| \documentclass[10pt,aps,prl, nofootinbib,superscriptaddress,reprint,tightenlines]{revtex4-1} |
| \usepackage{amssymb,amsmath,accents,mathrsfs} |
| \usepackage{verbatim,graphics,graphicx,color,slashed,textcomp,bbm,mathdots,multirow,array} |
| \usepackage{aas_macros} |
| |
| \usepackage{graphicx} |
| \usepackage{color,units} |
| \usepackage{xcolor} |
| \usepackage{hyperref} |
| \usepackage{bm} |
| \usepackage[title]{appendix} |
| \usepackage{listings} |
| \usepackage{soul} |
| \usepackage{tabularx} |
| \usepackage[normalem]{ulem} |
| \hypersetup{ |
| bookmarks=true, |
| unicode=false, |
| pdftoolbar=true, |
| pdfmenubar=true, |
| pdffitwindow=false, |
| pdfstartview={FitH}, |
| pdfauthor={} |
| colorlinks=true, |
| linkcolor=blue, |
| citecolor=red, |
| urlcolor=blue, |
| } |
| \graphicspath{{}} |
| \def\rem#1{{\bf\textcolor{red}{$\diamondsuit$ #1 $\diamondsuit$}}} |
|
|
| \newcommand{\ZX}[2]{{\color{red}\sout{#1}\color{brown}{#2}}} |
| \newcommand{\CY}[2]{{\color{purple}\sout{#1}\color{purple}{#2}}} |
| \newcommand{\LS}[1]{{\color{blue} [\bf #1]}} |
|
|
| \begin{document} |
|
|
| \definecolor{dkgreen}{rgb}{0,0.6,0} |
| \definecolor{gray}{rgb}{0.5,0.5,0.5} |
| \definecolor{mauve}{rgb}{0.58,0,0.82} |
|
|
| \lstset{frame=tb, |
| language=Matlab, |
| aboveskip=3mm, |
| belowskip=3mm, |
| showstringspaces=false, |
| columns=flexible, |
| basicstyle={\small\ttfamily}, |
| numbers=none, |
| numberstyle=\tiny\color{gray}, |
| keywordstyle=\color{blue}, |
| commentstyle=\color{dkgreen}, |
| stringstyle=\color{mauve}, |
| breaklines=true, |
| breakatwhitespace=true |
| tabsize=3 |
| } |
|
|
| \title{Detecting ultralight dark matter in the Galactic Center with pulsars around Sgr~A*} |
|
|
| \author{Jiang-Chuan Yu} |
| \affiliation{Department of Astronomy, School of Physics, Peking University, Beijing 100871, China} |
| \affiliation{Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China} |
| \author{Yan Cao} |
| \affiliation{School of Physics, Nanjing University, Nanjing 210093, China} |
| \author{Zexin Hu} |
| \affiliation{Department of Astronomy, School of Physics, Peking University, Beijing 100871, China} |
| \affiliation{Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China} |
| \author{Lijing Shao} |
| \email{lshao@pku.edu.cn} |
| \affiliation{Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China} |
| \affiliation{National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China} |
| \date{\today} |
|
|
| \begin{abstract} |
| Ultralight dark matter (ULDM) model is a leading dark matter candidate that arises naturally in extensions of the Standard Model. In the Galactic Center, ULDM manifests as dense hydrogen-like boson clouds or self-gravitating soliton cores. We present the first study of the gravitational effects of these ULDM structures on pulsar orbits around Sgr~A*, using pulsar timing as a precision dynamical probe, based on a comprehensive and practical framework that includes various kinds of black hole and orbital parameters. |
| Our analysis shows that long-term pulsar monitoring—one of the key objectives of future SKA science—could detect a boson cloud with a total mass as low as $\mathcal{O}(M_\odot)$ for boson mass $m \sim 10^{-18}\,\mathrm{eV}$, and probe a wide range of soliton core masses in the lower-mass regime, assuming a conservative timing precision of $\sigma_{\mathrm{TOA}}=1\,\mathrm{ms}$. |
|
|
| |
|
|
| |
| |
| \end{abstract} |
|
|
| \maketitle |
|
|
| \textit{Introduction.}---Ultralight bosons with mass $m \lesssim 10^{-6}\,\mathrm{eV}$ have emerged as compelling dark matter candidates, since they naturally alleviate challenges on the galactic scale in the traditional cold dark matter paradigm~\cite{Hu:2000ke,DelPopolo:2016emo,Hui:2016ltb,Ferreira:2020fam,Singh:2025uvp} and are predicted by various theories beyond the Standard Model~\cite{Peccei:1977hh,Weinberg:1977ma,Wilczek:1977pj,Hui:2016ltb,Marsh:2015xka,Agrawal:2021dbo,Polchinski:1998rr,Kim:1986ax,Fayet:1990wx,Svrcek:2006yi,Arvanitaki_2010,Arias:2012az,Lyth:1998xn,PhysRevD.93.103520,Ema:2019yrd,Ahmed:2020fhc,Kolb:2020fwh}. Independent constraints on ultralight dark matter (ULDM) arise from both cosmological and local measurements. Stellar kinematics in dwarf galaxies~\cite{Bar:2018acw,Marsh:2018zyw,Pozo:2023zmx,Pozo:2020ukk,PhysRevLett.134.151001,may2025updatedboundsultralightdark} and other substructures~\cite{Church:2018sro,Amorisco:2018dcn} favors solitonic cores formed by ULDM with boson mass $m \gtrsim 10^{-22}\,\mathrm{eV}$. Observations of the Lyman-$\alpha$ forest~\cite{Armengaud:2017nkf,PhysRevLett.119.031302,Zhang:2017chj,Kobayashi:2017jcf} and 21\,cm absorption line~\cite{Schneider:2018xba,Lidz:2018fqo} further limit the particle mass to $m \gtrsim 10^{-21}\,\mathrm{eV}$. Precision experiments, such as pulsar timing observations~\cite{Khmelnitsky:2013lxt,Porayko:2014rfa,Nomura:2019cvc,Sun_2022,Dror:2025nvg} and laser interferometers~\cite{Aoki:2016kwl,Kim:2023pkx,PhysRevD.110.023025} can also constrain oscillating ULDM fields via their coupling to the metric. Specifically, pulsar timing array experiments~\cite{EuropeanPulsarTimingArray:2023egv} showed that only ULDM with $m \gtrsim 10^{-23}\,\mathrm{eV}$ can constitute all of the local dark matter density. |
|
|
| |
|
|
| |
|
|
| Our Galactic Center (GC) provides a natural laboratory to probe the properties of dark matter~\cite{Gondolo_1999,Sadeghian:2013laa,Hu:2023ubk,Cheng:2024mgl,GRAVITY:2019tuf,Yuan:2022nmu,GRAVITY:2023cjt,GRAVITY:2023azi,Bar:2019pnz,Lacroix:2018zmg,Chan_2022,Shen:2023kkm,Zakharov:2007fj,Heissel:2021pcw}, as dense dark matter structure can form around the supermassive black hole (SMBH). Stellar orbital motion, especially that of S2, has been widely used to constrain the dark matter distribution near Sgr~A*~\cite{Lacroix:2018zmg,Shen:2023kkm,Zakharov:2007fj,Heissel:2021pcw}. Bound structures of ULDM, such as superradiant cloud~\cite{PhysRevD.22.2323,Cardoso:2005vk,PhysRevD.76.084001,Arvanitaki:2010sy,PhysRevD.87.043513,Brito:2014wla,Brito:2015oca,Baryakhtar:2017ngi,Frolov:2018ezx,Dolan:2018dqv,Brito_2020,B2,B3,Siemonsen:2022yyf} and soliton core~\cite{Schive:2014dra,Schive:2014hza,Chavanis:2019bnu,Bar:2019pnz,Davies_2020,Annulli_2020,Zagorac_2023,Aghaie_2024,liao2025decipheringsolitonhalorelationfuzzy}, may exist in the GC. Recent studies~\cite{GRAVITY:2019tuf,Yuan:2022nmu,Chen_2023,GRAVITY:2023cjt,GRAVITY:2023azi,Chen_2025,bai2025probingaxionsspectroscopicmeasurements,tomaselli2025probingdenseenvironmentssgr,Bar:2019pnz,Chan_2022} have investigated their observational signatures and constrained the total mass of such structures. |
|
|
| Pulsars are rapidly rotating neutron stars whose remarkably stable spin periods make them exquisite clocks for probing gravitational dynamics. Through precisely measuring times of arrival (TOAs) of their pulses, a pulsar in a tight orbit around Sgr~A*, potentially discoverable with next-generation instruments such as the Square Kilometre Array (SKA), would offer an unique opportunity to test gravity in the strong-field regime~\cite{Wex:1998wt,Liu:2011ae,Psaltis:2015uza,Zhang:2017qbb,Bower:2018mta,Dong:2022zvh,Hu:2023ubk,Hu:2023vsg,Hu:2024blq,Shao:2025vmb} and explore the dark matter properties in the Galactic Center~\cite{DeMartino:2017qsa,Hu:2023ubk}. |
|
|
| |
|
|
| In this {\it Letter}, we assess the detectability of nonrelativistic ULDM structures in the GC, including the gravitational atom (GA) and self-gravitating spherical soliton, through the timing observation of a pulsar with a realistic orbital period $P_b\in [0.5,5]\,\mathrm{yr}$. We develop a post-Newtonian timing framework for the pulsar–SMBH binary incorporating the leading-order gravitational perturbation induced by the ULDM struture and estimate the sensitivity of long-term observations to its total mass. Throughout this {\it Letter}, we adopt the natural units where $\hbar=G=c=1$. |
|
|
| \textit{ULDM in the GC.}---If the boson is sufficiently light, the ULDM environment of Sgr~A* locating far away from the BH horizon can be modeled as a nonrelativistic structure. In this limit, the slow mode of a massive spin-$s$ bosonic field is described by a rank-$s$ tensor wavefunction $\psi_I$ in the Cartesian basis, where $I$ refers schematically to the set of spatial tensor indices (e.g., $\psi$ for scalar, $\psi_i$ for vector) \cite{Baryakhtar:2017ngi,Brito_2020,Jain:2021pnk,Cao:2023fyv,Cao:2024wby} . Neglecting the possible non-gravitational self-interaction, the dynamics of $\psi_I$ around a static point mass $M$ is governed by the Schr\"odinger-Poisson (SP) equation, |
| \begin{align} |
| i\partial_t\psi_I &= -\frac{1}{2m}\nabla^2\psi_I + m\left(\Phi - \frac{M}{r} \right)\psi_I, \\ |
| \nabla^2\Phi &= 4\pi \rho,\quad \rho = m \sum_I |\psi_I|^2, |
| \end{align} |
| where $\rho$ and $\Phi$ are the mass density and Newtonian potential of the wavefunction respectively. We focus on two classes of bound states: the GA for which $\Phi$ is negligible, and the spherically symmetric ground state referred to as \textit{spherical soliton}. |
|
|
| The total mass of the bound state is given by $ |
| M_\mathrm{c} = \int d^3r\, m \sum_I |\psi_I|^2 \equiv \beta M$. If the mass ratio $\beta \ll 1$, self-gravity of the wavefunction can be neglected at the leading order. Setting $\Phi=0$, the SP equation reduces to the Shr\" {o}dinger equation of a hydrogen atom, with the gravitational fine-structure constant $\alpha\equiv m M\approx 0.032(\mu/10^{-18}\text{eV})(M/4.3\times 10^6M_\odot)$ and Bohr radius $r_\text{c}\equiv M/\alpha^2$. Consequently, $\psi_I$ is spanned by the hydrogenic bound states $\psi^{(nl\texttt{m})}\propto R_{nl}(r)\,Y_{l\texttt{m}}(\theta,\phi)$ for scalar field, labeled by the quantum numbers $|nl\texttt{m}\rangle$; in the case of vector field, the angular wavefunction is spanned by the pure-orbital vector spherical harmonics, $\mathbf{Y}_{lj\texttt{m}}(\theta,\phi)$, and an eigenstate can thus be labeled by $|nlj\texttt{m}\rangle$ (see \cite{SupplementaryMaterials} for details). For $\alpha \ll 1$, since $r_\text{c}\gg M$, this GA model provides an effective description for the superradiant cloud (quasibound states) around a spinning BH at radius $r\gg M$, with the $z$-axis aligned with the BH’s spin direction. The rotational superradiant instability varies between different states \cite{B2,B3}. In this work, we consider the GA occupied by the fastest growing state for an initial BH spin $\chi_i\sim 1$ (the so-called superradiant ground state), which is the $|211\rangle$ state for scalar field, the $|1011\rangle$ state for vector field, and the $|1022\rangle$ state for spin-2 field. The density profiles of the latter two are degenerate with that of the scalar $|100\rangle$ state. Neglecting the baryonic accretion, the saturated cloud mass is given by $\beta_\text{max}\approx \alpha \chi_i/\texttt{m}$, thus $\beta\ll 1$ is automatically satisfied. |
|
|
| The GA approximation breaks down when $\beta$ becomes sufficiently large, which is typical for a soliton core; in this regime, the bound state should be computed nonperturbatively. We consider the spherically symmetric ground state of the SP equation in the form $\psi(t,r)=f(r)\,e^{-iE t}$, which describes a nodeless self-gravitating configuration around the central mass. Due to a scaling symmetry, the SP equation for given $\alpha$ admits solutions of a one-parameter family~\cite{SupplementaryMaterials} in the dimensionless coordinate $y\equiv m r$, labeled by $\kappa^2\equiv \sqrt{{8\pi}/{m}}\,f(0)$. For $\beta=0$, these solutions reduce to the $|100\rangle$ state of the scalar GA, and for a moderately small $\beta$, the ground state is still well-described by a $|100\rangle$ state corrected by its own Newtonian potential, with the energy level $E = -(m\alpha^2/2)\,C(\beta)$ and $C(\beta)\approx 1+5\beta/4$. As the soliton mass increases, the configurations become more compact, and relativistic corrections are correspondingly more important \cite{Salehian_2021,zhang2025unifiedviewscalarvector}. We thus assume that $r_{0.5}\ge M+M_\text{c}$, with $r_{0.5}$ being the radius enclosing 50\% of the total soliton mass. This condition turns out to be well satisfied in the parameter space of interests. |
|
|
| The normalized density profile depends only on $\beta$ and $r/r_\text{c}$, hence for given $\beta$, a larger boson mass results in a more compact soliton. This is reflected in the flat density profile of the inner region~\cite{SupplementaryMaterials}: $M_\text{c}(r)/M\approx (4\pi/3)\,\rho(0)\, r^3/M=m^6\,D(\beta)\,(Mr)^3/6$, with $D(\beta)\approx 0.11 \beta^4$ for $\beta \gtrsim 100$, as depicted in Fig.~\ref{soliton_mass_profile}. For a large soliton mass, the enclosed mass within $r=100M$ in the GC is therefore $M_\text{c}(r)/M_\text{c}\approx 1.8\times 10^4\,\beta^3\alpha^6\approx 0.02\,(\mu/10^{-19}\text{eV})^6(\beta/1000)^3$, thus most of the soliton mass would indeed locate far away from the BH if $\beta \lesssim 1000\,(\mu/10^{-19}\text{eV})^{-2}$, in which case the accretion of the soliton onto the BH is suppressed~\cite{Annulli_2020}. |
|
|
| \textit{Forecast of sensitivities.}---To quantify the sensitivity of a single pulsar to the aforementioned ULDM structures in the GC, we simulate the timing signals using the numerical framework developed in Ref.~\cite{Hu:2023ubk}. The orbital evolution is obtained by integrating the post-Newtonian equations of motion, with the DM-induced acceleration included as $\ddot{\mathbf{r}}_{\text{DM}}\approx -\nabla \Phi$, since the dissipative effects and relativistic corrections from ULDM are negligible \cite{Cao:2024wby}. |
|
|
| |
|
|
| The model parameters are grouped into three sectors: the black hole parameters $\Theta_{\text{BH}} = \{M, \chi, q, \lambda, \eta\}$, describing the mass, spin, quadrupole moment, and spin orientation; the pulsar orbital and spin parameters $\Theta_{\text{PSR}} = \{P_b, e, \omega, i, \theta_0, N_0, \nu, \dot{\nu}\}$, where $\theta_0$ is the initial orbital phase and ${N_0,\nu,\dot \nu}$ is associated to the rotation number $N(T)=N_0+\nu T+\dot \nu T^2/2$ as function of proper time $T$; and the ULDM parameters $\Theta_{\text{DM}} = \{\beta\}$. For simplicity, we fix the unobservable longitude of the ascending node to $\Omega=0$. Timing residuals are computed including the Römer, Shapiro, and Einstein delays, and the effects due to the SMBH’s proper motion are also included. Additional numerical details are provided in \cite{SupplementaryMaterials}. |
|
|
| We employ the Fisher matrix formalism to estimate the projected constraints on ULDM parameters. The covariance matrix is given by $C_{\mu\nu} = \left(\partial^2 \mathcal{L}/\partial\Theta^\mu \partial\Theta^\nu\right)^{-1}$, where the log-likelihood takes the form |
| \begin{equation} |
| \mathcal{L} = \frac{1}{2\nu^2} \sum_{i=1}^{N_{\mathrm{TOA}}} \frac{\left[N_i(\Theta) - N_i(\bar{\Theta})\right]^2}{\sigma_{\mathrm{TOA}}^2}, |
| \end{equation} |
| with $N_i(\Theta)$ denoting the pulsar rotation number at the $i$-th TOA $t^{\mathrm{TOA}}_i$, and $\bar{\Theta}$ the true system parameters. We adopt a timing precision of $\sigma_{\mathrm{TOA}} = 1\,\mathrm{ms}$ and assume a total observation time of $T_{\mathrm{obs}} = 5\,\mathrm{yr}$, consistent with the projected capabilities of the Square Kilometre Array (SKA) and next-generation Very Large Array (ngVLA). |
|
|
| \begin{figure} |
| \includegraphics[scale=0.25]{soliton_mass_profile.png} |
| \caption{Enclosed mass of the spherical soliton in the GC for various boson masses and soliton masses. The inner profiles are well approximated by $m^6\,D(\beta)\,(Mr)^3/6$, shown as dashed lines. The semi-major axes of the considered pulsar orbits with given periods are indicated by the gray vertical lines. The mass of Sgr~A* is $M=4.3\times 10^6M_\odot$.} |
| \label{soliton_mass_profile} |
| \end{figure} |
|
|
| \begin{figure}[htbp] |
| \centering |
| \includegraphics[scale=0.25]{sum_sca.png}\\[0.3cm] |
| \includegraphics[scale=0.25]{sum_vec.png} |
| \caption{Projected sensitivities to the mass ratio $\beta$ versus $\alpha$ for scalar GA in the $|211\rangle$ state (top) and spherical soliton (bottom). The result for the soliton also applies to the vector or spin-2 GA in its superradiant ground state for $\beta<\alpha$. |
| |
| Red region is excluded by Schwarzschild precession of S2 star and |
| the green region is excluded by stellar dynamics of the clockwise rotating disk~\cite{Beloborodov:2006is}. |
| The blue dashed line denotes the sensitivities in case of timing precision $\sigma_{\text{TOA}}=10\, \mu\text{s}$. |
| The gray dashed line marks the boundary $\beta = \alpha$, corresponding to the threshold of superradiance. |
| Below the black dashed line, the soliton satisfies $r_{0.5}>M+M_\text{c}$. |
| The brown dashed line shows the soliton mass given by the soliton-halo relation for our galaxy~\cite{Schive:2014dra,Schive:2014hza}. |
| } |
| \label{fig:combined sensitivities} |
| \end{figure} |
|
|
| |
|
|
| Figure~\ref{fig:combined sensitivities} presents the projected sensitivities to the mass ratio of the cloud or soliton core to the central black hole, $\beta$, for two benchmark profiles: the scalar $|211\rangle$ GA and the spherical soliton. Estimated constraints~\cite{SupplementaryMaterials} from the periastron precession of the S2 star~\cite{GRAVITY:2020gka} and the stellar dynamics of the clockwise rotating disk (CWD)~\cite{Beloborodov:2006is} are shown for comparison. Note that the $3\sigma$ constraints obtained by the GRAVITY collaboration \cite{GRAVITY:2023cjt,GRAVITY:2023azi}, $\beta_\text{211}\lesssim 10^{-3}$ for $\alpha\in(0.015,0.045)$ and $\beta_\text{1011} \lesssim 10^{-3}$ for $m\in(10^{-19},10^{-18})\,\text{eV}$, are slightly stronger than our $1\sigma$ estimations in the corresponding mass range. The brown dashed line shows the soliton mass given by the halo–soliton relation, which is extrapolated from the dark-matter-only simulations~\cite{Schive:2014dra,Schive:2014hza} in the case of scalar ULDM. |
|
|
| |
|
|
| |
|
|
| For the scalar $|211\rangle$ configuration, a pulsar on a 0.5-year orbit yields new sensitivity on the mass ratio $\beta \sim 10^{-7}$ in the range $10^{-2} \lesssim \alpha \lesssim 10^{-1}$, improving upon S2 constraints by up to five orders of magnitude. Notably, the orientation of the orbit relative to the BH's spin also introduces angular dependence, leading to an order-of-magnitude variation in the |
| sensitivities; more details can be found in~\cite{SupplementaryMaterials}. For the spherical soliton, we find that a 0.5-yr pulsar orbit will place the strongest constraints on $\beta$ at $\alpha \gtrsim 10^{-3}$. The projected sensitivity extends to $\beta \sim 10^{-6}$ at high masses, improving significantly over current bounds, typically much helpful to search for superradiantly generated GAs ($\beta \lesssim \alpha$). |
| |
| |
| We can see that, in both ULDM scenarios, pulsars with long orbital periods tend to exhibit better sensitivities at smaller $\alpha$, whereas those with relatively short orbital periods tend to have better sensitivities at larger $\alpha$. For pulsars with the same period, those with a higher eccentricity perform better in the low-mass range. Importantly, our method can probe ULDM down to $\alpha \gtrsim 3\times10^{-4}$, surpassing the soliton mass scale inferred from the empirical halo–soliton relation. Moreover, this relation—extrapolated from galactic simulations—has been shown to overestimate the soliton mass by orders of magnitude for $m \gtrsim 10^{-21}$\,eV~\cite{Bar:2019pnz}, further highlighting the advantage of pulsar-based timing over traditional stellar dynamics in the GC. Moreover, if the timing precision reaches $\sigma_{\text{TOA}}=10 \mu$s in the future, the corresponding sensitivities would be 100 times better, as shown in Fig.~\ref{fig:combined sensitivities}. |
|
|
| |
|
|
| |
|
|
|
|
|
|
| The size of GA (with $\beta\ll 1$) is characterized by the Bohr radius $r_\text{c}=M/\alpha^2$. As a result, when the pulsar is within the Bohr radius, the enclosed mass and hence the gravitational influence of GA roughly increases with the size of the orbit. For a given pulsar orbit, this leads to enhanced sensitivity with increasing $\alpha$. The situation changes for orbits far outside the Bohr radius, due to an exponential decay in mass density with radius. The size of the spherical soliton for a given $\beta$ is also controlled by the Bohr radius. Although parametrically smaller than that of GA with the same $\alpha$, in the small-$\alpha$ regime the soliton size remains much larger than the considered pulsar orbits for the marginally sensitive values of $\beta$ in Fig.~\ref{fig:combined sensitivities}, leading to a decrease in sensitivity with decreasing $\alpha$. |
|
|
| In practice, the result of parameter estimation is also affected by other factors. For example, the presence of the black hole's spin and quadrupole moment can partially degenerate with the dark matter effect, degrading the sensitivities on $\beta$ by $\sim \mathcal{O}(1)-\mathcal{O}(10)$. However, our results show no significant dependence on the assumed spin magnitude of the black hole. |
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| \textit{Summary and discussions.}---We investigate the potential of using a pulsar near the GC to probe ULDM through pulsar timing observations. Assuming a pulsar with an orbital period \( P_b \sim 0.5\,\mathrm{yr} \), eccentricity \( e \sim 0.8 \), and timing precision $\sigma_{\text{TOA}}=1$ ms over a 5-yr observation with weekly TOA measurements, simulations show that with ULDM masses in the range \( 3\times10^{-19}\,\mathrm{eV} < m < 3\times10^{-18}\,\mathrm{eV} \), the mass ratio of the cloud or soliton core to the central black hole $\beta$ can be constrained to amplitudes of order \( \mathcal{O}(10^{-7}\text{--}10^{-6}) \), improving upon existing limits from S2 star at least by a few orders of magnitude. In the case of a spherical soliton, a pulsar with $P_b = 5\,{\rm yr}$ provides excellent sensitivity in the range of $5 \times 10^{-4} \lesssim \alpha \lesssim 10^{-2}$, significantly exceeding the total mass predicted by the empirical halo--soliton relation. These measurements improve proportionally if better timing precision is achieved. |
|
|
|
|
| |
| |
|
|
| In the mass window $0.066\lesssim \alpha \lesssim 0.16$ (scalar) and $0.015\lesssim \alpha \lesssim 0.039$ (vector), the timescale of the superradiant growth of GA for an initial BH spin $\chi_i\sim 1$ is shorter than the characteristic merger time ($\sim 10^{6}\,\mathrm{yr}$), while the saturated GA depletes via gravitational radiation on a timescale $\gtrsim 10^{8}\,\mathrm{yr}$. In the absence of other processes, the cloud would thus remain close to its saturated state (with $\beta \sim \alpha$), and future pulsar timing observations could either reveal its presence or place stringent bounds on the boson mass. In the more general scenario, the GA can be occupied by multiple states \cite{Ficarra_2019,Siemonsen_2020,Guo_2023,guo2025effectaccretionscalarsuperradiant} (see also \cite{SupplementaryMaterials}) and the soliton core will not be in a perfectly spherically symmetric ground state \cite{Zagorac_2022,Glennon_2023,salasnich2025collectiveexcitationsselfgravitatingultralight}. In such cases, the interference in the density profile can lead to oscillating gravitational potential and possibly distinctive orbital signatures. Additionally, relativistic effects would become important if the ULDM structure is sufficiently compact or resides closer to the BH horizon. These issues deserve further investigation. |
|
|
| |
|
|
| We have not taken into account the corrections to the Shapiro and Einstein delays due to the gravitational potential of the ULDM structure, which are estimated to be negligible compared with the correction to the Römer delay. According to the model of \cite{Khmelnitsky:2013lxt,DeMartino:2017qsa}, the metric perturbation sourced by the fast mode of the bosonic field gives a timing residue of order $\rho/m^3$, this is much smaller than $\sigma_{\text{TOA}} = 1 \, \text{ms}$ in the interested parameter space \cite{SupplementaryMaterials} and consequently can also be neglected. |
|
|
|
|
| Finally, throughout this work, we have focused on the gravitational signature of ULDM in pulsar orbits near Sgr~A$^*$, while neglecting environmental perturbations such as stellar clusters~\cite{PhysRevD.81.062002,Liu_2012}, gas dynamics, and other external gravitational fields, as well as their impact on the ULDM structure~\cite{Du_2022}. These environmental effects are expected to be subdominant for tight orbits and, while they may introduce noise or systematics in pulsar timing, a full assessment of their impact on ULDM sensitivity requires more realistic astrophysical modeling. |
|
|
| \begin{acknowledgments} |
| This work was supported |
| by the National Natural Science Foundation of China (12573042), the National SKA Program of China (2020SKA0120300), the Beijing Natural |
| Science Foundation (1242018), |
| the Max Planck Partner Group Program funded by the |
| Max Planck Society, and the High-Performance Computing Platform of Peking |
| University. |
| \end{acknowledgments} |
|
|
| |
|
|
| |
|
|
| |
|
|
|
|
| |
|
|
| |
| \bibliography{paper}{} |
| \bibliographystyle{unsrtnat} |
| |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| \clearpage |
|
|
| \onecolumngrid |
|
|
| \setcounter{section}{0} |
| \renewcommand{\thesection}{\arabic{section}} |
| \setcounter{equation}{0} |
| \renewcommand{\theequation}{S\arabic{equation}} |
| \setcounter{figure}{0} |
| \renewcommand{\thefigure}{S\arabic{figure}} |
| \setcounter{table}{0} |
| \renewcommand{\thetable}{S\arabic{table}} |
| \begin{center} |
| \Large\bfseries Supplementary Materials |
| \end{center} |
| |
|
|
| \section{Newtonian gravitational atom and spherical soliton}\label{sec2} |
| In the nonrelativistic (NR) limit, the slow mode of a massive spin-$s$ field can be described by a rank-$s$ tensor wavefunction $\psi_I$ in the Cartesian basis, here $I$ refers schematically to the set of tensor indices. E.g., for the scalar field, $\psi_I=\psi$; for the vector field, $I=i$; for the spin-2 filed, $I=ij$. Neglecting the possible non-gravitational self-interactions, the equation of motion of the wavefunction around a static point mass $M$ is given by the Shr\" {o}dinger-Poisson (SP) equation: |
| \begin{align} |
| i\partial_t\psi_I &=-\frac{1}{2m}\nabla^2\psi_I+m\left(\Phi-\frac{M}{r}\right) \psi_I, |
| \\ |
| \nabla^2\Phi &=4\pi \rho,\quad \rho =m \sum_I|\psi_I|^2, |
| \end{align} |
| where $m$ is the boson mass, $\Phi$ is the Newtonian potential sourced by the wavefunction, corresponding to the weakly perturbed metric $ds^2=\left[-1-2\left(\Phi-\frac{M}{r}\right)\right]dt^2+\left[1-2\left(\Phi-\frac{M}{r}\right)\right]|d\mathbf{x}|^2$. In the following, we consider the bound state solution of this equation in the limiting cases of gravitational atom and spherically symmetric soliton. We denote the total mass of the bound state by $M_\text{c}=\int d^3r\,m\sum_I|\psi_I|^2\equiv \beta M$. |
|
|
| \subsection{Gravitational atom limit} |
| If $\beta \ll 1$, the self-gravity of the wavefunction can be neglected at the leading order. Setting $\Phi=0$, the SP equation reduces to the Shr\" {o}dinger equation of hydrogen atom: |
| \begin{equation} |
| i\partial_t\psi_I=-\frac{1}{2m}\nabla^2\psi_I-\frac{\alpha}{r} \psi_I, |
| \end{equation} |
| where $\alpha\equiv m M$ is the gravitational fine-structure constant. For convenience, we introduce the Bohr radius $r_\text{c}\equiv M/\alpha^2$ and the dimensionless radial coordinate $x\equiv r/r_\text{c}=\alpha m r$. For each Cartesian component $I$, a general bound state can be written as |
| \begin{align}\label{general_bound} |
| \psi_I(t,r,\theta,\phi) &=\sum_{n\ge 1,\, |
| 0\le l\le n-1, \,|\mathtt{m}|\le l}c_{nl\mathtt{m}}\, \psi^{(nl\texttt{m})}, |
| \\ |
| \psi^{(nl\texttt{m})} &=r_\text{c}^{-3/2}R_{nl}(x)\,Y_{l\mathtt{m}}(\theta,\phi)\,e^{-i E^{(n)} t}, |
| \end{align} |
| with $c_{nl\mathtt{m}}\in\mathbb{C}$, $E^{(n)} =-\frac{m\alpha^2}{2n^2}$, $Y_{l\mathtt{m}}(\theta,\phi)$ being the spherical harmonics, and |
| \begin{equation} |
| R_{n l}(x)= \sqrt{\left(\frac{2}{n}\right)^3 \frac{(n-l-1) !}{2 n(n+l) !}}\left(\frac{2 x}{n}\right)^{l}e^{-\frac{x}{n}}\,L_{n-l-1}^{2 l+1}\left(\frac{2 x}{n}\right), |
| \end{equation} |
| where $\theta$ and $\phi$ are the angles written in the BH's coordinate system, $L^p_k(z)$ is the associated Laguerre polynomial. |
|
|
| For $\alpha \ll 1$, since $r_\text{c}\gg M$, this gravitational atom (GA) model provides an effective description for the superradiant cloud (quasi-bound states) around a spinning BH at radius $r\gg M$, if the $z$-axis is identified with the BH's spin direction. In the case of vector field, the NR limit of the quasibound states turns out to be~\cite{Baryakhtar:2017ngi,B2,B3} |
| \begin{equation} |
| \psi_i(t,r,\theta,\phi)=\sum_{n\ge 1,\,0\le l\le n-1,\,|l-1|\le j\le l+1,\,|\texttt{m}|\le j} c_{nlj\texttt{m}}\,\psi_i^{(nlj\texttt{m})}. |
| \end{equation} |
| with |
| \begin{equation} |
| \psi_i^{(nlj\texttt{m})}=r_\text{c}^{-3/2}R_{nl}(x)\,Y^i_{lj\texttt{m}}(\theta,\phi)\,e^{-iE^{(n)}t}, |
| \end{equation} |
| where the pure-orbital vector spherical harmonics\footnote{For the massive spin-2 field, the angular wavefunction is replaced by the pure-orbital tensor spherical harmonics~\cite{Brito_2020,RevModPhys.52.299} $Y_{lj\texttt{m}}^{ik} |
| =\sum_{|m_s|\le 2}Y^{(m_s)}_{lj\texttt{m}}\,\xi_{ik}^{m_s} $, with $Y^{(m_s)}_{lj\texttt{m}}=\left\langle 2, m_s;l, \texttt{m}-m_s \mid j, \texttt{m}\right\rangle Y_{l,\texttt{m}-m_s}(\theta, \phi)$ and $\xi_{ik}^{m_s}=\sum_{|m_1|\le 1, |m_2|\le 1}\langle 1, m_1;1, m_2 \mid 2, m_s\rangle \,\xi_i^{m_1}\xi_k^{m_2}$. For $l=0$, $j=2$, and $Y_{02\texttt{m}}^{ik}=Y_{02\texttt{m}}^{(\texttt{m})}\,\xi^\texttt{m}_{ik}$. For $\alpha\ll1$, the superradiant ground state of spin-2 GA is $|nlj\texttt{m}\rangle=|1022\rangle$~\cite{Brito_2020}. The mass density distribution of the latter is the same as that of $|100\rangle$ state of scalar GA.} $Y^i_{lj\texttt{m}}=\sum_{|m_s|\le 1}Y^{(m_s)}_{lj\texttt{m}}\,\xi_i^{m_s}$ (with $\boldsymbol{\xi}^0=\mathbf{e}_z$ and $\boldsymbol{\xi}^{\pm 1}=\mp (\mathbf{e}_x\pm i \mathbf{e}_y)/\sqrt{2}$) has spherical components: |
| \begin{equation} |
| Y^{(m_s)}_{lj\texttt{m}}(\theta,\phi) = \langle 1,m_s;l,\texttt{m}-m_s|j,\texttt{m}\rangle \,Y_{l,\texttt{m}-m_s}(\theta,\phi), |
| \end{equation} |
| with $\langle j_1,m_1;j_2,m_2|j,m\rangle$ being the Clebsch-Gordan coefficients. $Y_{lj\texttt{m}}^i$ is the simultaneous eigenfunction of $\hat L_z$ and $|\hat{\mathbf{L}}|^2$, with the eigenvalues $\texttt{m}$ and $l(l+1)$, respectively; here $\hat{\mathbf{L}}\equiv \mathbf{r}\times(-i\nabla)$ is the orbital angular momentum operator. Note that $\psi_i^{(nlj\texttt{m})}$ acquires a phase factor $(-1)^{l+1}$ under parity transformation, hence the $j=l\pm 1$ and $j=l$ modes are called electric and magnetic modes, respectively. For $l=0$, $j=1$, and $Y^i_{01\texttt{m}}=Y^{(\texttt{m})}_{01\texttt{m}}\,\xi_i^{\texttt{m}}$. |
|
|
| The phenomenology of GA and GA-companion system has been extensively studied~\cite{Baryakhtar:2017ngi,Brito_2020,Ferreira:2017pth,B1,B2,B3,B4,Tomaselli:2023ysb,Tomaselli:2024bdd,Boskovic:2024fga,Tomaselli:2025jfo,Brito_2023,Duque:2023cac,dyson2025environmentaleffectsextrememass,Takahashi:2021yhy,Takahashi:2023flk,Zhang:2018kib,ZJ,Guo:2024iye,Liu:2024mzw,Peng:2025zca,Guo:2025ckp,WY_1,WY_2,WY_3,Fan:2023jjj,Kavic:2019cgk,DeLuca:2021ite,Arana:2024kaz,Su:2021dwz,Cao:2023fyv,Cao:2024wby,DellaMonica:2025zby,Lyu:2025lue,Guo:2025pea}. In this work, we consider the GA occupied by the fastest growing state $|nl\texttt{m}\rangle=|211\rangle$ for initial BH spin $\chi\sim 1$ (so-called superradiant ground state) in the case of scalar field, and $|nlj\texttt{m}\rangle=|1011\rangle$ in the case of vector field~\cite{B3}. The mass density distribution of the latter is the same as that of $|100\rangle$ state of scalar GA. We have |
| \begin{align} |
| \rho_{211} &= m^2\beta \alpha^4\frac{x^2e^{-x}\sin^2\theta}{64\pi} |
| , |
| \\ |
| \rho_{100} &= m^2\beta \alpha^4\frac{e^{-2x}}{\pi} |
| . |
| \end{align} |
| Note that for $\beta\ll \alpha$, $\rho/m^3 \sim \beta \alpha^4/m \ll \alpha^5/m \approx \left(\frac{M}{4.3\times 10^6M_\odot}\right)\left(\frac{\alpha}{0.1}\right)^4\times 2.1\,\text{ms}$. |
|
|
| Using the general solution to the Poisson equation $\nabla^2\Phi=4\pi \rho$: |
| \begin{align} |
| \Phi(r,\theta,\phi)&=-\sum_{l\ge 0,\, |\mathtt{m}|\le l}\frac{4\pi}{2l+1}\left[\frac{q_{l\mathtt{m}}(r)}{r^{l+1}}+r^lp_{l\mathtt{m}}(r)\right]Y_{l\mathtt{m}}^*(\theta,\phi),\label{Poisson} |
| \\ |
| q_{l\mathtt{m}}(r)&=\int_0^rs^l\rho_{l\mathtt{m}}(s)s^2ds, |
| \\ |
| p_{l\mathtt{m}}(r)&=\int_r^\infty\frac{\rho_{l\mathtt{m}}(s)}{s^{l+1}}s^2ds, |
| \\ |
| \rho_{l\mathtt{m}}(r)&=\int_0^{2\pi}d\phi \int_0^\pi d\theta\,\sin\theta\int\rho(r,\theta,\phi)\, Y_{l\mathtt{m}}(\theta,\phi),\label{rho_sphere} |
| \end{align} |
| we obtain the Newtonian potential: |
| \begin{align} |
| \frac{\Phi_{211}}{\beta \alpha^2}=&-\frac{1}{x}-\frac{3}{x^3} |
| +\left(\frac{3}{x^3}+\frac{x^2}{16}+\frac{3}{x^2}+\frac{3 x}{8}+\frac{5}{2 x}+\frac{5}{4}\right) e^{-x}\nonumber |
| \\ |
| &+\left[\frac{9}{x^3}+e^{-x} \left(-\frac{9}{x^3}-\frac{x^2}{16}-\frac{9}{x^2}-\frac{3 x}{8}-\frac{9}{2 x}-\frac{3}{2}\right)\right] \cos ^2\theta |
| , |
| \\ |
| \frac{\Phi_{100}}{\beta \alpha^2}=& -\frac{1}{x}+e^{-2 x} \left(\frac{1}{x}+1\right)\label{Phi_1011} |
| . |
| \end{align} |
| The gravitational acceleration $\delta\mathbf{a}=-\nabla \Phi=-(\partial_r \Phi)\,\mathbf{e}_r-r^{-1}(\partial_\theta \Phi)\,\mathbf{e}_\theta$ is given by |
| \begin{align} |
| \frac{\delta \mathbf{a}_{211}}{\beta \alpha^4/M} =& \Big\{-\frac{1}{x^2}+\frac{e^{-x}}{16x^4}\Big[144-144e^x+144x+88x^2+40x^3+14x^4+4x^5+x^6\nonumber\\ |
| &+\left(-432+432e^x-432x-216x^2-72x^3-18x^4-4x^5-x^6\right)\cos^2\theta\Big]\Big\}\mathbf{e}_r\nonumber\\ |
| &+\frac{e^{-x}}{8x^4}\Big[\left(-144+144e^x-144x-72x^2-24x^3-6x^4-x^5\right)\cos \theta \sin \theta\Big]\mathbf{e}_{\theta}, |
| \\ |
| \frac{\delta \mathbf{a}_{100}}{\beta \alpha^4/M} =& \frac{e^{-2 x}}{x^2}\left[1-e^{2x}+2x(1+x)\right]\mathbf{e}_r. |
| \end{align} |
|
|
| Here we neglect the relativistic corrections, e.g., from the gravitomagnetic field of the cloud due to its angular momentum~\cite{Cao:2024wby}. At leading order in $\alpha$, the stationary gravitomagnetic potential of the $|211\rangle$ and $|1011\rangle$ states are given by |
| \begin{align} |
| \boldsymbol{\Xi} =-\frac{2 J_\text{c} \sin\theta}{r^2}\mathbf{e}_\phi |
| , |
| \quad |
| J_\text{c}(\mathbf{r}) =\frac{M_\text{c}}{\mu}\mathcal{J}(x), |
| \end{align} |
| with |
| \begin{align} |
| \mathcal{J}_{211}(x) & =1-\frac{1}{8}e^{-x}(x+2)(x^2+2x+4) |
| , |
| \\ |
| \mathcal{J}_{1011}(x) & =1 - e^{-2x}(1+2x+2x^2) |
| . |
| \end{align} |
| The gravitomagnetic acceleration $\mathbf{v}\times (\nabla\times\boldsymbol{\Xi})$ is suppressed by $\mathcal{O}(\alpha|\mathbf{v}|)$ relative to $-\nabla\Phi$. Meanwhile, dissipative effects are suppressed by the mass ratio $M_*/M$, where $M_*$ is the mass of the companion. Despite their possible relevance for the secular orbital evolution, they are much weaker than the conservative effects if the cloud is only weakly perturbed~\cite{Cao:2024wby}, and consequently are also neglected.\footnote{Although the dissipative effects are negligible for short-term evolution, they can potentially modify the distribution of orbital parameters in the nuclear star cluster, as recently discussed in \cite{tomaselli2025probingdenseenvironmentssgr}.} |
|
|
| \subsection{Spherically symmetric ground state} |
| Even if $\alpha \ll 1$, the GA approximation breaks down if $\beta$ becomes sufficiently large. In this section, we consider the exact spherically symmetric solution to the SP equation, focusing particularly on the ground state. Note that the solution taking into account the central mass has also been studied in \cite{Chavanis:2019bnu,Bar:2019pnz,Davies_2020}. |
|
|
| Without loss of generality, we consider the case of scalar field.\footnote{At the level of the SP equation, a spherically symmetric mass distribution of higher-spin fields can still possess spin angular momentum. In these cases, we again neglect the resulting gravitomagnetic fields.} For the spherically symmetric ansatz: |
| \begin{equation} |
| \psi(t,r)=f(r)\,e^{-iEt}, |
| \end{equation} |
| with $f\in\mathbb{R}$, the SP equation is reduced to |
| \begin{equation}\label{eq1} |
| Ef=-\frac{1}{2m}\frac{\partial_r\left(r^2 \partial_r f\right)}{r^2} + m \left(\Phi-\frac{M}{r}\right) f |
| ,\quad |
| \frac{\partial_r\left(r^2 \partial_r\Phi\right)}{r^2}=4\pi m f^2. |
| \end{equation} |
| Introducing the nondimensional radius $y\equiv m r$ and |
| \begin{equation} |
| V=2\left(\Phi-\frac{E}{m}\right), |
| \quad |
| F=\sqrt{\frac{8\pi}{m}}\,f, |
| \end{equation} |
| Eq.~\eqref{eq1} is further simplified to |
| \begin{equation}\label{eq2} |
| y \partial_y^2F+2\partial_yF=(yV-2\alpha )F |
| , |
| \quad |
| y\partial_y^2V+2\partial_y V=yF^2. |
| \end{equation} |
| A bound state ($E<0$) solution $\{V_\kappa(y,\alpha),F_\kappa(y,\alpha)\}$ can be specified by the boundary condition $F(0,\alpha)= \kappa^2$ with $\partial_yV(0,\alpha)=\partial_yF(0,\alpha)=F(\infty,\alpha)=0$. For a given number of nodes in the wavefunction, the value of $V_\kappa(0,\alpha)$ can be determined numerically by the shooting method. |
|
|
| Eq.~\eqref{eq2} has a scaling symmetry: |
| \begin{align} |
| F_\kappa(y,\alpha) &=\kappa^2 F_1(\kappa y,\alpha/\kappa), |
| \\ |
| V_\kappa(y,\alpha) &=\kappa^2 V_1(\kappa y,\alpha/\kappa), |
| \end{align} |
| it follows that |
| \begin{align} |
| \beta_\kappa(\alpha) &=\beta_1(\alpha/\kappa)= B(\alpha/\kappa), |
| \\ |
| \Phi_\kappa(y,\alpha) &=\kappa^2 \Phi_1(\kappa y,\alpha/\kappa), |
| \\ |
| E_\kappa(\alpha,\gamma) &=\kappa^2 E_1(\alpha/\kappa)=\left(-\frac{m\alpha^2}{2}\right)C(\beta).\label{E_kappa} |
| \end{align} |
| where we define the functions: |
| \begin{align} |
| B(\alpha) &\equiv \frac{1}{2\alpha}\int_0^\infty dy\,y^2\,[F_1(y,\alpha)]^2, |
| \\ |
| C(\beta) &\equiv V_1(\infty,g(\beta))\,[g(\beta)]^{-2}. |
| \end{align} |
| Therefore we only need to solve the bound state for $\kappa=1$. The $\alpha$-$\beta$ relation can be obtained from $\beta=B(\alpha/\kappa)$ as $g(\beta)=\alpha/\kappa$, with $g(x)$ being the inverse function of $B(x)$ (see Fig.~\ref{fig:g_beta}). Utilizing the scaling relations above, the mass density profile is explicitly given by |
| \begin{equation} |
| \begin{aligned} |
| \rho & = m|\psi|^2= |
| \frac{m^2}{8\pi}\left[\frac{\alpha}{g(\beta)}\right]^4 \left[F_1\left(\frac{\alpha m r}{g(\beta)},g(\beta)\right)\right]^2 |
| . |
| \end{aligned} |
| \end{equation} |
| Note that $\alpha m r=x=r/r_\text{c}$, the normalized density profile thus depends only on $\beta$ and $x$. |
|
|
| Correspondingly, the gravitational acceleration is given by $\delta \mathbf{a}=\delta a\,\mathbf{e}_r$, with |
| \begin{equation} |
| \begin{aligned} |
| \delta a(r)=-\frac{m}{2}\left[\frac{\alpha}{g(\beta)}\right]^3 A\left(\frac{\alpha m r}{g(\beta)},g(\beta)\right) |
| , |
| \end{aligned} |
| \end{equation} |
| and $A(y,\alpha)\equiv \partial_y V_1(y,\alpha)$ (see Fig.~\ref{fig:A_and_F1}). From this we also obtain the enclosed mass profile as $M_\text{c}(r)=-r^2\delta a(r)$. |
|
|
| The ground state is the state with minimum value of $E$ and is characterized by the absence of nodes in the wavefunction. Some limiting behaviours of the ground state are as follows: |
| \begin{itemize} |
|
|
| \item For $y\to 0$, $V_1(y,\alpha)\to y^2/6$, thus $A(y,\alpha)\to y/3$, independent of $\alpha$. The density profile is given by $M_\text{c}(r)/M \to (4\pi/3)\,\rho(0)\, r^3/M=m^6\,D(\beta)\,(Mr)^3/6$, with $D(\beta)\equiv \left[g(\beta)\right]^{-4}\left[F_1\left(0,g(\beta)\right)\right]^2$ (see Fig.~\ref{fig:D_beta}). For $\beta \gtrsim 100$, $D(\beta)\approx 0.11\beta^4$. |
|
|
| \item For $y\to \infty$, $\delta a\to -\beta M/r^2$, and $A(y,\alpha)\to 2B(\alpha)\,\alpha/y^2$. |
|
|
| \item For $\alpha=0$ (i.e., without the central point mass), a good fit is given by (compatible with \cite{Schive:2014dra}) |
| \begin{equation} |
| F_1(y,0)=\frac{1}{(1+0.03762\,y^2)^4}. |
| \end{equation} |
|
|
| \item For $\beta \to 0$, the ground state tends smoothly to the $|100\rangle $ state of scalar GA, with |
| \begin{equation} |
| F_1(y,\alpha)=e^{-\alpha y},\quad |
| V_1(0,\alpha)=\alpha^2,\quad |
| B(\alpha)=\frac{1}{8 \alpha ^4}, |
| \quad |
| g(\beta)=(8\beta)^{-1/4}, |
| \end{equation} |
| and |
| \begin{equation} |
| A(y,\alpha)=\frac{1-e^{-2 \alpha y}[1+2 \alpha y (1+\alpha y)]}{4 \alpha^3y^2}. |
| \end{equation} |
|
|
| \end{itemize} |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{g_beta.pdf} |
| \caption{$g(\beta)$.} |
| \label{fig:g_beta} |
| \end{figure} |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{F1_y_alpha.pdf} |
| \; |
| \includegraphics[width=0.45\textwidth]{A_y_alpha.pdf} |
| \caption{The function $F_1(y,\alpha)$ characterizing the radial profile of the ground state, and the function $A(y,\alpha)$ characterizing the gravitational acceleration.} |
| \label{fig:A_and_F1} |
| \end{figure} |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{D_beta.pdf} |
| \caption{The function $D(\beta)$ characterizing the central mass density of the ground state.} |
| \label{fig:D_beta} |
| \end{figure} |
|
|
| Note that for $\beta \gtrsim 100$, $\rho/m^3<\rho(0)/m^3=\frac{M}{8\pi}\alpha^3D(\beta) \approx \left(\frac{M}{4.3\times 10^6M_\odot}\right)\left(\frac{\alpha}{0.1}\right)^3\beta^4\times 0.1\,\text{ms}$. |
|
|
| For a sufficiently small value of $\beta$, the ground state can be approximated by a scalar $|100\rangle $ state with self-gravity correction, we take this chance to check its range of validity. Using Eqs.~\eqref{Phi_1011} and \eqref{E_kappa}, the correction to the energy level is |
| \begin{equation} |
| \langle 100|m \Phi_{100}| 100\rangle=\int d^3r\,\left|\psi^{100}\right|^2 m \Phi_{100} = -\frac{5}{8}m\beta \alpha^2, |
| \end{equation} |
| thus |
| \begin{equation} |
| C(\beta)\approx\frac{E^{(1)}+\langle 100|m \Phi_{100}| 100\rangle}{-m\alpha^2/2}=1+\frac{5}{4}\beta. |
| \end{equation} |
| As shown in Fig.~\ref{fig:C_beta}, this approximation remains good even for $\beta \sim 1$, while $C(\beta)\propto \beta^2$ in the large-$\beta$ limit. |
|
|
| The leading-order correction to the wavefunction of $|100\rangle$ state comes from its mixings with the $|n\ge 2,00\rangle$ states, given by |
| \begin{equation} |
| \Delta\psi^{(100)}=\sum_{n\ge 2}\frac{\langle n00|m\Phi_{100}|100\rangle}{E^{(1)}-E^{(n)}}\psi^{(n00)}. |
| \end{equation} |
| Compared with the exact radial wavefunction |
| \begin{equation} |
| \frac{f(r)}{\alpha^2\sqrt{m}}=\frac{1}{\sqrt{8\pi}\,[g(\beta)]^2} F_1\left(\frac{x}{g(\beta)},g(\beta)\right), |
| \end{equation} |
| the corrected radial wavefunction takes the form |
| \begin{equation} |
| \frac{f(r)}{\alpha^2\sqrt{m}}=\sqrt{\beta }\left[\frac{e^{-x}+\beta \mathcal{F}(x)}{\sqrt{\pi }}\right]. |
| \end{equation} |
| The function $\mathcal{F}(x)$ after including the mixing with $n=2,3,4$ $s$-states is |
| \begin{equation} |
| \mathcal{F}(x)\approx -\frac{8192 e^{-\frac{x}{2}} (x-2)}{194481}+\frac{141 e^{-\frac{x}{3}} [2 (x-9) x+27]}{200000}-\frac{2891776 e^{-\frac{x}{4}} [(x-12)^2 x-192]}{75418890625}, |
| \end{equation} |
| and the contribution from higher-$n$ states is small. As shown in Fig.~\ref{fig:wave_function_correction}, this correction is small for $\beta<1$, while for $\beta\gtrsim 1$ the approximation becomes poor. |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{C_beta.pdf} |
| \caption{The function $C(\beta)$ characterizing the energy level of the ground state.} |
| \label{fig:C_beta} |
| \end{figure} |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{wave_function_correction.pdf} |
| \caption{Comparison between the exact radial wavefunction of self-gravitating spherically symmetric ground state (solid line), the GA limit (dotted line) and the result with self-gravity correction (dashed line), for different values of $\beta$.} |
| \label{fig:wave_function_correction} |
| \end{figure} |
|
|
| \section{Quasibound states of massive scalar and vector fields in Kerr spacetime} \label{sec:appendixB} |
| In this section, we compare the nonrelativistic GA approximation with the relativistic quasibound state (QBS) solutions for the scalar $|211\rangle$ and vector $|1011\rangle$ states. |
|
|
| The relativistic field profiles associated with the NR wavefunction is given by |
| \begin{equation} |
| \phi=\frac{1}{\sqrt{2m}}\left(\psi\, e^{-im t}+\text{c.c.}\right), |
| \end{equation} |
| for the real\footnote{In the test-field limit, the situation is similar for a complex field, which can be represented by $\phi=\frac{1}{\sqrt{2m}}\left(\psi_+\, e^{-im t}+\psi_-\, e^{-im t}\right)$.} scalar field $\phi$, and |
| \begin{equation} |
| A_i=\frac{1}{\sqrt{2m}}\left(\psi_i\, e^{-im t}+\text{c.c.}\right), |
| \end{equation} |
| for the real vector field $A_i$, such that the energy density $-T^0_0$ reduces to the mass density $\rho=m\sum_I|\psi_I|^2$ in the NR limit. For the GA occupied by a single state, we thus have |
| \begin{align} |
| \phi^\text{(GA)}_{nl\texttt{m}}(t,r,\theta,\phi) &=\sqrt{\frac{M_\text{c}}{2m^2}} \left[\psi^{(nl\texttt{m})}\, e^{-im t}+\text{c.c.}\right], |
| \\ |
| \mathbf{A}^{\text{(GA)}}_{nlj\texttt{m}}(t,r,\theta,\phi) &= \sqrt{\frac{M_\text{c}}{2m^2}} \left[\boldsymbol{\psi}^{(nlj\texttt{m})}\, e^{-im t}+\text{c.c.}\right]. |
| \end{align} |
| Using |
| \begin{equation} |
| R_{21}(x)=\frac{e^{-x/2} x}{2 \sqrt{6}}, |
| \quad |
| Y_{11}(\theta,\phi)=-\sqrt{\frac{3}{8 \pi }} e^{i \phi } \sin \theta |
| , |
| \end{equation} |
| and |
| \begin{equation} |
| R_{10}(x)=2 e^{-x}, |
| \quad |
| Y_{00}(\theta,\phi)=\frac{1}{\sqrt{4\pi }} |
| , |
| \end{equation} |
| we obtain for the scalar $|211\rangle$ state, |
| \begin{equation} |
| \phi_{211}^\text{(GA)}(t,r,\theta,\phi)=-\frac{\alpha ^2 \sqrt{\beta } }{4 \sqrt{2 \pi }}e^{-x/2} x \sin \theta\, \cos (\omega_{211} t- \phi), |
| \end{equation} |
| and for the vector $|1011\rangle$ state (in the spherical coordinates), |
| \begin{equation} |
| \left(\begin{matrix} |
| A_{1011}^{\text{(GA)}t} |
| \\ |
| A_{1011}^{\text{(GA)}r} |
| \\ |
| A_{1011}^{\text{(GA)}\theta} |
| \\ |
| A_{1011}^{\text{(GA)}\phi} |
| \end{matrix}\right) |
| = |
| \left(\begin{matrix} |
| A_{1011}^{\text{(GA)}t} |
| \\ |
| \mathbf{A}_{1011}^{\text{(GA)}}\cdot\mathbf{e}_r |
| \\ |
| \mathbf{A}_{1011}^{\text{(GA)}}\cdot\mathbf{e}_\theta/r |
| \\ |
| \mathbf{A}_{1011}^{\text{(GA)}}\cdot\mathbf{e}_\phi/(r\sin\theta) |
| \end{matrix}\right) |
| =-\frac{\alpha ^2 \sqrt{\beta }}{\sqrt{\pi }} e^{-x}\left(\begin{matrix} |
| -\alpha \sin \theta\, \sin \left(\omega_{1011} t-\phi \right) |
| \\ |
| \sin \theta\, \cos \left(\omega_{1011} t-\phi \right) |
| \\ |
| \cos \theta\, \cos \left(\omega_{1011} t-\phi \right) |
| \\ |
| \sin \left(\omega_{1011} t-\phi \right) |
| \end{matrix}\right), |
| \end{equation} |
| with $ \omega_{211}=m+E^{(2)}$, $ \omega_{211}=m+E^{(1)}$ and $x=r/r_\text{c}$. |
|
|
| The superradiant cloud produced by a spinning black hole (in the test field limit) lives in the Kerr spacetime, and the GA approximation above unavoidably breaks down in the inner region sufficiently close to the event horizon. To check this deviation, we compute the QBSs of free scalar and vector fields corresponding to the GA states $|211\rangle$ and $|1011\rangle$ in the NR limit. |
|
|
| In the Boyer-Lindquist coordinates $(t,r,\theta,\phi)$, the metric of Kerr spacetime with mass parameter $M$ and spin parameter $a=M\chi$ is given by |
| \begin{equation} |
| g_{a b}=\left(\begin{smallmatrix} |
| -1+\frac{2 M r}{r^2+a^2 \cos ^2 \theta} & 0 & 0 & -\frac{2 M r a \sin ^2 \theta}{r^2+a^2 \cos ^2 \theta} \\ |
| 0 & \frac{r^2+a^2 \cos ^2 \theta}{r^2-2 M r+a^2} & 0 & 0 \\ |
| 0 & 0 & r^2+a^2 \cos ^2 \theta & 0 \\ |
| -\frac{2 M r a \sin ^2 \theta}{r^2+a^2 \cos ^2 \theta} & 0 & 0 & \sin ^2 \theta\left(r^2+a^2+\frac{2 M r a^2 \sin ^2 \theta}{r^2+a^2 \cos ^2\theta}\right) |
| \end{smallmatrix}\right). |
| \end{equation} |
|
|
| The seperable ansatz of a free real scalar field is |
| \begin{equation} |
| \phi^\text{(rel)}_{nl\texttt{m}}(t,r,\theta,\phi)\propto e^{-i\omega t+i\texttt{m}\phi}S(\theta)\,R(r)+\text{c.c.}, |
| \end{equation} |
| where $S=S_{l\texttt{m}}(\gamma,\theta)$ is the spheroidal harmonics with the oblateness parameter $\gamma=ia\sqrt{\omega^2-\mu^2}$. Quasibound states correspond to solutions that are purely ingoing at the horizon $r_+/M=1+\sqrt{1-\chi^2}$ and exponentially decaying at infinity, this results in a complex discrete spectrum: $\omega=\omega_R+i\omega_I$, with $\omega_R<m$ and $\omega_I\propto \texttt{m}\Omega_H-\omega_R$, $\Omega_H=\frac{1}{2M}\frac{\chi}{1+\sqrt{1-\chi^2}}$ being the angular velocity of the horizon. The superradiant growth ($\omega_I>0$) happens when $\texttt{m}\Omega_H>\omega_R$. Note that $n=l+1+\hat n_S$, with the overtone number $\hat n_S\ge 0$. We compute the $|211\rangle$ QBS numerically following the approach of \cite{Dolan_2007}. See also \cite{B2,Bao_2022} for the analytical approximations. |
|
|
| The seperable ansatz of a free real vector field (for all electric modes and a subset of magnetic modes \cite{B2}) is \cite{Frolov:2018ezx} |
| \begin{equation} |
| A^{\text{(rel)}\,a}_{nlj\texttt{m}}(t,r,\theta,\phi)=B^{ab}\,\partial_bZ, |
| \quad |
| Z\propto e^{-i\omega t+i\texttt{m}\phi}S(\theta)\,R(r)+\text{c.c.}, |
| \end{equation} |
| where $B^{ab}(\nu)$ is related to an eigenvalue $\nu$ by $B^{ab}(g_{bc}+i\nu h_{bc})=\delta^a_c$, and $h_{ab}$ is the principal Killing–Yano tensor of Kerr spaceime satisfying $\nabla_ah_{bc}=g_{ab}\xi_c-g_{ac}\xi_b$, with $\xi^c$ being the timelike Killing vector. In the Boyer-Lindquist coordinates, $\xi^c=\partial_t x^c$, |
| \begin{equation} |
| h_{ab}=\left(\begin{smallmatrix}0&r&a^2\cos\theta\sin\theta&0\\-r&0&0&ar\sin\theta^2\\-a^2\cos \theta\sin\theta&0&0&a\cos\theta \sin\theta\,(a^2+r^2)\\0&-ar\sin\theta^2&-a\cos\theta\sin\theta\,(a^2+r^2)&0\end{smallmatrix}\right). |
| \end{equation} |
| The meaning of $\{n,l,j,\texttt{m}\}$ comes from the $\alpha\to 0$ limit, with the correspondence $j=l-S$ and $|S|\le 1$. Note that in the convention of \cite{PhysRevD.98.104006}, $n=|\texttt{m}|+1+S+\hat n_V$, with the overtone number $\hat n_V\ge 0$. We compute the $|1011\rangle$ QBS using the numerical solver provided by \cite{Fell_2023}. |
|
|
| Fig.~\ref{fig:compare_relativistic} displays the results for $\alpha=0.11$ and $\chi=0.6$, where for concreteness we compare the radial functions $R_{10}(x)$ and $R_{21}(x)$ with $\left[\frac{R_{10}(1)}{A_{1011}^{\text{(rel)}\,r}(0,1,\pi/2,0)}\right]A_{1011}^{\text{(rel)}\,r}(0,x,\pi/2,0)$ and $\left[\frac{R_{21}(1)}{R_{211}^{\text{(rel)}}(1)}\right]R_{211}^{\text{(rel)}}(x)$, respectively. The agreement between GA and QBS profiles appears to be good at $r> 20M$ (this holds also for the angular profiles), and is better for smaller values of $\alpha$. This justifies our nonrelativistic modeling of the bosonic field and its metric perturbation at large radius, when the self-gravity of the field is negligible.\footnote{In this regime, the (non-radiative) linear metric perturbation sourced by the QBS can, in principle, be computed within black-hole perturbation theory, which would provide a more accurate description of its gravitational effects, particularly on the the conservative dynamics of a small companion.} |
|
|
| \begin{figure}[hbt!] |
| \includegraphics[width=0.45\textwidth]{compare_relativistic.pdf} |
| \caption{Comparison between the radial profiles of GA and relativistic QBS for the scalar $|211\rangle$ and vector $|1011\rangle$ states.} |
| \label{fig:compare_relativistic} |
| \end{figure} |
|
|
| \section{Orbital dynamics}\label{orbit} |
|
|
| |
|
|
| \begin{figure*} |
| \includegraphics[width=0.4\textwidth]{orbit.png} |
| \caption{Schematic of a pulsar orbit around Sgr A$^{*}$.} |
| \label{fig:frame} |
| \end{figure*} |
|
|
| In this work, we use PN expansion to solve for the two-body orbit, the relative acceleration can be expressed as |
| \begin{align}\label{acc} |
| \ddot{\mathbf{r}}\equiv\frac{\mathrm{d}^2\mathbf{r}}{\mathrm{d}t^2}=\ddot{\mathbf{r}}_\text{N}+\ddot{\mathbf{r}}_\text{1PN}+\ddot{\mathbf{r}}_\text{SO}+\ddot{\mathbf{r}}_\text{Q}+\ddot{\mathbf{r}}_\text{2PN}+\ddot{\mathbf{r}}_\text{DM}, |
| \end{align} |
| where $\mathbf{r}$ is a vector which refers to the relative coordinate position in the harmonic gauge, $t$ is the coordinate time, and the terms to the right of the equal sign represent Newtonian acceleration, 1PN acceleration, spin-orbit coupling acceleration, quadrupolar acceleration, 2PN acceleration and acceleration induced by ultralight field's gravitational potential, respectively. Here we ignore the higher PN terms. |
|
|
| Because the mass of the pulsar is far smaller than the mass of BH, the mass ratio $m_{\text{PSR}}/M < 10^{-6}$, we can ignore the mass of the pulsar and treat the pulsar as a test particle moving in the BH's spacetime. In this case, the terms in Eq.~\eqref{acc} are given by \cite{Hu:2023ubk} |
| \begin{align} |
| \ddot{\mathbf{r}}_\text{N}&=-\frac{M}{r^2}\mathbf{e}_r,\\ |
| \ddot{\mathbf{r}}_\text{1PN}&=-\frac{M}{r^2}\left[\left(-\frac{4M}{r}+v^2\right)\mathbf{e}_r-4\dot{r}\mathbf{v}\right], \label{1PN}\\ |
| \ddot{\mathbf{r}}_\text{SO}&=\chi\frac{6M^2}{r^3}\left[\hat{\mathbf{s}}\cdot(\mathbf{e}_r\times \mathbf{v})\mathbf{e}_r+\dot{r}(\mathbf{e}_r\times \hat{\mathbf{s}})-\frac{2}{3}(\mathbf{v}\times \hat{\mathbf{s}})\right],\\ |
| \ddot{\mathbf{r}}_\text{Q}&=-q\frac{3M^3}{2r^4}\left\{\left[5(\mathbf{e}_r\cdot \hat{\mathbf{s}})^2-1\right]\mathbf{e}_r-2(\mathbf{e}_r\cdot \hat{\mathbf{s}})\hat{\mathbf{s}}\right\},\\ |
| \ddot{\mathbf{r}}_\text{DM}&= \delta \mathbf{a}, |
| \end{align} |
| where $r\equiv|\mathbf{r}|$, $\dot{r}\equiv \mathrm{d}r/\mathrm{d}t$, $\mathbf{v}\equiv \mathrm{d}\mathbf{r}/\mathrm{d}t$ and $v\equiv|\mathbf{v}|$. |
|
|
| The Keplerian orbit of the pulsar and the corresponding coordinate system and notations are shown in Fig.~\ref{fig:frame}. The $\mathbf{K}_0$ represent the direction from Earth to Sgr~A*, and $(\mathbf{I}_0,\mathbf{J}_0)$ forms the sky plane. The orbital parameters of pulsars include orbital period $P_b$, eccentricity $e$, inclination $i$, longitude of the periastron $\omega$, longitude of ascending node $\Omega$ and initial true anomaly $\theta_0$. $\eta$ and $\lambda$ represent the spin orientation of the BH: $\hat{\mathbf{s}}=(\sin \lambda \cos \eta, \sin \lambda \sin \eta, \cos \lambda)$. In this simulation, the longitude of the ascending node ($\Omega$) is fixed at $\Omega = 0$, and the other parameters are listed in Table~\ref{tab:benchmark}. |
|
|
| \begin{table}[h!] |
| \centering |
| |
| |
| |
| |
| |
| |
| |
| \begin{tabularx}{\textwidth}{@{\extracolsep{\fill}} cccccccccc } |
| \hline |
| $M/M_\odot$ & $\chi$ & $q$ & $\lambda$ & $\eta$ & $P_b$ & $e$ & $i$ & $\omega$ & $\theta_0$\\ |
| \hline |
| $4.3\times 10^6$ & $0.6$ & $-0.36$ & $\frac{1}{3}\pi$ & $\frac{5}{9}\pi$ & $0.5\,\text{yr}$ & $0.8$ & $\frac{1}{5}\pi$ & $\frac{5}{7}\pi$ & $\frac{1}{3}\pi$\\ |
| \hline |
| \end{tabularx} |
| \caption{Parameters used in the simulation.} |
| \label{tab:benchmark} |
| \end{table} |
|
|
| |
| |
| |
| |
| |
| |
|
|
| The observation time in our simulation is 5 years, that contains $\sim 10$ orbits. Apart from that, we also considered cases with orbital period of $P_b\sim 5$ years and eccentricity of $e \sim 0.3$ to compare the effects of ULDM on different pulsar orbits. |
|
|
| It is worth mentioning that, in order to express the acceleration caused by ULDM in the scene we defined above, we need to calculate $\theta$ and $\mathbf{e}_{\theta}$ with $\cos \theta = \hat{\mathbf{s}}\cdot \mathbf{e}_r$ and |
| \begin{equation} |
| \mathbf{e}_{\theta}=\frac{(\hat{\mathbf{s}}\times\mathbf{e}_r)\times\mathbf{e}_r}{|\hat{\mathbf{s}}\times\mathbf{e}_r|}. |
| \end{equation} |
|
|
|
|
| |
| |
|
|
| |
| |
| |
| |
| |
| |
|
|
| |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
|
|
| \section{Pulsar timing}\label{pt} |
| \subsection{Timing model} |
| To explore the possibility of constraining the bosonic field near Sgr A$^*$ using pulsar timing, the first step is to develop a timing model that properly accounts for the extra attraction from the bosonic field stucture. In pulsar timing, the model connects the pulsar’s intrinsic rotation to the observed times of arrival (TOAs) at radio telescopes by incorporating various physical effects that influence the signal during its propagation from the pulsar to the Earth. |
|
|
| In pulsar's frame, the proper rotation numbers N of the pulsar can be expressed by |
| \begin{align} |
| N(T)=N_0+\nu\,T+\frac{1}{2}\dot{\nu}\,T^2, |
| \end{align} |
| where $\nu \equiv 1/P$ is the spin frequency, $T$ is the proper time of the pulsar, and $\dot{\nu}$ is the time derivative of $\nu$. The relation between the TOAs in the observer frame and the proper time is \cite{AIHPA_1986__44_3_263_0} |
| \begin{align} |
| t^{\text{TOA}}=T+\Delta_E+\Delta_R+\Delta_S, |
| \end{align} |
| where $\Delta_E, \Delta_R, \Delta_S$ refer to Einstein delay, Romer delay and Shapiro delay. First, the Einstein delay is defined by the difference between the coordinate time $t$ and the proper time $T$ \cite{AIHPA_1986__44_3_263_0,1976ApJ...205..580B}: |
| \begin{align}\label{einlay} |
| \Delta_E \equiv t - T. |
| \end{align} |
| At the lowest order, the proper time $T$ is connected to $t$ through |
|
|
| \begin{align} |
| \frac{\mathrm{d}T}{\mathrm{d}t}=1-\frac{M}{r}-\frac{v^2}{2}. |
| \end{align} |
| We can see from above that the Einstein delay includes both gravitational redshift and special-relativistic time-dilation effects. The next one is Romer delay $\Delta_R$, which is a geometric effect caused by the orbital motion of the pulsar. Its form is |
| \begin{align} |
| \Delta_R \equiv \hat{\mathbf{K}}_0 \cdot \mathbf{r}. |
| \end{align} |
| The last term in our timing model is 1PN Shapiro delay. Its existence is due to the propagation of light in curved spacetime \cite{1976ApJ...205..580B,PhysRevLett.13.789}: |
| \begin{align}\label{shalay} |
| \Delta_S=-2M \ln \left(r-\mathbf{r}\cdot \hat{\mathbf{K}}_0\right). |
| \end{align} |
| In our timing model, we assume that the form of the Einstein delay and Shapiro delay in Eqs.~\eqref{einlay} and \eqref{shalay} do not include the ULDM's effect, since it is negligible compared to the Romer delay. |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
| |
| |
| |
| |
| |
| |
|
|
| |
|
|
|
|
|
|
| |
| |
| |
| |
| |
|
|
|
|
|
|
| |
| |
|
|
| |
|
|
| |
| |
|
|
| |
|
|
| |
| \subsection{Timing residual induced by the ULDM}\label{tr} |
| If we do not include the ULDM's effect in our timing model, the extra time residual in $t^{\text{TOA}}$ will occur. Here we show the corresponding time residual produced by spherical soliton with $\alpha = 0.01$ and $\beta=10^{-6}$ in Fig.~\ref{fig:tr}. |
| \begin{figure*} |
| \includegraphics[width=0.55\textwidth]{before_3.png} |
| \caption{Time residuals caused by spherical soliton with $\alpha = 0.01$ and $\beta=10^{-6}$.} |
| \label{fig:tr} |
| \end{figure*} |
|
|
| In order to better match the actual situation, we also add the proper motion of Sgr A$^*$ in our timing model. We take $\mu_\alpha = -3.2$ mas/yr and $\mu_\delta=-5.6$ mas/yr. We can see the maximum amplitude of residual could reach $\sim 100$ ms in case of $\beta = 10^{-6}$, highly exceeding the timing precision $\sigma_\text{TOA}=1 $ms. However, $\beta\sim 10^{-6}$ is predicted in our parameter estimation. The reason is that the effects of ULDM will exhibit a certain degree of parameter degeneracy with the spin-orbit (SO) coupling term and other terms, thereby degrading the estimation accuracy of $\beta$. |
|
|
| |
|
|
|
|
| \section{Impacts of BH spin and orbital inclination on the parameter estimation}\label{sec:ad} |
| \subsection{Impacts of BH spin} |
| We select $\chi = 0.6$ in our simulation, however, the spin of BH is reduced to $\chi \sim 4\alpha/\texttt{m}$ in a saturated GA. From a theoretical standpoint, the difference in spin of $\mathcal{O}(1)$ leads to a corresponding $\mathcal{O}(1)$ change in the sensitivity to $\beta$. Our simulations further show that even for spins as small as $\chi=0.01$, the sensitivity to $\beta$ remains of the same order as in the case with $\chi=0.6$. The results are showed in Fig.~\ref{fig:chicom}. We therefore neglect the dependence of parameter estimation on the spin magnitude. |
| \begin{figure*} |
| \includegraphics[scale=0.3]{chi_com1.png} |
| \caption{Estimated sensitivity to $\beta$ for different spin $\chi$. For the solid (dashed) curve, the spin corresponds to 0.6 (0.01).} |
| \label{fig:chicom} |
| \end{figure*} |
| \subsection{Impacts of orbital inclination in the case of $|211\rangle$ state} |
| In case of scalar $|211\rangle$ state, we draw the result that how the sensitivity of $\beta$ evolves with the inclination angle $i$ and the longitude of the periastron $\omega$ in Fig.~\ref{fig:i}. The gray line shows the variation of $\beta$ with $i$, while the yellow curve represents the dependence of $\beta$ on $\omega$. We can conclude that the results are periodic for $\omega$ and the tighter and loosest sensitivities differ by $\sim\mathcal{O}(10)$ for both $i$ and $\omega$. |
| \begin{figure*} |
| \includegraphics[scale=0.3]{i-beta-n.png} |
| \caption{Estimated sensitivity to $\beta$ for different inclination angle $i$ and longitude of the periastron $\omega$. For the gray (yellow) curve, the horizontal axis corresponds to $i$ ($\omega$).} |
| \label{fig:i} |
| \end{figure*} |
|
|
| \section{Estimation of constraints from S2 star} \label{sec:appendixC} |
| We estimate the constraints on $\beta$ from the observation of S2 orbit based on its periastron precession rate reported by \cite{2020S2}. The instantaneous periastron precession rate of the osculating orbit defined in Fig.~\ref{fig:frame} under a perturbing acceleration $\mathbf{F}=F_r \mathbf{e}_r+F_t \mathbf{e}_t +F_n \hat{\mathbf{L}}$ (with $\hat{\mathbf{L}}$ being the unit vector parallel to the orbital angular momentum and $\mathbf{e}_t=\hat{\mathbf{L}}\times\mathbf{e}_r$) is given by\footnote{See for example \cite{Cao:2024wby}, with the replacement $\varphi_0+\pi/2\to \omega$ (argument of periastron), $\varphi \to \theta$ (true anomaly).} |
| \begin{equation} |
| \begin{aligned} |
| \dot \omega = \frac{\sqrt{1-e^2}}{ae(2\pi/T)}\left\{\left[1+\frac{r}{a(1-e^2)}\right]\sin\theta\, F_t-\cos\theta\, F_r\right\} |
| -\cot i |
| \frac{r \sin (\theta+\omega) F_n}{a^2(2\pi/T)\sqrt{1-e^2}}. |
| \end{aligned} |
| \end{equation} |
| The periastron shift per orbital period $T=2\pi/\sqrt{M/a^3}$ is approximately given by $\langle \dot\omega \rangle=\frac{1}{T}\int_0^T dt\,\dot\omega$, with the orbital elements in the integrand fixed to constant values. The 1PN correction \eqref{1PN} gives rise to the Schwarzschild precession: |
| \begin{equation} |
| \langle \dot\omega \rangle_\text{S}=\frac{3 M^{3/2}}{a^{5/2}(1-e^2)}, |
| \end{equation} |
| (at this level of estimation, we do not consider higher PN corrections) while $\mathbf{F}=-\nabla\Phi$ gives rise to an additional contribution $\langle \dot\omega \rangle_\text{DM}$. We then derive the constraint by imposing that \cite{2020S2} $(\langle \dot \omega \rangle_\text{S}+\langle \dot\omega \rangle_\text{DM}) T \in 12.1'\times [1.1+0.19\times(-1, 1)]$, taking $M=4.3\times 10^6\,M_\odot$, $e=0.886$ and $a=1002\,\text{AU}$. For the $|211\rangle$ state, we assume for simplicity that the BH spin is perpendicular to the orbital plane of S2. |
|
|
| \section{Newtonian potential of GA occupied by multiple bound eigenstates} |
| In our analysis, the GA is assumed to be dominated by a single eigenstate. In the more general scenario, the simultaneous occupation of multiple states (including the free states) is possible. In fact, a completely generic bound state can be expanded instantaneously as a superposition of stationary hydrogenic bound eigenstates, that fixes the instantaneous Newtonian potential. Here we briefly discuss the effects of multiple bound eigenstates, leaving a more detailed investigation to future work. |
|
|
| The mass density \eqref{rho_sphere} associated with the bound state \eqref{general_bound} is |
| \begin{equation} |
| \begin{aligned} |
| \rho_{l_*m_*} |
| &=\int Y_{l_*m_*}\sin\theta \,d\theta \,d\phi\,m|\psi_I|^2 |
| \\ |
| &=m\int Y_{l_*m_*}\sin\theta \,d\theta \,d\phi\left[\sum_{i,i'}c_{i}\,c_{i'}^*\psi^{(i)}\psi^{(i')*}\right] |
| \\ |
| &=mr_\text{c}^{-3}\sum_{i,i'}c_{i}\,c_{i'}^*\,I^{i,i'}_\Omega \,e^{i\left[E^{(i')}-E^{(i)}\right] t}R_{i}(x)R_{i'}(x) |
| , |
| \end{aligned} |
| \end{equation} |
| where $i$ refers to the set of quantum numbers $\{n,l,\texttt{m}\}$, and |
| \begin{equation} |
| I^{i,i'}_\Omega \equiv \int Y_{l_*m_*}\,Y_i\,Y_{i'}^*\sin\theta \,d\theta \,d\phi. |
| \end{equation} |
|
|
| The Newtonian potential is given by Eq.~\eqref{Poisson}: |
| \begin{equation}\label{multi-modes_Phi} |
| \begin{aligned} |
| \Phi &=\sum_{l_*\ge 0,\, m_*=\texttt{m}'-\texttt{m}}\frac{-4\pi}{2l_*+1} |
| \int_0^\infty ds\,\rho_{l_*m_*}(s)\,s^2 \left[\frac{s^{l_*}}{r^{l_*+1}}\Theta(r-s)+\frac{r^{l_*}}{s^{l_*+1}}\Theta(s-r)\right] |
| Y_{l_* m_*}^*(\theta,\phi) |
| \\ |
| &= |
| \sum_{l_*\ge |\texttt{m}'-\texttt{m}|}\frac{-4\pi m}{2l_*+1} |
| \sum_{i,i'}c_i\,c_{i'}^*\,I^{i,i'}_\Omega\,\frac{I_r^{i,i'}}{r_\text{c}}\, |
| Y_{l_*,\texttt{m}'-\texttt{m}}^*(\theta,\phi)\, \,e^{i\left[E^{(i')}-E^{(i)}\right] t} |
| , |
| \end{aligned} |
| \end{equation} |
| (where $\Theta(x)$ is the Heaviside unit step function) with |
| \begin{equation} |
| \begin{aligned} |
| I_r^{i,i'}\equiv \int_0^\infty dx\,x^2\,R_i(x)\,F_{l_*}(x)\,R_i(x), |
| \end{aligned} |
| \end{equation} |
| and |
| \begin{equation} |
| F_{l_*}(x) = \frac{x^{l_*}}{x_*^{l_*+1}}\Theta(x_*-x)+\frac{x_*^{l_*}}{x^{l_*+1}}\Theta(x-x_*),\quad x_*=x/r_\text{c}. |
| \end{equation} |
| Note that the structure of Eq.~\eqref{multi-modes_Phi} is almost identical with the bound-bound transition matrix element of GA in the presence of a companion\footnote{This connection is natural, since (in the nonrelativistic regime) it is the Newtonian potential of the cloud that is responsible for the backreaction of the companion-induced GA transitions, although the backreaction is typically analyzed using the flux-balance equations.} \cite{Cao:2024wby}, where the dipole term ($l_*=1$) is modified in the comoving frame of the central object due to its acceleration. In the present context, the dipole term in Eq.~\eqref{multi-modes_Phi} accelerates the central object, since from Eq.~\eqref{Poisson} the gravitational acceleration is given by |
| \begin{equation} |
| \begin{aligned} |
| \frac{-\nabla\Phi}{\frac{4\pi}{2l_*+1}} &=\sum_{l_*m_*}\left\{Y_{l_*m_*}\partial_r\left[\frac{q_{l_*m_*}}{r^{l_*+1}}+r^{l_*}p_{l_*m_*}\right]\mathbf{e}_r+\left[\frac{q_{l_*m_*}}{r^{l_*+2}}+r^{l_*-1}p_{l_*m_*}\right]\left(\mathbf{e}_\theta\partial_\theta+\mathbf{e}_\phi\frac{\partial_\phi}{\sin\theta}\right)Y_{l_*m_*}\right\}, |
| \end{aligned} |
| \end{equation} |
| while at the center, |
| \begin{equation} |
| \begin{aligned} |
| \lim_{r\to 0}\partial_r\left[\frac{q_{l_*m_*}}{r^{l_*+1}}+r^{l_*}p_{l_*m_*}\right] & = \lim_{r\to 0}l_* r^{l_*-1}p_{l_*m_*} = p_{1m_*}\,\delta_{l_*,1} |
| , |
| \\ |
| \lim_{r\to 0}\left[\frac{q_{l_*m_*}}{r^{l_*+2}}+r^{l_*-1}p_{l_*m_*}\right] & = \lim_{r\to 0} r^{l_*-1}p_{l_*m_*} = p_{1m_*}\,\delta_{l_*,1}, |
| \end{aligned} |
| \end{equation} |
| we thus obtain |
| \begin{equation} |
| \begin{aligned} |
| -\nabla\Phi|_{\mathbf{r=0}} & =\frac{4\pi}{3}\sum_{|m_*|\le 1}p_{1m_*}\left(\mathbf{e}_r+\mathbf{e}_\theta\partial_\theta+\mathbf{e}_\phi \frac{\partial_\phi}{\sin\theta}\right)Y_{1m_*} |
| \\ |
| &=\sqrt{\frac{4\pi}{3}}\sum_{|m_*|\le 1}p_{1m_*}\,\boldsymbol{\xi}^{m} |
| , |
| \end{aligned} |
| \end{equation} |
| with $\boldsymbol{\xi}^0=\mathbf{e}_z$ and $\boldsymbol{\xi}^{\pm 1}=\mp (\mathbf{e}_x\pm i \mathbf{e}_y)/\sqrt{2}$. |
|
|
| As an illustration, the density distribution of a two-level system is |
| \begin{align} |
| \rho &=\frac{M_\text{c}(t=0)}{r_\text{c}^3}|C_1\psi^{(1)}+C_2\psi^{(2)}|^2 |
| \\ |
| &=\frac{M_\text{c}(t=0)}{r_\text{c}^3}\left[ |C_1|^2|\psi^{(1)}|^2+|C_2|^2|\psi^{(2)}|^2+C_1C_2^*\psi^{(1)}\psi^{(2)*}+C_1^*C_2\psi^{(1)*}\psi^{(2)} |
| \right] |
| \\ |
| &\equiv \frac{M_\text{c}(t=0)}{r_\text{c}^3}g(x,\theta,\phi), |
| \end{align} |
| here the normalization is taken to be $|C_1|^2+|C_2|^2=1$ at $t=0$. For convenience we choose $\psi^{(i=1,2)}$ to be time-independent, with $C_i=A_i(t) e^{i\chi_i(t)}$ and $A,\chi\in\mathbb{R}$. As a concrete example, for $|1\rangle=|211\rangle$ and $|2\rangle=|21,-1\rangle$, |
| \begin{equation} |
| g=\frac{1}{64 \pi } e^{-x} x^2 \sin ^2\theta \left[A_1^2+A_2^2-2 A_1A_2 \cos (\chi_1-\chi_2+2 \phi )\right], |
| \end{equation} |
| correspondingly the Newtonian potential is |
| \begin{equation} |
| \begin{aligned} |
| \frac{\Phi}{\frac{M_\text{c}(t=0)}{r_\text{c}}} &=\frac{e^{-x}}{16 x^3} \left(x^5+6 x^4+24 x^3+72 x^2+144 x-144 e^x+144\right) \\ |
| &\qquad \times \sin ^2\theta \left[A_1^2+A_2^2-2 A_1 A_2 \cos (\chi_1-\chi_2+2 \phi )\right] |
| \\ |
| &\quad -\frac{e^{-x}}{4 x^3}\left[x^3+8 x^2+4 e^x \left(x^2-6\right)+24 x+24\right] \left(A_1^2+A_2^2\right). |
| \end{aligned} |
| \end{equation} |
| There are only $l_*= |m_*|=2$ terms. For the purely hydrogenic GA, $\chi_{211}=-E^{(2)}t=\chi_{21,-1}$, while $A_{211},A_{21,-1}$ are constant, the potential is thus stationary but not axially symmetric. The spin of the central object can break the degeneracy of this pair of hyperfine levels, and consequently the potential felt by a companion in circular orbit with angular velocity $\Omega$ depends on $\chi_1-\chi_2+2\phi=\left\{2\Omega-\left[E^{(211)}-E^{(21,-1)}\right]\right\}t$, which is stationary only at resonance, i.e., when $\Omega={\Delta E}/{\Delta \texttt{m}}=\left[E^{(211)}-E^{(21,-1)}\right]/2$. |
|
|
| |
| |
|
|
| \end{document} |
