the thermal shear tensor are identical for \Lambda and \bar{\Lambda} , resulting in nearly the same values of c_{\Lambda\Lambda}^{yy} , c_{\bar{\Lambda}\bar{\Lambda}}^{yy} , and c_{\Lambda\bar{\Lambda}}^{yy} when the SFF is switched off. The spin Hall effect could also result in a nonzero c_{\text{net}} , but its magnitude is much smaller than the contribution of SFF. Therefore c_{\text{net}} is shown to be sensitive to the SFF and is almost independent from the other hydrodynamic fields. Figure 3: Components of the net spin correlation c^{ab}_{\text{net}} as functions of the azimuthal-angle difference. Results with the SFF are shown by the orange, red, and cyan lines, while the other lines are results in absence of the SFF. Figure 4: Components of the net spin correlation c^{ab}_{\text{net}} as functions of the rapidity difference. Notations are the same as in Fig. 2 . The dependence of the net spin correlation on \delta\phi or \delta Y is a consequence of the position-momentum correlation of hyperons. Due to the transverse expansion of the QGP, the mean transverse position of hyperons emitted from the freeze-out hypersurface is approximately aligned with the transverse momentum, \left\langle{\bf x}_{T}\right\rangle\propto\hat{\bf p}_{T} . Similarly, the longitudinal expansion leads to a nearly linear dependence of \left\langle z\right\rangle on the rapidity. Given such significant position-momentum correlations, a larger separation of two hyperons in momentum space corresponds to a larger distance between their production positions, which leads to weaker field-field and spin-spin correlations according to Eqs. ( 8 ) and ( 9 ). The \delta\phi or \delta Y dependence is similar to the \Lambda\bar{\Lambda} spin correlation observed in p+p collisions STAR 2025 , but arises from a different mechanism. In p+p collisions (or for high-momentum hyperons in A+A collisions Li et al. 2024 ; Huang et al. 2025 ), it is dominated by the parton fragmentation, while the correlation in this study is dominated by interactions between constituent quarks and the fluctuating SFF. Summary and Discussion. Based on the constituent-quark coalescence model, we have investigated the spin correlation between \Lambda ( \bar{\Lambda} ) hyperons using the CLVisc hydrodynamic model with SMASH initial conditions. We considered the spin polarization due to both the hydrodynamic fluid evolution and the SFF with a short-distance Gaussian correlation. Due to opposite contributions of the SFF to the \Lambda and \bar{\Lambda} polarization, the spin correlation induced by the SFF fluctuation is negative for \Lambda\bar{\Lambda} while it is positive for \Lambda\Lambda and \bar{\Lambda}\bar{\Lambda} . We thus proposed the net spin correlation, which describes the difference between “same-sign” pairs and “opposite-sign” pairs, as a unique signature of spin polarization induced by SFF in heavy-ion collisions. Given the SFF fluctuation constrained by the experimental data on the \phi meson spin alignment, we have provided quantitative predictions on the relative momentum dependence ( \delta\phi and \delta Y ) of the net spin correlations of hyperons, which all decrease with their separation in momentum space. We also predicted the dependence of different spin correlation Components on the collision energy, which shows a pattern similar to the \phi meson spin alignment. Provided that the typical space-time scale of the SFF fluctuation is comparable to the size of a light meson, \sigma_{t}=\sigma_{x}=0.6 fm, the net spin correlation is \mathcal{O}(10^{-4}) . Larger values of \sigma_{t} and \sigma_{x} result in stronger spin correlations. For example, if \sigma_{t}=\sigma_{x}=1.2 fm, the correlation induced by SFF will become twice as large. These predictions provide a perfect opportunity to test the mechanism of spin polarization in the QGP and constraints on the scale of the SFF fluctuation in high-energy heavy-ion collisions. For simplicity, we have neglected the internal motion of s quark inside the hyperon in the constituent-quark coalescence model. Additionally, the classical electromagnetic fields are not considered, as they should be weak at the freeze-out stage of heavy-ion collisions. These effects, along with further refinements, can be addressed in future studies. Acknowledgments. The authors thank Diyu Shen and Aihong Tang for enlightening discussions. This work is supported in part by the Italian Ministry of University and Research, project PRIN2022“Advanced probes of the Quark Gluon Plasma”, Next Generation EU, Mission 4 Component 1, X.-Y.W. is supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC) [SAPIN-2020-00048 and SAPIN-2024-00026] and in part by US National Science Foundation (NSF) under grant number OAC2004571. D.H.R. is supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through CRC-TR 211 “Strong-interaction matter under extreme conditions” - project number 315477589 - TRR 211. X.-N.W. is supported by NSFC under Grant No. 1193507, the Guangdong MPBAR with No. 2020B0301030008 and by the Alexander von Humboldt Foundation through a Humboldt Research Award.