Scalarization and descalarization in hyperbolic encounters of black holes Frederick C.L. Pardoe, Helvi Witek Abstract We use numerical relativity to study the scalar field evolution sourced by hyperbolic encounters of black holes in quadratic scalar Gauss-Bonnet gravity. In this theory, single black holes are known to acquire a scalar hair through scalarization for certain values of their mass and spin. We work in the decoupling limit and evolve the scalar field on top of a background metric. Seeding binary black holes with an initial scalar field, we find that configurations which initially cannot sustain a scalar hair temporarily scalarize during an encounter and thereby exhibit dynamical scalarization . This is possible for both positive and negative couplings between the scalar field and curvature in black hole binaries with zero and non-zero initial spins, respectively. Furthermore, we find that the change in the spin magnitude of black holes during certain hyperbolic encounters can lead to permanent spin-induced scalarization (or descalarization ), which we refer to as spin-up (de)scalarization . I Introduction Ever since the publication of General Relativity (GR) more than a century ago, the theory has been subject to a litany of tests. Historically, most tests were conducted in the weak-field regime, and only a few strong-field tests based on electromagnetic observations of pulsars were possible; see Refs. 138 ; 30 ; 139 ; 144 for a review. However, since the first direct detection of gravitational waves (GWs) in 2015 6 , tests of GR in the strong-field regime have become far more common. In the decade since, hundreds of GW observations have been subject to tests 7 ; 8 ; 9 ; 10 ; 11 ; 2 ; 1 ; 3 ; 4 ; 5 . Further strong-field tests have also been conducted using observations of supermassive black holes (BHs) from the Event Horizon Telescope 86 ; 12 . To date, these tests have found observations to be consistent with GR. However, with pulsar timing arrays 144 ; 93 and next generation GW detectors such as LISA 14 , the Einstein Telescope 108 , and Cosmic Explorer 110 , opportunities to test GR will only continue improve in the coming decades. With these advancements in experimental capabilities has come an increased interest in theoretical alternatives to GR. Numerous alternatives to GR have been proposed, motivated in part by the seeming inevitability that quantum effects must lead to deviations from GR in the strong-field regime 119 ; 30 ; 48 . One such set of theories is scalar Gauss-Bonnet (sGB) gravity, a subset of the broader class of Horndeski theories 75 that are known to lead to second order field equations. The Horndeski theories are derived by adding a dynamical scalar field to Lovelock gravity 95 ; 97 ; 96 , either heuristically or via Kaluza-Klein reduction from higher dimensions; see Refs. 44 ; 62 ; 84 ; 85 for a review. Certain forms of sGB gravity also appear as part of a dimensional reduction of heterotic string theory 40 ; 98 ; 81 in the low energy limit. One of the key properties of sGB gravity is that BHs can be endowed with a scalar hair sourced by the spacetime curvature 80 ; 128 ; 129 . This is due to a nonminimal coupling between the curvature and the scalar field. In some versions of the theory, such as those with a linear or diatonic coupling between the scalar and curvature, this hair is always present. However, for quadratic or gaussian forms of the coupling, hairy solutions can only appear below a certain BH mass, or equivalently, above a certain value of the coupling constant 57 ; 126 . In this case, initially hairless BHs can acquire a scalar hair. This phenomenon, known as scalarization 52 ; 53 , has been studied in great detail both in sGB gravity, as well as in other theories; see Ref. 55 for a review. In practice, scalarization is often classified into a number of phenomenological subcategories. Single compact objects can experience spontaneous scalarization , which occurs on account of the object’s compactness 52 ; 53 ; 57 ; 126 . Under certain conditions, BHs in sGB gravity can also undergo spin-induced scalarization , which occurs above a certain (dimensionless) spin determined by the coupling constant and BH mass 54 ; 72 ; 74 . Binary systems can be subject to induced scalarization , where one scalarized compact object induces a scalar hair in another, and dynamical scalarization , where the gravitational interaction of two compact objects causes the scalarization of both 22 ; 123 ; 105 . Equivalent processes in which the scalar hair is lost are referred to as descalarization 55 . In recent years, numerous works have studied binary BH dynamics in sGB gravity 127 ; 61 ; 56 ; 59 ; 17 ; 101 ; 91 ; 79 . These studies have relied on numerical relativity, both in the decoupling limit 140 ; 127 ; 61 ; 56 ; 92 and with backreaction 141 ; 60 ; 59 ; 17 ; 101 ; 91 ; 90 ; 50 , as well as on semi-analytic methods 79 ; 78 ; 124 . See Ref. 113 for a review on the application of numerical relativity to various Horndeski theories. Simulations have shown that remnants formed from the merger of two scalarized BHs can descalarize if the remnant BH has a larger mass 127 or lower spin 61 than its progenitors. These processes are often referred to as dynamical descalarization and spin-induced dynamical descalarization , respectively. However, it is pertinent to note that the descalarization found in these cases occurs after the formation of a common apparent horizon (CAH) and is thus not dynamical in quite the same sense as defined above; originally, dynamical scalarization was coined to describe the scalarization of inspiraling neutron stars prior to merger in other theories 22 ; 123 ; 105 . The “dynamical” descalarization of remnant BHs has also been studied in varieties of sGB gravity where there is a “jump” between solutions with a minimum allowed scalar hair and no scalar hair at all, as opposed to a gradual transition depending on the BH mass (or coupling constant) 56 . It is conjectured that this jump may create an abrupt qualitative difference between GW emissions from systems with scalarized and unscalarized remnants. In addition to spin-induced “dynamical” descalarization , there are also examples of systems where the inspiral and merger of initially unscalarized BHs leads to the spin-induced dynamical scalarization of the remnant 61 . Here, scalarization begins prior to the formation of a CAH and is thus dynamical in the original sense of the term. Nonetheless, for all but exceptionally large couplings, the scalarization in these systems was found to be small prior to the formation of a CAH. Thus, the scalarization would likely affect merger and ringdown, but have minimal impact on the dynamics and GW emission during the inspiral. This phenomena is known as stealth scalarization . In addition to the inferences drawn from studies in the decoupling limit, GW emissions have also been investigated directly with backreaction. The GW emissions emitted from stealth scalarization were directly computed by Ref. 17 , which confirmed that scalarization had minimal effect on these systems prior to merger. Backreaction has also been used to study the merger of initially scalarized BHs and their GW emissions 59 . It was found that the GW luminosity is substantially greater than in GR when non-spinning BHs are considered, but that there is negligible difference when the BHs are scalarized due to spin (i.e., due to spin-induced scalarization ). Using semi-analytic methods, it has been shown that non-spinning BHs should also undergo dynamical scalarization during the inspiral phase of a binary BH merger 79 . This is intriguing, given that such systems should descalarize if they form a CAH. This phenomenon was further explored numerically through the construction of constraint satisfying initial data for binary BHs in a quasicircular configuration 101 ; 91 . It was found that the BHs had a greater value of the scalar field on their horizons than would have been the case if the BHs were isolated and that such binaries could acquire a scalar hair for lower values of the coupling than would have been possible were they isolated. In addition to the above, scalarization has been studied in BH binaries with unequal masses and scalar hairs of opposite sign 101 . In unequal mass binaries, it was found that the larger BH could support a scalar hair despite the fact that it would be unable to do so individually, in effect demonstrating induced scalarization . It was also found that the scalar hair is smaller in systems with opposite sign than in systems with equal sign. Furthermore, a study considering oppositely signed BHs in a backreacted inspiral found a sign flip in one of the BHs leading to the introduction of eccentricity in the orbit 91 . The studies above largely considered BH binaries in head-on collisions or quasicircular inspirals. However, BH binaries are known to display a variety of additional morphologies 107 ; 122 ; 131 ; 71 ; 68 ; 69 . Namely, hyperbolic encounters between BHs can result in scattering or zoom-whirls, where the BHs perform a series of small close orbits punctuated by larger more eccentric orbits before eventually merging. 1 1 1 Neutron stars also exhibit these morphologies; see Refs. 64 ; 103 ; 67 ; 58 ; 109 ; 106 ; 45 . See Refs. 105 ; 133 ; 29 for eccentric orbits and scattering in other theories beyond GR. These temporary close encounters between BHs provide a fascinating environment in which to explore the dynamical processes discussed above. Furthermore, it is known that scattering BHs can “spin up” in the direction of their orbital angular momentum 130 ; 102 ; 77 ; 116 ; 47 ; 88 . This makes such encounters interesting environments in which to consider spin-induced phenomena. In this work, we study the scalarization and descalarization of equal-mass binary BHs as they undergo hyperbolic encounters in quadratic sGB gravity. Working in the decoupling limit, we consider a variety of systems with different couplings, spins, and morphologies. We observe regular and spin-induced dynamical scalarization and descalarization of the BHs. Furthermore, we observe spin-induced scalarization and descalarization resulting from the spin-up of the BHs. We call these phenomenon spin-up scalarization and spin-up descalarization . We structure this work as follows. In Sec. II , we introduce the theory behind sGB gravity and discuss more precisely the conditions for the spontaneous scalarization and spin-induced scalarization of individual BHs. In Sec. III , we discuss the computational framework of our study. In Sec. IV , we discuss the physical setup of our simulations and summarize the simulation suite considered. The results are presented in Sec. V . In Sec. VI , we discuss our conclusions and directions for further work. In the Appendix, we describe a suite of convergence tests. Throughout the following, we use geometric units where G=c=1 . II Scalar Gauss-Bonnet Gravity II.1 Action and equations of motion In sGB gravity, a real (dynamical) scalar field is non-minimally coupled to the spacetime curvature according to the action, S=\frac{1}{16\pi}\int d^{4}x\sqrt{-g}\left[R-\frac{1}{2}\left(\nabla\phi\right)^{2}+\frac{\alpha_{\rm GB}}{4}f(\phi)\mathcal{G}\right]\,, (1) where g is the determinant of the spacetime metric, g_{\mu\nu} , R is the Ricci scalar associated with the metric, \phi is the scalar field, and \alpha_{\rm GB} is a (dimensionful) coupling constant. Several theories fall under the umbrella of sGB gravity; the precise version is set by the coupling function f(\phi) , which couples the scalar field to the spacetime curvature via the Gauss-Bonnet invariant, \mathcal{G}=R^{2}-4R_{\mu\nu}R^{\mu\nu}+R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}\,, (2) where R_{\mu\nu} is the Ricci tensor and R_{\mu\nu\rho\sigma} is the Riemann tensor. The Gauss-Bonnet invariant is a topological invariant and total derivative in four dimensions; it thus only results in modification to the Einstein field equations when coupled to a dynamical scalar field. We work in the test field approximation, where the scalar field does not backreact onto the spacetime, which evolves according to GR. The equations of motion are, \displaystyle\Box\phi=-\frac{\alpha_{\rm GB}}{4}\frac{\textrm{d}f(\phi)}{\textrm{d}\phi}\mathcal{G}\,, (3a) \displaystyle R_{\mu\nu}-\frac{1}{2}Rg_{\mu\nu}=0\,, (3b) where the first equation is given by the extremization of Eq. ( 1 ) with respect to the scalar field, and the second corresponds to the Einstein field equations in vacuum. If the coupling function contains a stationary point, \frac{df(\phi)}{d\phi}|_{\phi=\Psi}=0 , at some value of the scalar field, \phi=\Psi , then Eq. ( 3a ) is trivially solved by the constant scalar field \phi(x^{\mu})=\Psi=\text{const} . In this case, even the backreacted equations of motion can admit any metric solution of GR. However, depending on the parameters of the theory and the geometry of a given system, the GR solution may not be stable. This can lead to scalarization, whereby systems initially described by GR develop a non-trivial scalar field. In this model both the solutions of vacuum GR and scalarized solutions are thus allowed. When expanded around the GR solution such that \phi=\Psi+\delta\phi , any coupling function with a stationary point takes the form f(\phi)=f(\Psi)+f^{\prime\prime}(\Psi)(\delta\phi-\Psi)^{2}/2+\mathcal{O}(\delta\phi^{3}) . Due to the topological invariance of the Gauss-Bonnet invariant, the field and coupling constant can always be redefined such that \Psi=0 and f(\phi)=\phi^{2}+\mathcal{O}(\phi^{3}) without loss of generality. The second order term determines whether a GR solution is stable and thus the onset of scalarization. Therefore, any theory with the same coupling constant scalarizes under the same conditions, regardless of the particular coupling function used. However, the end state of scalarization is determined by higher order terms in the coupling function and thus does depend on the exact theory 57 ; 126 ; 55 ; 125 . For example, systems with coupling functions of the form f(\phi)\propto\phi^{2} 57 , f(\phi)\propto 1-e^{-\eta\phi^{2}} 126 , and f(\phi)\propto\phi^{2}+\zeta\phi^{4} 125 (where \eta and \zeta are constants) scalarize under the same circumstances, but have different end states. In the following, we are primarily interested in the onset of scalarization and descalarization. Therefore, we choose the coupling function f(\phi)=\phi^{2} such that \Psi=0 . The theory corresponding to this choice of the coupling function is known as quadratic sGB gravity. Substituting this coupling function into Eq. ( 3a ) gives a Klein-Gordon like equation, \displaystyle\left(\Box-\mu_{\rm eff}^{2}\right)\phi=0,\,\, (4a) \displaystyle\mu_{\rm eff}^{2}=-\frac{\alpha_{\rm GB}}{2}\mathcal{G}, (4b) where the curvature acts as an effective mass, \mu_{\rm eff} . We can gain some insight into how an instability develops by considering Eq. ( 4a ) in flat space. If we forget for a moment that the effective mass is curvature dependent, then we might suppose that Eq. ( 4a ) would admit exponentially growing plane wave solutions with wavenumber, k , when \mu_{\rm eff}^{2}<-k^{2} . In principle, these plane waves could have infinite wavelength and vanishingly small wavenumber, so scalarization would occur for any tachyonic effective mass, \mu_{\rm eff}^{2}<0 . However, the effective mass is curvature dependent, so for any non-zero effective mass, spacetime cannot truely be flat. Nevertheless, spacetime is always locally flat within a sufficiently small region. This means that plane waves can only be solutions in an approximate local sense and must therefore have finite wavelength, which imposes a lower bound on the wavenumber. Consequently, as we increase our curvature in an attempt to satisfy \mu_{\rm eff}^{2}<-k^{2} , we also make the constraint more restrictive. To know when the conditions for an instability are satisfied, one must consider the particular geometry of a given system. In stationary radially symmetric systems, one can find a condition on the existence of scalarized solutions using an effective potential formalism 55 . However, in dynamical systems, there is no simple analytic prescription to find a condition for scalarization. II.2 Black hole scalarization in a nutshell In this work, we consider the scalarization of BH binary systems. Let us begin by reviewing the conditions under which individual BHs scalarize as a guide. In GR, BHs are uniquely described by three parameters: (1) their mass, m , (2) their (dimensionless) spin, \chi , and (3) their electric charge 2 2 2 BHs can only acquire an electric charge in Einstein-Maxwell theory, where GR is minimally coupled to electromagnetism. 76 ; 115 ; 137 ; 26 ; 25 ; 43 . These parameters influence the curvature of the spacetime around a BH and the corresponding Gauss-Bonnet invariant. Here we consider BHs with zero electric charge. The conditions under which a BH scalarizes thus depends on the BH mass, the BH spin, and the coupling constant. The coupling constant is dimensionful and has units of mass squared. Therefore, it is typical to reexpress it as the dimensionless coupling, \beta=\alpha_{\rm GB}/\mathrm{M}^{2}\,, (5) where \mathrm{M} is some characteristic mass, which we take to be the total mass of a system. We thus use the dimensionless coupling when considering the criterion for scalarization. Since scalarization requires a tachyonic effective mass, the dimensionless coupling and Gauss-Bonnet invariant must either both be positive or both be negative; see Eq. ( 4b ). In the vicinity of slowly spinning BHs, the Gauss-Bonnet invariant is entirely positive. However, for BHs with spin magnitude |\chi|>0.5 , the Gauss-Bonnet invariant becomes negative around the spin axis. Two types of scalarization are thus possible depending on the sign of the dimensionless coupling: spontaneous scalarization 57 ; 126 and spin-induced scalarization 54 ; 72 ; 74 . Spontaneous scalarization ( \beta>0 ): Let us first consider the criterion for the spontaneous scalarization of BHs with positive dimensionless coupling. A single non-spinning BH of mass m=\mathrm{M}=1 experiences the onset of scalarization for a critical coupling of \beta_{\rm c,1}=1.451 126 . In the test field approximation, the BH scalarizes for any dimensionless coupling above this value. For a BH with a different mass, m^{\prime} , expressed in the same unit system, the (dimensionless) critical coupling would be the same, but the total mass would differ (i.e., m^{\prime}=\mathrm{M}\neq 1 ). Consequently, the dimensionful critical coupling would be \alpha_{\rm GB,c}(m^{\prime})=m^{\prime 2}\beta_{\rm c,1} . If we consider a system of isolated BHs with individual masses, m_{\rm\mathfrak{n}} , the expression for the dimensionful critical coupling still applies. However, the mass of a single BH no longer coincides with the total system mass, \mathrm{M}=\sum m_{\rm\mathfrak{n}} . Thus, the formula for the (dimensionless) critical coupling of the \mathrm{\mathfrak{n}} -th BH is \beta_{\rm c}(m_{\rm\mathfrak{n}})=(m_{\rm\mathfrak{n}}/\mathrm{M})^{2}\beta_{\rm c,1}\,. (6) This formula can be used to find the critical coupling for any non-spinning BH. In the test field approximation, a BH scalarizes if the dimensionless coupling is greater than its critical coupling, \beta\geq\beta_{\rm c}(m_{\rm\mathfrak{n}}) . In a similar manner, if a system has a dimensionless coupling \beta=\beta_{\rm c}(m_{\rm c}) , then BHs with mass m_{\rm\mathfrak{n}}\leq m_{\rm c} scalarize and BHs with mass m_{\rm\mathfrak{n}}>m_{\rm c} descalarize (or remain unscalarized). For systems with a positive coupling constant, increasing spin tends to suppress scalarization 51 . However, we do not consider this effect here. Spin-induced scalarization ( \beta<0 ): Let us now consider the criterion for the spin-induced scalarization of BHs with spin magnitude |\chi|>0.5 and negative dimensionless coupling. We are unaware of an exact solution for the spin dependence of the critical coupling. However, Ref. 61 used data from Ref. 72 to find a best fit for the spin-dependent critical coupling of a BH expressed in units of its own mass (i.e., m=\mathrm{M}=1 ), \beta_{\rm c,1}(\chi)=-\frac{0.422}{(|\chi|-1/2)^{2}}+1.487|\chi|^{7.551}\,. (7) The fit has an error of \sim 5 % for spin magnitudes |\chi|<0.74 and of \sim 15 % for spin magnitudes 0.74<|\chi|<1 61 . The dimensional argument used to derive the mass scaling in Eq. ( 6 ) also applies here. Thus, in tandem with Eq. ( 6 ), this formula can approximate the critical coupling for the spin-induced scalarization of a BH with any mass. In this case, a BH scalarizes if the dimensionless coupling is smaller (i.e., more negative) than its critical coupling, \beta\leq\beta_{\rm c}(m_{\rm\mathfrak{n}},\chi) . Furthermore, if a system has a dimensionless coupling \beta=\beta_{\rm c}(m_{\rm\mathfrak{n}},\chi_{\rm c}) , then BHs with mass m_{\rm\mathfrak{n}} scalarize for spins |\chi|\geq|\chi_{\rm c}| and descalarize (or remain unscalarized) for spins |\chi|<|\chi_{\rm c}| . Scalar charge: Once a BH scalarizes, it is said to acquire a scalar hair. Asymptotically, the scalar field has a radial dependence similar to the electric potential of a point charge, \phi(r)=\Psi+Q_{\rm s}/r+\mathcal{O}(r^{-2})\,. (8) Therefore, the BH is also said to acquire a scalar charge, which is denoted by Q_{\rm s} 126 . III Numerical Framework In this work, we simulate a series of BH binaries in quadratic sGB gravity using the decoupling approximation, where the scalar field does not backreact onto the metric. The metric thus evolves according to GR. We follow Refs. 127 ; 61 and adopt the procedures laid out in Refs. 140 ; 27 ; 28 . III.1 Spacetime evolution We utilize a 3+1 implementation of numerical relativity, where the four dimensional spacetime is foliated into a set of three dimensional spatial hypersurfaces parameterized by a time coordinate, t ; see e.g. Refs. 13 ; 24 ; 120 . The hypersurfaces are described by the 3-metric, \gamma_{ij} , and embedded into the four dimensional spacetime as prescribed by their orthogonal unit normal vector, n^{\mu} . The unit normal vector is then used to define the projection operator, P^{\alpha}_{\beta}=\delta^{\alpha}_{\beta}+n^{\alpha}n_{\beta} , which projects tensors onto the hypersurfaces. For example, the covariant derivative on a 3-metric is given by D_{i}=P^{\alpha}_{i}\nabla_{\alpha} . Furthermore, the extrinsic curvature of a hypersurface is given by K_{ij}=-P^{\alpha}_{i}P^{\beta}_{j}\nabla_{\alpha}n_{\beta}=-\frac{1}{2}\mathcal{L}_{\bm{n}}\gamma_{ij}\,, (9) where \mathcal{L}_{\bm{n}} is the Lie derivative with respect to the normal vector. Given initial data for the 3-metric of a hypersurface and its extrinsic curvature, a system can be evolved in time. Here we summarize the mathematical formulation. In terms of the 3-metric, the spacetime metric is ds^{2}=-(\alpha^{2}-\beta^{i}\beta_{i})dt^{2}+2\beta_{i}dx^{i}dt+\gamma_{ij}dx^{i}dx^{j}, (10) where the lapse function, \alpha , and the shift vector, \beta^{i} , relate the hypersurfaces at different times. Namely, an observer that is stationary with respect to a hypersurface is separated from their position on the next hypersurface by a proper time of \alpha dt . Their position within the next hypersurface is then given by x^{i}_{t+dt}=x^{i}_{t}-\beta^{i}dt . The lapse and shift are ultimately coordinate dependent and thus provide a choice of gauge. To generate initial data for the GR background, we use the Bowen-York method as extended by Brandt and Brügmann 33 ; 35 to solve the constraint equations \displaystyle{}^{(3)}R+K^{2}-K_{ij}K^{ij}=0\,, (11a) \displaystyle D_{j}\left(K^{ij}-\gamma^{ij}K\right)=0\,, (11b) for the 3-metric and extrinsic curvature in vacuum. Here, {}^{(3)}R is the three dimensional Ricci scalar describing the intrinsic curvature of a hypersurface, and K=\gamma^{ij}K_{ij} is the trace of the extrinsic curvature; see e.g. Refs. 13 ; 24 ; 120 for further details on the constraint equations. In Arnowitt-Deser-Misner-York form, the vacuum evolution equations are written as \displaystyle\partial_{t}\gamma_{ij}=-2\alpha K_{ij}+\mathcal{L}_{\bm{\beta}}\gamma_{ij}\,, (12a) \displaystyle\begin{aligned} \partial_{t}K_{ij}=&-D_{i}D_{j}\alpha+\mathcal{L}_{\bm{\beta}}K_{ij}\\ &+\alpha\left[{}^{(3)}R_{ij}+KK_{ij}-2K_{ik}K^{k}_{j}\right]\,,\end{aligned} (12b) where {}^{(3)}R_{ij} is the three dimensional Ricci tensor describing the intrinsic curvature of a hypersurface, and \mathcal{L}_{\bm{\beta}} is the Lie derivative along the shift vector. To evolve the simulations, we use the Baumgarte-Shapiro-Shibata-Nakamura (BSSN) implementation of the 3+1 formalism 121 ; 23 together with moving puncture gauge 39 ; 21 . III.2 Scalar evolution Let us now consider the numerical description of the scalar field. In analogy to the extrinsic curvature in Eq. ( 9 ), we introduce the momentum of the scalar field as K_{\phi}=-\frac{1}{2}\mathcal{L}_{\bm{n}}\phi\,. (13) We initialize the scalar field around each BH using the bound state solution of an isolated, non-spinning BH obtained by Ref. 126 . Following Refs. 127 ; 61 , we thus approximate the solution of a many-body system using the formulae, \displaystyle\phi|_{t=0}=\sum_{\rm\mathfrak{n}}\phi_{\rm\mathfrak{n}}\,,\hskip 19.91684ptK_{\phi}|_{t=0}=0\,, (14a) \displaystyle\phi_{\rm\mathfrak{n}}=c_{\rm 0}\left[c_{\rm 1}\rho_{\rm\mathfrak{n}}+c_{\rm 2}\rho_{\rm\mathfrak{n}}^{2}+c_{\rm 3}\rho_{\rm\mathfrak{n}}^{3}\right]\,, (14b) \displaystyle\rho_{\rm\mathfrak{n}}=\frac{m_{\rm\mathfrak{n}}r_{\rm\mathfrak{n}}}{(m_{\rm\mathfrak{n}}+2r_{\rm\mathfrak{n}})^{2}}\,, (14c) where \phi_{\rm\mathfrak{n}} is the bound state solution of the \mathfrak{n} -th BH and r_{\rm\mathfrak{n}} is the distance to the \mathfrak{n} -th BH in quasiisotrpic radial coordinates. The coefficients c_{\rm i} are expansion parameters used to fit the numerical solution in Ref. 126 . They are fit to be c_{\rm 1}=3.68375 , c_{\rm 2}=4.97242 , and c_{\rm 3}=2.29938\times 10^{2} . The parameter c_{\rm 0} sets the initial scalar field amplitude. In the following, we consider BH binaries, where the BHs begin widely separated. Note that Eq. ( 4a ) is linear and homogeneous with respect to the scalar field. It is therefore reasonable to approximate the initial scalar field as a sum of bound states around the individual BHs. Moreover, the initial scalar field amplitude only influences the evolution up to an overall constant and is thus arbitrary. Note that in Eq. ( 14 ) we assume the field is momentarily at rest, i.e., its momenta is set to zero. Furthermore, Eq. ( 14 ) does not take into account a BH’s spin. Therefore, the scalar field takes time to equilibrate. For a backreacted solution see Ref. 101 . Once the initial data for the scalar field is specified, it is evolved according to the equations, \displaystyle(\partial_{t}-\mathcal{L}_{\bm{\beta}})\phi= \displaystyle-\alpha K_{\phi}\,, (15a) \displaystyle(\partial_{t}-\mathcal{L}_{\bm{\beta}})K_{\phi}= \displaystyle-D^{i}\alpha D_{i}\phi (15b) \displaystyle-\alpha\left(D^{i}D_{i}\phi-KK_{\phi}+\frac{1}{2}\alpha_{\rm GB}\mathcal{G}\phi\right)\,. Since we work in the decoupling approximation, the Gauss-Bonnet invariant is calculated from the curvature of the time-dependent background spacetime 140 ; 27 ; 28 . We note that every system must be provided with some initial scalar field as otherwise there would be no scalar dynamics; see Eq. ( 4a ). Therefore, we seed every BH with an initial scalar field using Eq. ( 14 ). However, depending on the parameters of a system, the BHs may not be able to sustain a scalar field, and thus the scalar field decays; see Sec. II.2 . In the following, we refer to these BHs as unscalarized. III.3 Code We run simulations with the Einstein Toolkit 114 ; 94 ; 145 , an open-source software for computational astrophysics, and the Canuda code 142 ; 104 ; 146 ; 140 ; 127 ; 112 for fundamental physics. The Einstein Toolkit is built upon the Cactus computational framework 70 ; 37 and uses Carpet 118 ; 42 to implement box-in-box adaptive mesh refinement as well as hybridized message passing interface and open multiprocessing parallelization. We generate the initial data for the background metric using the TwoPunctures spectral thorn 16 . The initial data for the scalar field is generated using Canuda ’s EdGB_dec_Init thorn. We then evolve the system using Canuda ’s LeanBSSNMoL 3 3 3 This thorn is adapted from the Lean code 132 . and EdGB_dec_Evol thorns, which handle the background and scalar evolutions, respectively. LeanBSSNMoL and EdGB_dec_Evol provide up to eighth order finite differences for spatial derivatives. Here, we use fourth order finite differencing for spatial derivatives and employ the fourth order Runge-Kutta scheme for the time integration. The system is evolved via the method of lines. III.4 Extraction of variables In order to analyze the simulations, we require a variety of information about the scalar field and spacetime evolution. The scalar field is handled by Canuda ’s EdGB_dec_Base thorn, and the Gauss-Bonnet invariant is calculated from the background in Canuda ’s EdGB_dec_Evol thorn. In order to visualize the behavior of these quantities, we use the CarpetIOHDF5 thorn to output the values of the quantities across two dimensional cross sections of our grid as a function of time. Furthermore, we decompose the scalar field into a basis of spherical harmonics, \phi_{\rm\mathfrak{l}\mathfrak{m}}(t,r_{\rm ex})=\int\phi(t,r_{\rm ex},\theta,\phi)Y^{*}_{\mathfrak{l}\mathfrak{m}}(\theta,\phi)d\Omega\,, (16) on spheres of constant extraction radii, r_{\rm ex} , using the Multipole thorn 73 . Here, Y^{*}_{\mathfrak{l}\mathfrak{m}}(\theta,\phi) are the complex conjugates of the spherical harmonics, Y_{\mathfrak{l}\mathfrak{m}}(\theta,\phi) , with spin-weight zero. The effective scalar charge of a system is then given by r_{\rm ex}\phi_{\rm 00} for a sufficiently large extraction radius; see Eq. ( 8 ). We compute the BH apparent horizons and their properties, such as the horizon area, A_{\rm H} , and equatorial circumference, C_{\rm e} , using the AHFinderDirect thorn 135 ; 136 . Furthermore, we use these quantities in postprocessing to compute the (dimensionless) spin, \chi=\sqrt{1-\left(\frac{2\pi A_{\rm H}}{C_{\rm e}^{2}}-1\right)^{2}}\,, (17) of the BHs. Note that the dimensionless spin is only well defined when the BHs are relatively isolated. We find that other common methods for computing the spin, such as via a ratio of horizon circumferences or quasilocal measures, give similar results; see also Refs. 15 ; 83 ; 102 . We also use the AHFinderDirect thorn to track the position of the BH punctures in tandem with the PunctureTracker thorn 89 . Finally, the AHFinderDirect thorn outputs the minimum and maximum radii of the BH horizons, which we use to reconstruct the approximate shape of the horizons during the evolution. III.5 Grid setup and refinement We evolve the BH binaries on a three dimensional cartesian grid with outer boundary located at x,y,z=\pm 256\mathrm{M} , where \mathrm{M} is the total mass of the system, and \mathrm{M=1} in code units. To reduce the computational cost, we impose rotation symmetry about the z-axis and reflection symmetry across the xy-plane. We use Carpet to employ seven levels of box-in-box adaptive mesh refinement centered around the BHs. The outermost refinement level has a resolution with step size dx=1\mathrm{M} . Within consecutive refinement levels, we halve the step size such that the innermost refinement level has step size dx=\frac{1}{64}\mathrm{M} . We set the Courant factor to dtfac =0.225 . We use two different setups for the refinement levels around the BHs. The first (setup A) places the refinement boundaries at radii r/\rm{M}=\{64.0,16.0,6.0,3.0,1.5,0.75\} around the BH centers. The second setup (setup B) places the refinement boundaries at radii r/\rm{M}=\{64.0,16.0,4.0,2.0,1.0,0.6\} . We use setup A in the majority of cases and setup B in systems where the BHs have zero initial spin. We present a series of convergence tests in the Appendix. IV Setup In the following, we analyze the evolution of a scalar field in the (GR) background of BH binaries with the initial setup depicted in Fig. 1 . The BHs have equal initial masses m_{\rm i}=\mathrm{M}/2 , where \mathrm{M} is the total system mass. Note that the BHs have equal, but opposite, initial momenta, |\bm{P}_{\rm i}| , inclined at an incident angle, \theta , with respect to the x-axis. The BHs are equidistant from the z-axis with initial separation d=100\mathrm{M} . Consequently, the systems are rotationally symmetric about the z-axis. In several of the simulations, we also provide the BHs with an initial (dimensionless) spin given by the parameter \chi_{\rm 0} . We orient the BH spins along the z-axis such that they are both aligned (+ z ) or anti-aligned (- z ) to the orbital angular momentum. When referring to the spin, \chi , we therefore refer solely to the z-component and take positive (negative) values to refer to aligned (anti-aligned) spins. Note that the spins maintain rotational symmetry. The BHs’ (GR) configuration is the same as in Ref. 88 . Here, we consider BHs with initial spin parameter \chi_{\rm 0}=0.0 and initial momentum |\bm{P}_{\rm i}|=0.245\mathrm{M} , as well as BHs with initial spin parameters \chi_{\rm 0}=\{-0.7,0.7\} and initial momentum |\bm{P}_{\rm i}|=0.490\mathrm{M} . m_{\rm i} |\bm{P}_{\rm i}| \theta m_{\rm i} |\bm{P}_{\rm i}| \theta x y z d Figure 1: Diagram illustrating the initial setup of the GR background for the BH binary systems that we consider. The system has rotational symmetry about the z-axis such that the BHs have equal initial mass, m_{\rm i} , initial momentum, |\bm{P}_{\rm i}| , and incident angle, \theta . The BHs have initial separation d=100\mathrm{M} . For sufficiently large initial momenta, such BHs are initially unbound in a Newtonian sense and are thus considered to be in a hyperbolic configuration. However, as they pass one another during a close encounter, they lose energy to gravitational radiation, which can cause them to become bound and undergo dynamical capture. This results in three distinct morphologies as the incident angle is increased: (1) direct mergers, where the BHs merge during the first encounter; (2) zoom whirls, where the BHs undergo a series of close encounters (i.e., small orbits) punctuated by relatively large eccentric orbits until they lose sufficient energy such that they merge; and (3) scattering, where the BHs separate and escape to infinity after a single close encounter 107 ; 122 ; 131 ; 71 ; 68 ; 69 . IV.1 Spin-up of scattering black holes When BHs undergo a close encounter, they emit and reabsorb gravitational radiation. This gravitational radiation carries energy and angular momentum, which changes the mass, angular momentum, and ultimately the (dimensionless) spin of the BHs 130 ; 102 ; 77 ; 116 ; 47 ; 88 . This change in parameters usually leads to a spin-up, where the spin increases along the direction of a system’s orbital angular momentum (i.e., the z-axis), although a spin-down, where the spin decreaces with respect to the direction of the orbital angular momentum, is also possible 130 ; 116 ; 47 ; 88 . The change in spin can be quantified as the difference between the final spin and the initial spin, \Delta\chi=\chi_{\rm f}-\chi_{\rm i}\,, (18) where here we compute the final spin, \chi_{\rm f} , and initial spin, \chi_{\rm i} , according to Eq. ( 17 ). We compute the initial spin at a time, t_{\rm i}=45\mathrm{M} , approximately halfway between the start of the simulation and the close encounter (i.e., after the initial gauge adjustment). We compute the final spin at a time, t_{\rm f}=270\mathrm{M} , long after the encounter. Note that the initial spin computed from Eq. ( 17 ) typically differs slightly ( <1\% ) from the initial spin parameter, \chi_{\rm 0} , due to numerical and theoretical limitations in the initial data construction of spinning BHs. Figure 2: The change in spin, \Delta\chi , of scattering binary BHs with initial momentum |\bm{P}_{\rm i}|=0.490\mathrm{M} and initial spin parameters \chi_{\rm 0}=\{-0.7,0.0,0.7\} as a function of the incident angle, \theta . The plot is adapted from Fig. 6 in Ref. 88 . In this work, we consider systems with the same GR parameters as those denoted by the red squares. In this work, we are concerned with the role that BH spins, and their change during an encounter, may play in the spin-induced scalarization of BHs, which can only occur for spin magnitudes |\chi|>0.5 ; see Eq. ( 7 ). Therefore, we consider systems with initial spin parameters of magnitude |\chi_{\rm 0}|=0.7 . In Fig. 2 , we plot the change in spin of scattering BH binaries against their incident angle for BHs with initial momenta |\bm{P}_{\rm i}|=0.490\mathrm{M} and initial spin parameters \chi_{\rm 0}=\{-0.7,0.0,0.7\} ; the plot is adapted from Fig. 6 in Ref. 88 . We include initial spin parameter \chi_{\rm 0}=0.0 for comparison. Typically, the spin-up of scattering BHs increases with decreasing initial spins and incident angles. The behavior of simulations with initial spin parameter \chi_{\rm 0}=0.7 differs and exhibits a spin-down at intermediate angles. At large incident angles the change in spin approaches zero. The red squares in Fig. 2 denote systems with the same GR parameters as those studied below. The setup with initial spin parameter \chi_{\rm 0}=0.7 and small incident angle yields an increase in spin magnitude. The setup with initial spin parameter \chi_{\rm 0}=-0.7 leads to a decrease in spin magnitude. The setup with initial spin parameter \chi_{\rm 0}=0.7 and large incident angle corresponds to a minute change in spin. Such systems are useful as we consider the construction of hyperbolic encounters in which different types of scalarization may occur. IV.2 Simulation Suite \overline{S} \overline{S} \overline{S}\uparrow S{-}S S{-}S \overline{S} \overline{S} (a) ExptBp0355X0 \overline{S}\downarrow \overline{S}\downarrow \overline{S}\downarrow \overline{S}\downarrow S{-}S (b) ExptBm300Xm07 \overline{S}\uparrow \overline{S}\uparrow S\Uparrow S\Uparrow S{-}S (c) ExptBm300Xp07 S\Downarrow S\Downarrow \overline{S}\downarrow \overline{S}\downarrow S{-}S (d) ExptBm350Xm07 Figure 3: Schematic depiction of the phenomenology of close encounters in quadratic sGB gravity. Time moves from left to right. Black lines denote the trajectories of the BHs. Wavy lines denote the interaction of the BHs during a close encounter. The symbol S ( \overline{S} ) denotes scalarized (unscalarized) BHs. Arrows \uparrow ( \downarrow ) denote spins aligned (anti-aligned) with the orbital angular momentum. When the change in spin is important to the dynamics of the process, arrows \Uparrow ( \Downarrow ) denote aligned (anti-aligned) spins with greater magnitude. Panel 3a : The dynamical scalarization of two BHs during a zoom-whirl and merger. Panel 3b : The spin-induced dynamical scalarization of two scattering BHs. Panel 3c : The permanent spin-up scalarization of two scattering BHs as a consequence of their spin-up. Panel 3d : The permanent spin-up descalarization of two scattering BHs as a consequence of their spin-up, which leads to a decrease their spin magnitude. Run \beta c_{\rm 0} \chi_{\rm 0} \chi_{\rm i} \chi_{\rm f} \Delta\chi Type |\bm{P}_{\rm i}|/\mathrm{M} \theta Phenomenon ExptBp0355X0 0.355 1.0 0.0 N/A N/A N/A Zoom-Whirl 0.245 0.0580 Dynamical scalarization CtrlBm300Xp07 -3.00 1.0 0.7 0.695 0.687 -0.008 Scattering 0.490 0.05685 Control, unscalarized ExptBm300Xm07 -3.00 1.0 -0.7 -0.695 -0.391 0.304 Scattering 0.490 0.05685 Dynamical scalarization ExptBm300Xp07 -3.00 1.0 0.7 0.695 0.718 0.023 Scattering 0.490 0.0425 Spin-up scalarization CtrlBm350Xp07 -3.50 0.1 0.7 0.695 0.687 -0.008 Scattering 0.490 0.05685 Control, scalarized ExptBm350Xm07 -3.50 0.1 -0.7 -0.695 -0.391 0.304 Scattering 0.490 0.05685 Spin-up descalarization Table 1: List of simulations. For each simulation we provide a run name, the dimensionless coupling, \beta , and the initial scalar field amplitude, c_{\rm 0} . Furthermore, we provide the BHs’ initial spin parameter, \chi_{\rm 0} , as well as the initial spin, \chi_{\rm i} , final spin, \chi_{\rm f} , and change in spin, \Delta\chi , computed from Eq. ( 17 ). We also provide the type of the encounter, the initial momentum of the BHs, |\bm{P}_{\rm i}| , and the incident angle, \theta . Finally, we state the phenomenon each simulation is designed to display. Each system has an initial separation of d=100\mathrm{M} . The fact that BHs can undergo extremely close flybys without forming a CAH while scattering and during the first encounter(s) of a zoom-whirl, makes such morphologies natural settings in which to search for dynamical scalarization 79 ; 101 ; 91 . Furthermore, the spin-up of BHs during such encounters makes the scattering of initially spinning BHs an interesting setting in which to look for spin-induced scalarization (or descalarization ). Here we have two goals: We seek to observe instances of (1) temporary dynamical scalarization during the course of these encounters and (2) permanent spin-induced scalarization (or descalarization ) owing to the spin-up of the BHs, which we call spin-up scalarization (or descalarization ). In order to achieve these goals, we run a series of simulations depicted schematically in Fig. 3 and described in Table 1 . Fig. 3 contains “Feynmann”-like diagrams illustrating the phenomenology of different numerical “experiments.” Time moves from left to right. We denote BH trajectories by solid black lines. Close encounters are denoted by wavy lines, and a merger is denoted by the intersection of two solid lines. We denote segments of the trajectories in which BHs are scalarized (unscalarized) with the symbol S ( \overline{S} ). Note that every BH is seeded with some initial scalar field; see Sec. III.2 . However, around “unscalarized” BHs, the scalar field decays. We denote the combined scalarization of the BHs due to a close encounter as S{-}S . We denote positive (negative) spins aligned (anti-aligned) to the orbital angular momentum with arrows \uparrow ( \downarrow ). In systems where the change in spin magnitude is important to the scalar dynamics, we use arrows \Uparrow ( \Downarrow ) to denote spins with relatively large magnitude. Table 1 lists the simulation parameters and other information describing the experiments. Table 1 also details a set of “control” simulations to be discussed in due course. We now summarize these simulations and their outcomes. We first consider a binary of non-spinning BHs that undergoes a zoom-whirl and merger. We then discuss two control simulations for binaries with spinning BHs and negative dimensionless coupling. Finally, we describe a series of scattering encounters between spinning BHs, which constitute the remainder of our experiment runs. IV.2.1 Numerical experiment without spin ExptBp0355X0 , depicted in Fig. 3a and listed in Table 1 , is set up to achieve the first goal of finding dynamical scalarization during the course of a close encounter. In this case, we consider BHs that are initially non-spinning. The parameters of ExptBp0355X0 are chosen such that the BHs undergo a single zoom-whirl and then merge. In order to see if we can observe dynamical scalarization , we set the dimensionless coupling just below the critical value of the positive dimensionless coupling given by Eq. ( 6 ) in Sec. II , such that the BHs begin unscalarized. For a BH binary system with equal initial masses m_{\rm i}=\mathrm{M}/2 , the critical coupling given by Eq. ( 6 ) is \beta_{\rm c}=0.36275 . In this simulation, we therefore set the dimensionless coupling slightly below that value at \beta=0.355 . The results are displayed in Sec. V.1 . We ultimately find that the system does temporarily display dynamical scalarization during the first encounter. We then find that the BHs begin to descalarize during the orbit between encounters, briefly show some signs of dynamically rescalarizing during the inspiral, and then descalarize after merger. The remnant BH has a spin of \chi=0.733 . A natural followup question to ask is whether one can observe a spin-induced version of dynamical scalarization . However, in order to consider scalarization in systems of initially spinning BH, we must take extra care in considering the initial conditions of the scalar field. IV.2.2 Control simulations When considering setups with a positive coupling constant and non-spinning BHs, determining the BHs’ critical coupling and selecting an appropriate dimensionless coupling for the system is straightforward. In this case, the formula for the critical coupling, Eq. ( 6 ), is exact. Moreover, the initial data reliably produces the intended initial masses with high accuracy. Therefore, we can confidently control whether the BHs begin scalarized or unscalarized by setting the dimensionless coupling above or below the critical value given by Eq. ( 6 ). For spinning BHs in systems with a negative coupling constant, this process is far less straightforward. First, the formula for the (negative) critical coupling in Eq. ( 7 ) is merely a fit to numerical data and thus carries an inherent uncertainty. For example, the critical coupling \beta_{\rm c}=-2.61 , associated with a spin magnitude of |\chi|=0.7 and a mass m_{\rm\mathfrak{n}}=\mathrm{M}/2 , has an error of \pm 0.13 ; see Ref. 61 . We also note that Ref. 61 estimates an error of 15\% if the spin magnitude is above |\chi|=0.74 . Another issue is that the initial spin parameter, \chi_{\rm 0} , deviates from the initial spin, \chi_{\rm i} , found using Eq. ( 17 ) after gauge adjustments. As reported in Table 1 , for simulations with an initial spin parameter of \chi_{\rm 0}=0.7 , we find an initial spin of \chi_{\rm i}=0.695 after gauge adjustments. The critical coupling associated with a spin magnitude of |\chi|=0.695 and mass m_{\rm\mathfrak{n}}=\mathrm{M}/2 is \beta_{\rm c}=-2.75 , and it has an error of \pm 0.14 . A further issue is that the measured spin can vary depending on the method used to extract it. The data in Ref. 72 , upon which Eq. ( 7 ) is based, measures spin using the asymptotic behavior of semianalytically computed metric functions for a single BH. We cannot apply such a method to a binary system simulated on a full numerical grid and thus do not know how closely our spin measurements correspond to those of Ref. 72 . While we use Eq. ( 7 ) as an approximate guide for what the critical coupling of a spinning BH in our simulations may be, we must find a way to verify whether initially spinning BHs begin scalarized or unscalarized. The initial scalar field is constructed assuming a non-spinning configuration; see Sec. III.2 . Consequently, there is an initial period during which the scalar field adjusts to the geometry, making it difficult to determine whether a BH is scalarizing or descalarizing in the short interval preceding an encounter. We thus run a series of control simulations in which the incident angle is large, such that the spin-up during the encounter is marginal (see Fig. 2 ). We can be reasonably certain that the behavior of the scalar field after the encounter matches its behavior before the encounter in these systems. Therefore, we can use them to verify the initial state of spinning BHs in other systems with the same dimensionless coupling. CtrlBm300Xp07 , listed in Table 1 , is set up as a control run to verify scenarios in which BHs with an initial spin parameter magnitude of |\chi_{\rm 0}|=0.7 are initially unscalarized. The simulation is chosen to have a large incident angle, such that the change in BH parameters is marginal; the BHs only experience a spin-down of \Delta\chi=-0.008 . Through trial and error, the dimensionless coupling is chosen to be \beta=-3.00 . The scalar field ultimately decays, suggesting that BHs with this initial spin magnitude and dimensionless coupling are unscalarized. CtrlBm350Xp07 , listed in Table 1 , is set up as a control run to verify scenarios in which BHs with an initial spin parameter magnitude of |\chi_{\rm 0}|=0.7 are initially scalarized. The background spacetime is the same as in CtrlBm300Xp07, but here the dimensionless coupling is chosen to be \beta=-3.50 . Note that the initial scalar field amplitude is set to c_{\rm 0}=0.1 , as is also the case in the ExptBm350Xm07 run. The scalar field ultimately grows exponentially, suggesting that BHs with the same initial spin magnitude and dimensionless coupling are scalarized. A more detailed analysis of the results from both control simulations can be found in Sec. V.2 . IV.2.3 Numerical experiments with spin Now that we understand the initial conditions for scalarized and unscalarized BH binaries with spin, we proceed with the study of spin-induced (de)scalarization . We begin by returning to the matter of dynamical scalarization . ExptBm300Xm07 , depicted in Fig. 3b and listed in Table 1 , is set up to achieve the first goal of finding dynamical scalarization in systems with initially spinning BHs and a negative coupling constant. The parameters of ExptBm300Xm07 are chosen such that the BHs scatter and thus undergo a single close encounter before separating. The BHs have an initial spin of \chi_{\rm i}=-0.695 and experience a spin-up of \Delta\chi=0.304 during the encounter, leading to a final spin of \chi_{\rm f}=-0.391 . In order to see if we can observe spin-induced dynamical scalarization , we want the BHs to begin (and end) unscalarized and thus set the dimensionless coupling to \beta=-3.00 in order to match the unscalarized control CtrlBm300Xp07. The results are displayed in Sec. V.3 . We ultimately find that the system does temporarily display spin-induced dynamical scalarization during the encounter. Note that the change in spin is not necessary for this process and is, in fact, likely to hinder it. We then find that the BHs descalarize after the encounter because the individual BHs’ spins are still too low for spin-induced scalarization . ExptBm300Xp07 , depicted in Fig. 3c and listed in Table 1 , is set up to achieve the second goal of finding permanent spin-up scalarization due to an increase in the BH spin. The initial parameters of ExptBm300Xp07 are chosen such that the BHs scatter and thus undergo a single close encounter, which results in a small spin-up of \Delta\chi=0.023 . The BHs have an initial spin of \chi_{\rm i}=0.695 and a final spin of \chi_{\rm f}=0.718 . In order to see if we can observe spin-up scalarization , we want the BHs to begin unscalarized and thus set the dimensionless coupling to \beta=-3.00 . The change in spin may then result in a system where the critical coupling of the final BHs is less negative than the dimensionless coupling, leading the BHs to scalarize. The results are displayed in Sec. V.4 . Like ExptBm300Xm07, the system first scalarizes during the encounter. However, after the encounter, the scalar field continues to grow, rather than decay, indicating that the BHs become permanently scalarized. We thus find that the system does display spin-up scalarization over the course of the encounter. ExptBm350Xm07 , depicted in Fig. 3d and listed in Table 1 , is set up to achieve the second goal of finding permanent spin-up descalarization due to changes in the BH spin. The initial parameters of ExptBm350Xm07 are chosen such that the BHs scatter and thus undergo a single close encounter, which results in a spin-up of \Delta\chi=0.304 . The initial spin of the BHs is negative, \chi_{\rm i}=-0.695 , so the spin-up results in an overall decrease in the spin magnitude such that the final spin is \chi_{\rm f}=-0.391 . In order to see if we can observe spin-up descalarization , we want the BHs to begin scalarized and thus set the dimensionless coupling to \beta=-3.50 . As the final spin is below the minimum value for spin-induced scalarization , |\chi_{\rm f}|<0.5 , we expect the final BHs to descalarize. Since the dimensionless coupling has a particularly large magnitude, we set the initial scalar field amplitude to c_{\rm 0}=0.1 . As discussed in Sec. III , this does not change the dynamics in the decoupling limit; however, it helps minimize reflection off the inner refinement boundaries. The results are displayed in Sec. V.5 . During the encounter, scalarization appears to be enhanced and the scalar field grows even faster than it does initially. However, after the encounter, the scalar field decays, indicating that the BHs become permanently unscalarized. We thus find that the system does display spin-up descalarization over the course of the encounter. Having described the simulation suites and setup, we next describe the results in Sec. V . V Results Figure 4: Snapshots of the scalar field, \phi , and Gauss-Bonnet invariant, \mathcal{G} , in the xy-plane (i.e., the orbital plane) from the ExptBp0355X0 run (see Fig. 3a and Table 1 ). Positive scalar field is shown in blue and negative scalar field is shown in yellow. Positive contours of the Gauss-Bonnet invariant, \mathcal{G}\mathrm{M}^{4}=\{10^{-3},10^{-2},10^{-1}\} , are depicted using dotted, dash-dotted, and dashed black lines, respectively. The trajectories of the BHs are denoted using solid black and red lines. In the following, we analyze the detailed results of the simulations summarized in Sec. IV . In Sec. V.1 we detail the results of the numerical “experiment” without spin, ExptBp0355X0. In Sec. V.2 we detail the results of the “control” simulations for spinning BHs, CtrlBm300Xp07 and CtrlBm350Xp07. In Secs. V.3 , V.4 , and V.5 we detail the results of the numerical “experiments” with spin: ExptBm300Xm07, ExptBm300Xp07, and ExptBm350Xm07. Animations depicting the different simulations can be found at Ref. 41 . V.1 Dynamical scalarization of two initially non-spinning black holes In this section, we analyze the results of the ExptBp0355X0 run listed in Table 1 and depicted in Fig. 3a . The BHs are initially non spinning and begin unscalarized. They undergo a zoom whirl and merger. In the following, we comment upon the successive dynamical scalarization , descalarization, and brief rescalarization of the binary BHs as they perform the zoom whirl, followed by their ultimate descalarization after merger. Animations created from this simulation can be found at Ref. 41 . In order to understand the behavior of the system, we begin by plotting snapshots of the scalar field in the orbital plane at a series of representative times in Fig. 4 . The value of the scalar field is shown using a color scale, where darker blues show more positive values and darker yellows show more negative values. The BH trajectories are plotted over the scalar field. Furthermore, we plot contours of the Gauss-Bonnet invariant, \mathcal{G}\mathrm{M}^{4}=\{10^{-3},10^{-2},10^{-1}\} . Qualitatively, we organize the evolution of the system into five phases, each of which aligns with one of the panels in Fig. 4 . Note that the time of the first close encounter (i.e., the smallest BH separation prior to merger) and the time at which the CAH is formed during the final inspiral are t_{\rm CE}=140\mathrm{M} and t_{\rm CAH}=987\mathrm{M} , respectively. Phase I ( 0\mathrm{M}0.5 to undergo spin-induced scalarization . In the case where the coupling constant is positive, we consider the behavior of initially unscalarized, non-spinning BHs, which undergo a single zoom whirl and then merge. We observe that the BHs temporarily undergo dynamical scalarization during the close encounter. The BHs descalarize following the first encounter, but then undergo a second round of dynamical scalarization during the final inspiral and briefly rescalarize. After the merger, the remnant BH descalarizes due to its larger mass (and smaller curvature). In the case where the coupling constant is negative, we consider the behavior of BHs with initial spin parameters of magnitude |\chi_{\rm 0}|=0.7 , which undergo a close encounter and scatter. When the BHs begin unscalarized, we find that the BHs can temporarily undergo spin-induced dynamical scalarization . This is because the regions of negative Gauss-Bonnet invariant along the BHs’ spin axes are temporarily enhanced and combine during the close encounter. This enhancement is caused by the orbital angular momentum of the BHs. We have further considered how the spin-up of BHs during scattering can affect the scalar field when the coupling constant is negative. We find that when the BHs begin unscalarized and the BHs’ spin magnitudes increase (here from |\chi_{\rm i}|=0.695 to |\chi_{\rm f}|=0.718 ) due to the encounter, the BHs can become permanently scalarized. Furthermore, we find that when the BHs begin scalarized and the BHs’ spin magnitudes decrease (here from |\chi_{\rm i}|=0.695 to |\chi_{\rm f}|=0.391 ) due to the encounter, the BHs can become permanently unscalarized. We refer to these processes as spin-up scalarization and spin-up descalarization , respectively. We thus find that close encounters have the power to bestow scalar hair upon a BH or to rescind it. While we have primarily considered the impact of hyperbolic encounters on the scalarization of BHs, we can also make inferences about how that scalarization might influence the evolution of the encounters if backreaction were taken into account. Properties such as the scattering angle or the number of the zoom-whirl orbits undertaken prior to merger can be very sensitive to the initial conditions of a binary; see e.g. Refs. 69 ; 111 . In GR, these quantities depend on the amount of kinetic energy lost to gravitational radiation. When more energy is emitted, it results in larger scattering angles. If sufficient energy is lost such that the system becomes bound, then the BHs undergo a series of zoom-whirls until they loose enough energy to merge. In quadratic sGB gravity, scalar radiation provides an additional mechanism through which kinetic energy can be lost. Consequently, one might expect to find larger scattering angles and fewer zoom-whirl orbits than one would in GR. Similarly, one might expect to find the transition between scattering, zoom-whirl, and direct merger morphologies at larger incident angles. Furthermore, close encounters result in a brief emission of GWs 66 ; 134 ; 38 ; 117 ; 20 ; 63 , potentially including uniquely identifiable quasi-normal modes 20 , which may be detectable in the future 65 ; 82 ; 87 ; 100 ; 99 ; 31 . The temporary or permanent scalarization of BHs during close encounters could leave imprints on these emissions. For example, one might expect to see a change in frequency, or a phase shift, as is sometimes observed during BH mergers in theories beyond GR and in scalar dark matter environments; see e.g. Refs. 19 ; 50 ; 18 ; 91 ; 46 ; 49 . These phenomena could be interesting to explore in systems with full backreaction. In the future, there are numerous other initial conditions and phenomena which may be interesting to explore within quadratic sGB gravity and other varieties of sGB gravity that allow for scalarization. Many of these theories (including quadratic sGB gravity) are invariant under a change in the sign of the scalar field. The finding that the scalar charge oscillates (and changes sign) in the aftermath of spin-induced dynamical scalarization , suggests the possibility that close encounters could result in permanent sign flips, where the scalar charge becomes negative. Such behaviors have already been found during inspirals 91 . One could further consider systems in which the two BHs are seeded by scalar fields with opposite signs. It would be interesting to see whether this enhances or suppresses dynamical scalarization . It may also be interesting to study systems where one BH begins scalarized, but the other does not. This could potentially lead to a permanent form of induced scalarization , where one BH seeds another with scalar field thus allowing it to scalarize. Furthermore, one could consider the scattering of BHs with unequal mass. The findings of Refs. 101 demonstrate that BHs too massive to scalarize can experience induced scalarization from a smaller BH. This suggests that a temporary dynamical form of induced scalarization may occur in close encounters. Finally, sGB gravity is far from the only theory beyond GR that posits a scalar field coupled to higher order curvature terms. Among these theories, there are others which exhibit spin-dependence or scalarization; see e.g. Refs. 55 ; 112 . It would be interesting to see whether the equivalents of the phenomena presented here can be found in the close encounters of other such theories. VII Acknowledgments We thank C. - H. Cheng, H. Kogan, N. Ghadiri, and H. O. da Silva, for insightful discussions and comments. The authors acknowledge support provided by the National Science Foundation under NSF Award No. No. OAC-2411068 and No. PHY-2409726. We acknowledge the Texas Advanced Computing Center (TACC) at the University of Texas at Austin for providing HPC resources on Frontera via allocation PHY22041. This work used Delta at NCSA through allocation PHY260102 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services and Support (ACCESS) program, which is supported by U.S. National Science Foundation grants No. 2138259, 2138286, 2138307, 2137603, and 2138296 32 . This research used resources provided by the Delta research computing project, which is supported by the NSF Award No. OAC-2005572 and the State of Illinois. This research was supported in part by the Illinois Computes project which is supported by the University of Illinois Urbana-Champaign and the University of Illinois System. This work made use of the Illinois Campus Cluster, a computing resource that is operated by the Illinois Campus Cluster Program (ICCP) in conjunction with the National Center for Supercomputing Applications (NCSA) and which is supported by funds from the University of Illinois Urbana-Champaign. This work used the open-source softwares xTensor 143 ; 36 , the Einstein Toolkit 114 ; 94 ; 145 , Canuda 142 , and kuibit 34 . Appendix A Convergence Tests We perform a series of convergence tests to assess the validity and numerical accuracy of our simulations. In this aim, we rerun a subset of the simulations (ExptBp0355X0, CtrlBm350Xp07, and ExptBm300Xm07) with step sizes dx_{\rm low}=1\mathrm{M} , dx_{\rm med}=0.95\mathrm{M} , and dx_{\rm high}=0.85\mathrm{M} on the outermost refinement level (see Table 2 ). Within successive (inner) refinement levels, the step sizes are halved. Note that in the ExptBp0355X0 test we use a dimensionless coupling of \beta=0.36281 , rather than \beta=0.355 as in the production run. This is because the test was conducted as a preliminary check before the production runs. We expect the ExptBp0355X0 production run to have better convergence than the test as its dimensionless coupling is below the critical value for the scalarization of individual BHs. In the following, we show convergence plots and make error estimates for \phi_{\rm 00} , which is related to the scalar charge, and \phi_{\rm 22} , which is related to the scalar radiation. Both quantities are evaluated at an extraction radius of r_{\rm ex}=100\mathrm{M} . We use the same simulations to produce convergence plots and error estimates for the spin and other properties of the GR background in our previous work Ref. 88 , which we do not reproduce here. Run \beta c_{\rm 0} \chi_{\rm 0} |\bm{P}_{\rm i}|/\mathrm{M} \theta ExptBp0355X0 0.36281 1.0 0.0 0.245 0.0580 CtrlBm350Xp07 -3.50 0.1 0.7 0.490 0.05685 ExptBm300Xm07 -3.00 1.0 -0.7 0.490 0.05685 Table 2: Parameters for the suite of convergence tests. In each case, we run three simulations with step sizes dx_{\rm low}=1.0\mathrm{M} , dx_{\rm med}=0.95\mathrm{M} , and dx_{\rm high}=0.85\mathrm{M} at the outermost refinement level. The dx_{\rm low}=1.0\mathrm{M} simulations correspond to those in Table 1 . The dimensionless coupling of the ExptBp0355X0 test is larger than that of the production run. Here we list the run name, the dimensionless coupling, \beta , the initial scalar field amplitude, c_{\rm 0} , the BHs’ initial spin parameter, \chi_{\rm 0} , and the BHs’ initial momentum, |\bm{P}_{\rm i}| . Each system has a separation of d=100\mathrm{M} . As we increase the resolution of the simulations, we expect them to converge to the true solution at a rate determined by the convergence order, n . For a given quantity, q , the difference between the low and medium resolutions, q_{\rm low}-q_{\rm med} , should be proportional to the difference between the medium and high resolutions, q_{\rm med}-q_{\rm high} , times the convergence factor, Q_{n}(dx_{\rm low},dx_{\rm med},dx_{\rm high})=\frac{dx_{\rm low}^{n}-dx_{\rm med}^{n}}{dx_{\rm med}^{n}-dx_{\rm high}^{n}}\,. (19) In the simulations, we use fourth order finite differencing to evaluate spatial derivatives and a fourth order Runge-Kutta scheme to evolve in time. Therefore, we expect fourth order convergence and use the convergence factor Q_{4}(1.0,0.95,0.85)=0.634 . However, at the refinement boundaries, we use second order and fifth order interpolations in time and space, respectively. Therefore, the convergence order may be mixed. For comparison, the third and second order convergence factors are Q_{3}=0.586 and Q_{2}=0.542 , respectively. We compute the approximate percent error of a quantity by using the highest resolution simulation as a reference, \%\mathrm{Error}=100\,\left|\frac{q_{\rm low}-q_{\rm high}}{q_{\rm high}}\right|\,. (20) Whenever the high resolution quantity crosses zero, the percent error diverges. Neither the scalar charge, nor the scalar radiation are positive definite, so we must take this into consideration in both cases. The scalar charge changes sign on rare occasions in the midst of encounters and mergers. We thus comment on its typical error outside of these isolated incidents. The scalar radiation, however, undergoes frequent oscillations. We thus report the error found at the maxima of the radiation. A.1 Zoom-Whirl Figure 21: Convergence plot of the scalar charge in the ExptBp0355X0 run, which corresponds to a zoom-whirl and merger. Top: Convergence centered on the first encounter between the BHs. Bottom: Convergence centered on the BH merger. The simulation times are shifted by t_{\rm S} , which is defined such that their times of merger coincide (see text). Figure 22: Convergence plot of the scalar radiation in the ExptBp0355X0 run, which corresponds to a zoom-whirl and merger. Top: Convergence centered on the first encounter between the BHs. Bottom: Convergence centered on the BH merger. The simulation times are shifted by t_{\rm S} , which is defined such that their times of merger coincide (see text). Figure 23: Convergence plot of the scalar field in the CtrlBm350Xp07 run, which corresponds to a scattering encounter between the BHs. Both panels are centered on the time of the close encounter. Top: Convergence of the scalar charge. Bottom: Convergence of the scalar radiation. We first present the convergence test for the ExptBp0355X0 run. In this simulation, the BHs undergo a single zoom-whirl before merging. Depending on the resolution, the separation of the BHs during this intermediate phase varies. The BH trajectories realign as they approach merger. However, this variation leads to a shift in the merger time. For this reason, we perform convergence tests in two blocks, where the first is focused on the close encounter, and the second is focused on the merger. In the second block, we shift the time by the value, t_{\rm S}=t_{\mathrm{merge,}\,dx}-t_{\mathrm{merge,\,}dx_{\rm low}} , in order to align the data with the time of merger in the dx_{\rm low}=1\mathrm{M} simulation. The convergence of the scalar charge ( \sim\phi_{\rm 00} ) is depicted in Fig. 21 . The convergence during the first encounter is displayed in the top panel. While the trends are somewhat different, the difference between the scalar charges is of order 10^{-4} . The convergence during the merger is displayed in the bottom panel. We find fourth order convergence. We find percent errors of <3\% across both intervals. The only exception being a brief period following the merger when the scalar charge changes sign. The convergence of the scalar radiation ( \sim\phi_{\rm 22} ) is displayed in Fig. 22 . The convergence during the first encounter is displayed in the top panel. The convergence during the merger is displayed in the bottom panel. We find fourth order convergence. The percent error evaluated at the waveform peak is approximately 0.35\% during the first encounter and about 2.34\% during the merger. Figure 24: Convergence plot of the scalar field in the ExptBm300Xm07 run, which corresponds to a scattering encounter between the BHs. Both panels are centered on the time of the close encounter. Top: Convergence of the scalar charge. The simulation times are shifted by t_{\rm S} , which is defined such that the maxima of their scalar charges coincide (see text). Bottom: Convergence of the scalar radiation. A.2 Scattering black holes with spin Here we present the convergence tests for the CtrlBm350Xp07 and ExptBm350Xm07 runs, which each result in scattering. Therefore, we center the convergence tests on the close encounter between the BHs. The convergence test for the CtrlBm350Xp07 run is shown in Fig. 23 , with the scalar charge ( \sim\phi_{\rm 00} ) in the top panel and the scalar radiation ( \sim\phi_{\rm 22} ) in the bottom panel. We find fourth order convergence. Prior to and during the encounter we find a percent error of <20\% for the scalar charge and a percent error of <30\% for the scalar radiation. After the encounter, both quantities grow exponentially as the system scalarizes. At different resolutions the scalar field grows with somewhat differing time constants, which causes the percent error to increase with time. However, we are principally concerned with the qualitative, rather than the quantitative, behavior. The convergence test for the ExptBm350Xm07 run is shown in Fig. 24 , with the scalar charge \sim(\phi_{\rm 00}) in the top panel and the scalar radiation \sim(\phi_{\rm 22}) in the bottom panel. We find a slight phase shift in the scalar charge and adjust the simulation time by the shift time, t_{\rm S}=t_{\mathrm{peak,}\,dx}-t_{\mathrm{peak,\,}dx_{\rm low}} , to align the data with the time of the scalar charge maximum in the dx_{\rm low}=1\mathrm{M} simulation. We find fourth order convergence. The percent error of the scalar charge is <10\% prior to the encounter and <40\% after the encounter. The percent error evaluated at the waveform peak of the scalar radiation is about 20.2\% . As we are primarily interested in the qualitative behavior of the scalar field, this is within reason.