betaFeedback

WorksPhysics and Astronomy

Magnetic monopoles produced by first-order phase transition

Abstract

We present the first 3D lattice simulations of magnetic monopole production from a first-order phase transition, using the Press-Ryden-Spergel method, capturing the subsequent annihilation stage and the entry into the scaling regime in an expanding universe. We establish the quantitative relationship between the initial bubble number and the residual monopole abundance, and calculate the magnetic field and gravitational wave power spectra sourced by monopole dynamics. Crucially, we find that the magnetic field spectrum displays a steep k−5k^{-5} decline beyond the monopole size wMw_M, whereas the peak of the gravitational wave spectrum aligns with the scale of the mean monopole separation dMd_M, governed by k3k^3 and k−3k^{-3} scalings on the low- and high-wavenumber sides of the peak, respectively. However, since the resulting magnetic field strength is excluded by blazar observations, magnetic monopoles cannot be the origin of primordial cosmic magnetic fields. Finally, we determine the allowed monopole mass under the constraints of the Parker bound (mM≲2.18×1011 GeVm_M\lesssim2.18\times10^{11}\,\mathrm{GeV}) and dark matter abundance (mM≲1.58×1012 GeVm_M\lesssim1.58\times10^{12}\,\mathrm{GeV}).

Introduction

Spontaneous symmetry breaking (SSB) in the early universe routinely gives rise to a variety of topologically stable solitons. In particular, when a grand unified gauge group GG undergoes a phase transition down to a subgroup containing an explicit U(1)U(1) factor, ’t Hooft-Polyakov monopoles [1, 2] are inevitably produced. A single ’t Hooft-Polyakov monopole carries a fundamental magnetic charge of eM=4π/ge_M = 4\pi/g and possesses an extremely large mass, typically scaled as mM∼vm_M \sim v, where vv represents the symmetry-breaking scale, typically the Grand Unified Theories (GUTs) scale. According to standard cosmological estimates, the creation of these superheavy defects yields an abundance that would severely overclose the universe today, which is known as the “monopole problem” [3]. While inflation [4, 5] can resolve this puzzle by drastically diluting the energy density, an alternative approach is to consider that SSB of GUT occurs through a strongly first-order phase transition (FOPT) with supercooling. Substantial supercooling before the phase transition, and large effective monopole mass relative to the temperature right after the transition, are proposed to suppress the final number density of monopole [6].

A FOPT proceeds via the thermal or quantum nucleation of true vacuum bubbles within a supercooled false vacuum background [7, 8, 9, 10, 11]. Driven by the vacuum energy difference, these bubbles expand outward at relativistic velocities and eventually undergo violent collisions. Within this framework, topological defects can be produced by the Kibble mechanism [12, 13], which is realized through the geodesic rule [14, 15, 16]: different bubbles carry random vacuum phases, and when they collide, the phases interpolate along geodesics in the vacuum manifold. If the closed path traced by several colliding bubbles is non-contractible in the vacuum manifold, the central region cannot be smoothly filled with vacuum values, and a topological defect forms due to the topological obstruction. Furthermore, bubble coalescence also generates magnetic fields [17, 18, 19, 20, 21, 22, 23], sphalerons [24, 25, 26], and gravitational waves (GWs) [10, 27, 28, 29, 30].

In this work, we systematically investigate the monopole production through the real-time dynamics of the FOPT and the subsequent evolution of magnetic monopoles using 3D lattice numerical simulations in an expanding Friedmann–Lemaître–Robertson–Walker (FLRW) universe for the first time. Specifically, we simulate the full evolutionary history from bubble nucleation to coalescence, which dynamically generates monopoles via the geodesic rule of the Kibble mechanism [12]. We further trace the subsequent pairwise annihilation of these monopole-antimonopole pairs until the system enters the scaling regime, thereby establishing a quantitative relationship between the initial bubble number density and the surviving monopole abundance. The spectra of the generated magnetic field and the gravitational wave sourced by monopole dynamics are also explored. By incorporating observational constraints from cosmic monopole searches, dark matter relic density, and magnetic fields, we separately evaluate scenarios where the monopoles reside in either the dark or visible sectors to derive rigorous bounds on the monopole mass and corresponding temperatures of the FOPTs.

Our results demonstrate that the previous studies overestimate the dark matter relic density from the monopoles and anti-monopoles after considering annihilation, and the GW spectra from the monopoles’ evolution are far too small to be probed by current detectors. Furthermore, our results demonstrate that the magnetic field strength generated by magnetic monopoles is exceedingly weak and falls well below the lower limits inferred from blazar observations. Consequently, magnetic monopoles are ruled out as the primary origin of primordial cosmic magnetic fields, necessitating an alternative magnetogenesis scenario.

This work adopts the natural units where ℏ=c=1\hbar= c = 1. To reasonably simplify the simulation, and considering that the phase transition completes extremely rapidly, we neglect the effects of the cosmic fluid and temperature variations.

The Model

We consider the Georgi-Glashow model with the SU(2)SU(2) Higgs Lagrangian being [31]:

L=−14WaμνWμνa+12DμϕaDμϕa−V(ϕ).(1)\mathcal{L}=-\frac{1}{4}W^{a\mu\nu}W^a_{\mu\nu}+\frac{1}{2}D^\mu\phi^aD_\mu\phi^a-\mathcal{V}(\phi). \tag*{(1)}

Where, Wμνa=∂μWνa−∂νWμa+gϵabcWμbWνcW^a_{\mu\nu}=\partial_\mu W^a_\nu-\partial_\nu W^a_\mu+g\epsilon^{abc}W^b_\mu W^c_\nu is the field strength tensor of the SU(2)SU(2) gauge field WμaW^a_\mu. The adjoint scalar field ϕa\phi^a is represented as ϕ=ϕaσa/2\phi=\phi^a\sigma^a/2, where σa\sigma^a denotes the Pauli matrices. The covariant derivative is defined as Dμϕa=∂μϕa+gϵabcWμbϕcD_\mu\phi^a=\partial_\mu\phi^a+g\epsilon^{abc}W^b_\mu\phi^c, and g=0.5g=0.5 is the coupling constant. The potential V(ϕ)\mathcal{V}(\phi) featuring a potential barrier

V(ϕ)=λϕaϕa[14ϕaϕa−(v+vB)3ϕaϕa+vvB2](2)\mathcal{V}(\phi)=\lambda\phi^a\phi^a\left[\frac{1}{4}\phi^a\phi^a-\frac{(v+v_B)}{3}\sqrt{\phi^a\phi^a}+\frac{vv_B}{2}\right] \tag*{(2)}

facilitates a first-order phase transition, where v=1v=1 is the vacuum expectation value (VEV) and vB=0.35vv_B=0.35v is the location of the potential barrier. We incorporate the model into an expanding cosmological background using the FLRW metric:

ds2=−a2(τ)(dτ2−δijdxidxj)(3)ds^2=-a^2(\tau)\left(d\tau^2-\delta_{ij}dx^idx^j\right) \tag*{(3)}

where a(τ)a(\tau) is the scale factor and τ\tau denotes the conformal time. Using the temporal gauge W0a=0W^a_0=0, we obtain a self-consistent set of dynamical equations for the fields in the FLRW spacetime:

(a2ϕa′)′−a2DiDiϕa=−a4∂V∂ϕa(4)(a^2\phi^{a\prime})'-a^2D^iD_i\phi^a=-a^4\frac{\partial\mathcal{V}}{\partial\phi^a} \tag*{(4)}
Wia′′+DjWija=−a2gϵabcϕb(Diϕc).(5)W_i^{a\prime\prime}+D^jW^a_{ij}=-a^2g\epsilon^{abc}\phi^b(D_i\phi^c). \tag*{(5)}

The prime (′\prime) indicates the derivative with respect to conformal time τ\tau. We assume the expansion is driven by a radiation-dominated fluid (equation of state ω=1/3\omega=1/3); the scale factor evolves as a(τ)=[1+H0(τ−τ0)]a(\tau)=[1+\mathcal{H}_0(\tau-\tau_0)], where H0\mathcal{H}_0 is the initial conformal Hubble parameter, and a0=a(τ0)=1a_0=a(\tau_0)=1. The discretized equations of motion are detailed in the Supplemental Material.

The first-order phase transition proceeds through the nucleation and subsequent coalescence of vacuum bubbles. Each bubble carries a local SU(2)SU(2) phase; consequently, upon collision, the non-trivial SU(2)SU(2) vacuum configurations lead to the formation of point-like topological defects, magnetic monopoles. The four-bubble-vertex monopole picture is a heuristic prediction built upon the geodesic rule assumption for the cosmic strings from three bubbles’ collision [14, 15, 16]. The rigorous mathematical generalization to π2\pi_2 monopoles via 2-simplex constructions applies primarily to global-symmetry models [32]. For gauged Higgs theories, the geodesic rule is not guaranteed to hold. Two competing contributions can coexist: topological defect formation at multi-bubble vertices, and dynamical pair-production driven by Higgs-modulus oscillations and transient local symmetry restoration in collision regions. We disentangle these two channels in our lattice simulations via spatial correlation diagnostics, monitoring of the Higgs modulus, controlled-bubble tests, and parameter scans.

Since the system initially possesses no net magnetic charge, these monopoles are produced in pairs, strictly maintaining a total magnetic charge of zero at all times. Following the completion of the phase transition, the newly formed monopoles undergo mutual attraction and pairwise annihilation until the system enters (or approaches) a scaling regime. During this stage, the most characteristic behavior is the remarkable consistency in the time evolution of the comoving mean monopole separation: [33]

dM=(V2NM)1/3,(6)d_M=\left(\frac{V}{2N_M}\right)^{1/3}, \tag*{(6)}

where VV is the simulation volume and NMN_M is the number of monopole-antimonopole pairs. In this work, we focus on establishing the quantitative relationship between the number of nucleated bubbles NbN_b and NMN_M. The procedure for identifying monopoles on the lattice is detailed in the Supplemental Material.

After SU(2)SU(2) symmetry is broken to U(1)U(1), a massless vector field Aμ=Wμaϕ^aA_\mu=W_\mu^a\hat{\phi}^a is produced, with field strength [1]

Aμν=Wμνaϕ^a−1gϵabcϕ^aDμϕ^bDνϕ^c,(7)A_{\mu\nu}=W_{\mu\nu}^a\hat{\phi}^a-\frac{1}{g}\epsilon^{abc}\hat{\phi}^aD_\mu\hat{\phi}^bD_\nu\hat{\phi}^c, \tag*{(7)}

where ϕ^a=ϕa/ϕbϕb\hat{\phi}^a=\phi^a/\sqrt{\phi^b\phi^b}. So the magnetic field is Bi=ϵijkAjk/2B_i=\epsilon_{ijk}A_{jk}/2. We can compute its energy spectrum:

dρBdln⁡k=4πk3B~i(k)B~i∗(k),(8)\frac{d\rho_B}{d\ln k}=4\pi k^3\widetilde{B}_i(k)\widetilde{B}_i^*(k), \tag*{(8)}

where B~i(k)B~i∗(k)=⟨B~i(k)B~i∗(k)⟩∣k∣=k\widetilde{B}_i(k)\widetilde{B}_i^*(k)=\langle\widetilde{B}_i(k)\widetilde{B}_i^*(k)\rangle_{|k|=k}, and Bi(k)B_i(k) is the Fourier transform of Bi(x)B_i(x).

Also, we focus on the gravitational waves (GWs) generated during the phase transition and the subsequent contribution from monopoles after the phase transition is complete. The evolution of the GW field is governed by by the linearized Einstein equation in the transverse-traceless (TT) gauge:

hij′′+2a′ahij′−∇2hij=a22mp2ΠijTT,(9)h_{ij}''+2\frac{a'}{a}h_{ij}'-\nabla^2h_{ij}=a^2\frac{2}{m_{\mathrm{p}}^2}\Pi_{ij}^{\mathrm{TT}}, \tag*{(9)}

where hijh_{ij} denotes the TT component of the metric perturbation. The source term ΠijTT\Pi_{ij}^{\mathrm{TT}} is the TT part of the anisotropic stress tensor Πij=a−2{DiϕaDjϕa−(EiaEja+BiaBja)/a2}\Pi_{ij}=a^{-2}\{D_i\phi^aD_j\phi^a-(E_i^aE_j^a+B_i^aB_j^a)/a^2\} [34]. Here, mp=(8πG)−1/2≃2.44×1018m_{\mathrm{p}}=(8\pi G)^{-1/2}\simeq2.44\times10^{18} GeV is the reduced Planck mass. The energy density of the GWs is given by ρGW(τ)=mp232πa2⟨hij′hij′⟩\rho_{\mathrm{GW}}(\tau)=\frac{m_{\mathrm{p}}^2}{32\pi a^2}\langle h_{ij}'h_{ij}'\rangle, where ⟨⋯ ⟩\langle\cdots\rangle denotes the volume average. The GW energy spectrum is then defined as:

ΩGW(k,τ)=1ρc(τ)dρGW(τ)dln⁡k,(10)\Omega_{\mathrm{GW}}(k,\tau)=\frac{1}{\rho_c(\tau)}\frac{\mathrm{d}\rho_{\mathrm{GW}}(\tau)}{\mathrm{d}\ln k}, \tag*{(10)}

where ρc=3H2mp2\rho_c=3H^2m_{\mathrm{p}}^2 is the critical energy density of the universe.

Simulation setup

In this section, we present the initialization, the modified Press-Ryden-Spergel (PRS) method for “growing” the size of the monopole core, as well as parameter setup for the lattice simulation.

Initialization

We numerically solve the complete set of equations of motion for the scalar field (4) and gauge field (5) on a three-dimensional lattice with periodic boundary conditions. Initially, we instantaneously introduce a fixed number of NbN_b vacuum bubbles onto the lattice, while the gauge fields and their conjugate momenta are both set to zero. The critical bubble profile is obtained by using FindBounce [35]

ϕa(r)=v2[1−tanh⁡(r−R0lw)]ηa,(11)\phi^a(r)=\frac{v}{2}\left[1-\tanh\left(\frac{r-R_0}{l_w}\right)\right]\eta^a, \tag*{(11)}
ϕ˙a(r)=0,(12)\dot{\phi}^a(r)=0, \tag*{(12)}

where ηa\eta^a is a random unit vector satisfying ηaηa=1\eta^a\eta^a=1, rr is the radial distance from the bubble center, R0R_0 is the initial bubble radius, and lwl_w denotes the bubble wall thickness. These bubbles subsequently expand and coalesce until the entire simulation volume has transitioned to the true vacuum. Before the phase transition is complete, we first turn off the expansion of the universe and introduce damping using the Crank-Nicolson method [36] with σ=1.5\sigma=1.5 (the detailed implementation is provided in the Supplemental Material). The purpose of this is to allow the scalar field configuration in the box to transition stably from bubbles to monopoles, facilitating the subsequent evolution of the monopole network. When the phase transition is completed (ϕaϕa∼1\phi^a\phi^a\sim1), we turn on the expansion and simultaneously enter the “core growth” phase of the monopoles.

PRS Method

When performing simulations in an expanding universe, the shrinking of the comoving size of monopoles must be taken into account. In general, the physical size of a monopole wM=1/mHw_M=1/m_H is determined by the mass of the scalar field mH=2λvm_H=\sqrt{2\lambda}v (or the massive boson mW=gvm_W=gv) and is independent of the scale factor aa. Meanwhile, the lattice spacing Δx\Delta x is comoving, with the physical grid spacing being aΔxa\Delta x. Consequently, the number of lattice sites occupied by a monopole, wM/(aΔx)=(amHΔx)−1∝a−1w_M/(a\Delta x)=(am_H\Delta x)^{-1}\propto a^{-1}, will decrease as aa increases. When wM/(aΔx)<1w_M/(a\Delta x)<1, the dynamics of the monopoles become unreliable. Regarding this issue, we employ an improved PRS method [37] adapted for gauge theories [38], which involves the topological defects—in our case, magnetic monopoles—to a “core growth” stage before physical evolution. This “grow” their physical core size and enables them to occupy multiple lattice sites. Concretely, this “core growth” stage is implemented by allowing the monopole core width wM=1/2λvw_M=1/\sqrt{2\lambda}v to “grow” with the scale factor aa. Keeping the VEV vv invariant, this requirement dictates that the coupling parameters λ\lambda and gg decrease as aa increases. We elaborate on the exact implementation below. After the phase transition completes, the expansion begins, and the system simultaneously enters the “core growth” stage of the monopoles. During this period, the coupling constants λ\lambda and gg depend on aa as follows:

λ=λ0a−6,g=g0a−3,(13)\lambda=\lambda_0a^{-6},\qquad g=g_0a^{-3}, \tag*{(13)}

where λ0\lambda_0 and g0g_0 are constant. Thereby, as aa increases, the number of lattice sites occupied by a monopole also increases, wM/(aΔx)∝a2w_M/(a\Delta x)\propto a^2, thereby realizing the “core growth” of the monopoles. The advantage of this approach is that it both fixes the VEV of the scalar field and keeps the mass ratio between the scalar field and the massive bosons mH/mWm_H/m_W constant, thereby keeping the monopole configuration stable. It should be emphasized that Ref. [33, 39] generally uses λ=λ0a−4\lambda=\lambda_0a^{-4} (therefore wM/(aΔx)∝aw_M/(a\Delta x)\propto a), whereas in our simulations, the initial λ0\lambda_0 is chosen to be larger to make the bubble radius relatively smaller, thereby allowing more bubbles to fit inside the simulation box. At the onset of expansion, the monopole size is actually less than one lattice spacing. Therefore, to “grow” the monopoles within a shorter period of time, we have chosen a more aggressive wM/(aΔx)∝a2w_M/(a\Delta x)\propto a^2. The “core growth” stage lasts for 1300 time steps, after which λ\lambda and gg are fixed at their values at that time until the end of the simulation, a period that can be referred to as the physical evolution stage. Figure 1 shows the division of the simulation into the phase transition stage (yellow region), the “core growth” stage (green region), and the physical evolution stage (remaining white region), as well as the time evolution of the number of lattice sites occupied by a monopole wM/(aΔx)w_M/(a\Delta x), the scale factor aa, and the coupling constants λ\lambda and gg. At the end of the simulation, a monopole still occupies more than one lattice site. Since the number of bubbles varies, the time at which the phase transition completes also differs. For convenience in presenting the data, we define the moment when the phase transition ends and expansion begins as τ=τ0=0\tau= \tau_0 = 0.

Figure 1. The time evolution of the number of lattice sites occupied by a monopole wM/(aΔx)w_M/(a\Delta x), the scale factor aa, and the coupling constants λ\lambda and gg. We set the onset of expansion as the time τ=0\tau= 0. We apply smoothing functions to λ\lambda and gg when entering and exiting the “core growth” stage to make the transitions smoother. The yellow region corresponds to the phase transition stage without expansion, the green region represents the monopole “core growth” stage, and the remaining white region denotes the physical evolution stage.

When simulating an expanding universe, the number of Hubble volumes contained within the simulation box gradually decreases. We terminate the simulation when only a single Hubble volume remains inside the box.

Parameter Setup

For all simulations, the lattice box size is N3=5123N^3 = 512^3, the lattice spacing is Δx=0.5\Delta x = 0.5, and the time step is Δτ=0.05\Delta\tau= 0.05. This choice of parameters provides sufficient resolution to capture all the dynamical details of both the bubbles and the monopoles. If a smaller lattice spacing Δx\Delta x were adopted, individual bubbles would occupy an excessively large number of lattice sites, preventing the simulation box from containing a large number of bubbles. Conversely, if Δx\Delta x were larger, it would be more difficult to identify the bubble walls and monopoles during the evolution, leading to a loss of dynamical details. At the beginning of the PRS stage, our simulation box encompasses 64 Hubble volumes. After evolving for 3840 time steps, only a single Hubble volume remains within the simulation box.

For each parameter configuration listed in Table 1, we perform three independent realizations and report the ensemble-averaged results.

vvvBv_Bg0g_0λ0\lambda_0R0R_0lwl_wNbN_b
10.353710Δx10\Delta x2Δx2\Delta x100
10.353710Δx10\Delta x2Δx2\Delta x250
10.353710Δx10\Delta x2Δx2\Delta x500
10.353710Δx10\Delta x2Δx2\Delta x750
10.353710Δx10\Delta x2Δx2\Delta x1000

Table 1. Parameters setup in the simulations.

Numerical Results

In this section, we investigate the evolution of the monopole network following the completion of the phase transition, along with the resulting magnetic fields and gravitational waves. Detailed results regarding the phase transition dynamics itself are provided in the Supplemental Material.

Evolution of Monopole Anti-Monopole Network

The top panel of Figure 2 displays the spatial distributions of the magnetic charges at the conclusion of the simulation for an initial bubble number of Nb=500N_b = 500. For the purpose of clear visualization, the (anti)monopoles in the top panel have been magnified by a factor of 216. The simulated volume contains 92 monopoles (highlighted in red) and 92 antimonopoles (highlighted in blue) at the end of the PRS stage. These topological defects carry magnetic charges of ±eM\pm e_M, respectively, which strictly satisfy the conservation law of zero net magnetic charge inside the box. The bottom panel of Figure 2 displays the spatial distribution of the energy density at the end of the PRS stage. Compared with the top panel, it can be seen that the energy is concentrated inside the (anti)monopoles. However, there are also some regions, such as the middle right of the box, where the energy density is high but there is no magnetic charge; these are precisely the regions where monopoles and antimonopoles have annihilated.

3D spatial distribution snapshots of the magnetic charge and energy density configuration

Figure 2. 3D spatial distribution snapshots of the magnetic charge (Top) and the energy density configuration (Bottom) at the conclusion of the PRS stage for an initial bubble number of Nb=500N_b = 500. For clarity of visualization, the (anti)monopoles in the top panel have been magnified by a factor of 216. Each monopole carries a positive magnetic charge of +1+1 (indicated in red), while each antimonopole carries a negative charge of −1-1 (indicated in blue).

Next, we explore the evolutionary dynamics of the monopole-antimonopole pairs produced during the phase transition. Crucially, since the system initially possesses no net magnetic charge, monopoles and antimonopoles are produced in exact symmetry, ensuring that the net topological charge remains strictly zero throughout the evolution. These topological defects are generated via bubble collisions and subsequently undergo pairwise annihilation once the PT completes. The top panel of Figure 3 illustrates the time evolution of the comoving monopole separation dMd_M for various values of NbN_b. We observe that a larger NbN_b yields a higher abundance of both produced and surviving monopoles, thus a shorter separation. With the exception of the Nb=100N_b = 100 case, all other runs exhibit a similar growth in separation with time, signifying the scaling behavior. This behavior occurs because closely separated pairs have already annihilated, whereas the attraction between the remaining widely spaced monopoles and antimonopoles is effectively counterbalanced by the cosmological expansion, halting further annihilation. Consequently, these stable relics can potentially persist into the present-day universe. It is worth noting that at the very beginning of the PRS stage, the number of monopoles exhibits a plateau. This is because we initially choose relatively large values of λ\lambda and gg, making the monopole size smaller than one lattice spacing Δx\Delta x, which causes the monopoles to be “pinned” to the lattice site and unable to move freely. As the PRS stage progresses, λ\lambda and gg gradually decrease and the monopole size grows, allowing them to move within

Figure 3. Top: Time evolution of the comoving monopole separation dMd_M for various bubble numbers NbN_b, plotted from τ>0\tau> 0 immediately following the completion of the phase transition. A higher initial bubble density results in a larger population of surviving monopole pairs. Bottom: Number of remaining monopole-antimonopole pairs at the end of the simulation as a function of NbN_b. The red dashed line denotes the theoretical fit, and the error bars represent the 95% confidence intervals.

the box and annihilate due to mutual attraction, which leads to a decrease in the monopole count.

To further quantify this relationship, we extract the population of surviving monopole-antimonopole pairs at the end of the simulation for different values of NbN_b, as presented in the bottom panel of Figure 3. We find that as the number of bubbles increases, the monopoles exhibit a tendency toward saturation. We speculate that this is because below a certain distance, monopoles always annihilate very quickly. Furthermore, a larger number of bubbles leads to more densely generated monopoles. Due to their close proximity, these extra generated monopoles cannot survive for long. Therefore, we choose the following function for fitting:

NM=70.71[1−exp⁡(−0.13Nb)].(14)N_M = 70.71 \left[1 - \exp(-0.13N_b)\right]. \tag*{(14)}

The dependence of the monopole abundance on the bubble number is not particularly sensitive, as the number of bubbles per Hubble volume increases roughly from 1 to 10, the number of surviving monopole pairs rises from 10 to 60.

Magnetic Fields and GWs Power Spectra

In Figure 4, we present the generated MF spectra for different bubble numbers NbN_b. We find that the magnetic field spectra corresponding to different monopole numbers exhibit qualitative similarities: they all feature a broken power-law profile with comparable peak frequencies. In the low-frequency (low-kk) regime, the spectrum rises as k−1.2k^{-1.2}. On scales smaller than the monopole size (dotted line in Figure 4), the spectrum falls off rapidly with a slope of k−5k^{-5}, demonstrating the magnetic screening effect inside the monopoles. Furthermore, as the number of monopoles increases, the peak energy density of the magnetic field also grows.

Figure 4. Magnetic field spectra at the end of the simulation for various values of NMN_M. The black dotted line denotes the physical size of the monopoles k/a=2π/wMk/a = 2\pi/w_M at the final time step.

Subsequently, we investigate the GW signals generated by the monopole dynamics. Figure 5 displays the time evolution of the GW power spectrum for various values of NbN_b. We observe that, first, across different monopole numbers, the power spectra similarly exhibit

Figure 5. GW power spectra at the end of the simulation for various values of NMN_M. The physical scales corresponding to the monopole separation k/a=2π/dMk/a = 2\pi/d_M for different values of NMN_M are indicated by dashed lines of the matching colors.

a broken power-law behavior: rising as k3k^3 in the low-frequency (low-kk) regime (consistent with the predication in [40]), followed by a brief plateau, and falling off as k−3k^{-3} in the high-frequency (high-kk) regime. Second, the peak frequency of the power spectrum consistently occurs at the scale of the mean monopole separation dM=L(2NM)−1/3d_M = L(2N_M)^{-1/3} (dashed line in the figure). As the number of monopoles NMN_M increases, the separation decreases, shifting the spectral peak to the right (toward higher frequencies).

We also plot the time evolution of both the magnetic field and gravitational wave spectra for various NbN_b, which are provided in the Supplemental Material. From these figures, the effects of the PRS method and the expansion on the spectral wavenumbers can be clearly observed.

Cosmological constraints

Since searches for magnetic monopoles to date have yielded null results, they must either be extremely scarce or constitute dark matter that participates in no interactions other than gravity. We discuss below the constraints on both the monopole mass

mM=4πvgf(mHmW)=43.7v,(15)m_M = \frac{4\pi v}{g}f\left(\frac{m_H}{m_W}\right)=43.7v , \tag*{(15)}

and the phase transition temperature T∗T_\ast under each of these two scenarios. Here we take mH/mW=1.247m_H/m_W = 1.247 and f(1.247)≃1.273f(1.247)\simeq1.273 [41].

If magnetic monopoles are treated as visible matter, the most stringent limit arises from the Parker bound for nonrelativistic monopoles [42, 43] F<10−16 s−1cm−2sr−1F < 10^{-16}\,\mathrm{s}^{-1}\mathrm{cm}^{-2}\mathrm{sr}^{-1}. Assuming that monopoles move isotropically within intergalactic space or inside galaxies, and taking their typical velocity to be comparable to the galactic virial velocity vM∼10−3c≈3×107v_M\sim10^{-3}c\approx3\times10^7 cm/s, the contemporary monopole number density is given by [44]:

nM,0≲4πFvM=4.19×10−23 cm−3.(16)n_{M,0}\lesssim\frac{4\pi F}{v_M}=4.19\times10^{-23}\ \mathrm{cm}^{-3}. \tag*{(16)}

Then the energy density of monopoles is

ρM,0=nM,0mM≲1.83v×10−21 cm−3.(17)\rho_{M,0}=n_{M,0}m_M\lesssim1.83v\times10^{-21}\ \mathrm{cm}^{-3}. \tag*{(17)}

By substituting today’s critical energy density ρc,0/h2≈1.05×10−5 GeV/cm3\rho_{c,0}/h^2\approx1.05\times10^{-5}\ \mathrm{GeV}/\mathrm{cm}^3, we can calculate the present-day monopole abundance:

ΩPBh2=ρM,0ρc,0/h2≲1.74×10−16(v1 GeV).(18)\Omega_{PB}h^2=\frac{\rho_{M,0}}{\rho_{c,0}/h^2}\lesssim1.74\times10^{-16}\left(\frac{v}{1\ \mathrm{GeV}}\right). \tag*{(18)}

Meanwhile, under the hypothesis that monopoles account for dark matter with the observational constraint:

ΩDMh2≤0.12.(19)\Omega_{DM}h^2\leq0.12 . \tag*{(19)}

Next, we establish the relationship between the density parameter ΩMh2\Omega_M h^2 and the number of monopoles NMN_M. We assume the phase transition occurs at a critical temperature of T∗∼vT_* \sim v. Today’s relic abundance of monopoles is given by:

ΩMh2=YMs0mMρc,0/h2=1.42×10−45NM(v1 GeV)4=7.10×10−44(v1 GeV)4,(20)\begin{aligned} \Omega_M h^2 = \frac{Y_M s_0 m_M}{\rho_{c,0}/h^2} = 1.42 \times10^{-45} N_M \left(\frac{v}{1\ \mathrm{GeV}}\right)^4 \\ &= 7.10 \times10^{-44} \left(\frac{v}{1\ \mathrm{GeV}}\right)^4 , \tag*{(20)} \end{aligned}

where YM=nM/s=nM,0/s0Y_M = n_M/s = n_{M,0}/s_0 is the monopole-to-entropy ratio, s0s_0 is the today entropy density. According to our simulations, we take the number of surviving monopole pairs as NM=50N_M = 50.

Finally, comparing Eqs. (18) and (20), we find that under the assumption that monopoles constitute visible matter, their mass must satisfy mM≲2.18×1011 GeVm_M \lesssim2.18 \times10^{11}\ \mathrm{GeV}, with the corresponding phase transition temperature being approximately T∗≲4.98×109 GeVT_* \lesssim4.98 \times10^9\ \mathrm{GeV}. Comparing Eqs. (19) and (20), under the hypothesis that monopoles account for dark matter, their mass should be bounded by mM≲1.58×1012 GeVm_M \lesssim1.58 \times10^{12}\ \mathrm{GeV}, leading to a phase transition temperature of roughly T∗≲3.60×1010 GeVT_* \lesssim3.60 \times10^{10}\ \mathrm{GeV}.

For a non-helical, inhomogeneous primordial magnetic field generated during the phase transition at temperature T∗T_*, the field undergoes self-similar decaying magnetohydrodynamic (MHD) turbulence in the radiation-dominated plasma. Defining the comoving magnetic field as B~≡a2B\widetilde{B} \equiv a^2 B, the root-mean-square field strength decays as B~(τ)∝τ−5/7\widetilde{B}(\tau) \propto\tau^{-5/7} with conformal time τ\tau [45, 46]. This turbulent dissipation continues until recombination (Trec≈0.3 eVT_{\mathrm{rec}} \approx0.3\ \mathrm{eV}), after which the drop in ionization fraction freezes the fluid velocity and the magnetic field redshifts purely adiabatically. The physical root-mean-square field strength today, B0B_0, is therefore related to the initial field strength B∗B_* by

B0=B∗(a∗a0)2(τrecτ∗)−5/7≃B∗(a∗a0)2[(g∗(T∗)g∗(Trec))1/6T∗Trec]−5/7,(21)B_0 = B_* \left(\frac{a_*}{a_0}\right)^2 \left(\frac{\tau_{\mathrm{rec}}}{\tau_*}\right)^{-5/7} \simeq B_* \left(\frac{a_*}{a_0}\right)^2 \left[\left(\frac{g_*(T_*)}{g_*(T_{\mathrm{rec}})}\right)^{1/6}\frac{T_*}{T_{\mathrm{rec}}}\right]^{-5/7}, \tag*{(21)}

where g∗(T∗)g_*(T_*) denotes the effective relativistic energy degrees of freedom. For temperatures well above the electroweak scale, g∗(T∗)=106.75g_*(T_*) = 106.75. Substituting the results from Figure 4 and adopting v∼T∗∼1010 GeVv \sim T_* \sim10^{10}\ \mathrm{GeV}, the present-day magnetic field strength generated by monopoles is estimated to be B0∼4.68×10−23 GaussB_0 \sim4.68 \times10^{-23}\ \mathrm{Gauss}, which falls below the lower bounds, B0∼10−17 GaussB_0 \sim10^{-17}\ \mathrm{Gauss}, reported by Fermi-LAT [47, 48]. This indicates that the origin of cosmological magnetic fields cannot be attributed to monopoles, pointing instead to alternative magnetogenesis mechanisms.

The GW power spectrum generated during the radiation-dominated era, ΩGW,∗(f∗)\Omega_{\mathrm{GW},*}(f_*), is redshifted to the present day as the Universe expands. Because gravitational waves decouple immediately upon generation and propagate freely, their energy density scales as ρGW∝a−4\rho_{\mathrm{GW}} \propto a^{-4}. Combining this with the conservation of entropy in the cosmic plasma, s∝g∗sa3T3=consts \propto g_{*s}a^3T^3 = \mathrm{const}, the present-day amplitude ΩGW,0(f0)h2\Omega_{\mathrm{GW},0}(f_0)h^2 is related to the initial spectrum by

ΩGW,0(f0)h2=Ωrad,0h2(g∗(T0)g∗(T∗))(g∗s(T∗)g∗s(T0))4/3ΩGW,∗(f∗),(22)\Omega_{\mathrm{GW},0}(f_0)h^2 = \Omega_{\mathrm{rad},0}h^2 \left(\frac{g_*(T_0)}{g_*(T_*)}\right) \left(\frac{g_{*s}(T_*)}{g_{*s}(T_0)}\right)^{4/3} \Omega_{\mathrm{GW},*}(f_*), \tag*{(22)}

where Ωrad,0h2≈4.15×10−5\Omega_{\mathrm{rad},0}h^2 \approx4.15 \times10^{-5} represents the radiation relic density today, and g∗s(T∗)g_{*s}(T_*) denotes the effective relativistic entropy degrees of freedom at the transition temperature T∗T_*. Assuming g∗(T∗)=g∗s(T∗)≈106.75g_*(T_*) = g_{*s}(T_*) \approx106.75 for temperatures well above the electroweak scale, this relation reduces to the standard parameterized form:

ΩGW,0(f0)h2≃1.63×10−5ΩGW,∗(f∗).(23)\Omega_{\mathrm{GW},0}(f_0)h^2 \simeq1.63 \times10^{-5}\Omega_{\mathrm{GW},*}(f_*). \tag*{(23)}

Substituting the numerical results from Figure 5 and adopting v∼T∗∼1010 GeVv \sim T_* \sim10^{10}\ \mathrm{GeV} along with mp=2.44×1018 GeVm_p = 2.44 \times10^{18}\ \mathrm{GeV}, the present-day peak amplitude of the gravitational waves produced by monopoles is approximately ΩGWh2∼2.30×10−38\Omega_{\mathrm{GW}}h^2 \sim2.30 \times10^{-38}, which falls far below the sensitivity range of any current detector. This is consistent with the findings obtained in Ref. [49].

Conclusion and discussion

In this work, we numerically study the monopole generation from a cosmological first-order phase transition on a 3D lattice, and derive the relation between the number density of magnetic monopole pairs and the vacuum bubbles, and the associated generation of the MF and GW spectra. The constraints on the monopole number density from the Parker bound and the relic abundance all preclude the detection of GW signatures.

In order for the monopole abundance to satisfy current observational constraints, both its mass mMm_M and the temperature T∗T_* of the first-order phase transition must be stringently constrained to values far below the Planck mass mpm_p. Consequently, the gravitational waves generated by monopoles, when redshifted to the present day, would fall well below the sensitivity thresholds of O3-QAR, ALP [50], ARCAD [51], and GW-photon conversion in geomagnetic and galactic magnetic fields [52]. Regarding the magnetic field, owing to the magnetohydrodynamic (MHD) dissipation of non-helical magnetic fields, a higher phase transition temperature leads to a weaker residual field strength today. According to our model, the magnetic field strength generated by monopoles also falls below the observational sensitivity B0∼10−17 GaussB_0 \sim10^{-17}\ \mathrm{Gauss} established by Fermi-LAT [47, 48] observations of blazars. If we require the generated magnetic field to avoid being ruled out by this observational constraint, the phase transition temperature would need to satisfy T∗∼350T_\ast\sim350 GeV.

Regarding the dark matter relic abundance constraint, the relic density of monopoles produced via the Kibble-Zurek mechanism in a second-order phase transition incorporating monopole annihilation requires the phase transition temperature to satisfy T∗<106∼109T_\ast< 10^6 \sim10^9 GeV [53], which is stricter than our bound. Ref. [54, 55, 56] derived an analytical result incorporating monopole annihilation, yielding YM=2.64×10−15∝T∗/mpY_M = 2.64 \times10^{-15} \propto T_\ast/m_p. In comparison, our numerical simulation gives a much smaller value, YM=7.80×10−22∝(T∗/mp)3Y_M = 7.80 \times10^{-22} \propto(T_\ast/m_p)^3. Therefore, our simulations show that, compared with their results, monopoles produced during the FOPT are far less likely to account for the full dark matter relic abundance.

We further note that, in the presence of monopoles, axions or axion-like particles acquire an effective mass via the Witten effect [57], thereby modifying predictions for the dark matter relic abundance [58, 59]. Ref. [60] propose to suppress axion isocurvature fluctuations with the hidden monopole. It has also been observed that, if hidden monopoles constitute a fraction of dark matter, the effective mass can resolve the domain-wall problem [61, 62, 63]. Moreover, the H0H_0 tension may be alleviated through axion decay [64].

Acknowledgements

We thank Ke-Pan Xie for helpful discussion. L.B. is supported by the National Natural Science Foundation of China (NSFC) under Grants Nos. 12322505, 12547101. L.B. also acknowledges Chongqing Natural Science Foundation under Grant No. CSTB2024NSCQ-JQX0022. R.G.C. is supported by the National Natural Science Foundation of China Grants Nos.12588101, 12235019. The numerical calculations in this study were carried out on the ORISE Supercomputer.

References

Equations of motion on lattice

Firstly, we introduce the link fields UiU_i to describe the gauge fields on the lattice

Ui(τ,x)=exp⁡[−igΔxWi(τ,x)]=exp⁡[−igΔxσa2Wia(τ,x)].(24)U_i(\tau,x)=\exp[-ig\Delta x W_i(\tau,x)]=\exp\left[-ig\Delta x\frac{\sigma^a}{2}W_i^a(\tau,x)\right]. \tag*{(24)}

Since we adopt the temporal gauge, we have W0=0W_0=0. The spatial covariant derivatives are

Diϕ(τ,x)=1Δx[Ui(τ,x)ϕ(τ,x+i)Ui†(τ,x)−ϕ(τ,x)].(25)D_i\phi(\tau,x)=\frac{1}{\Delta x}\left[U_i(\tau,x)\phi(\tau,x+i)U_i^\dagger(\tau,x)-\phi(\tau,x)\right]. \tag*{(25)}

We define ±0\pm0 to denote a forward or backward step Δτ\Delta\tau in the temporal direction, and ±i\pm i to denote a forward or backward step Δx\Delta x along the spatial ii-direction. We introduce the conjugate momentum fields π\pi and FiF_i corresponding to ϕ\phi and UiU_i, respectively. Temporally, ϕ\phi and UiU_i are defined at integer time steps, whereas π\pi and FiF_i are defined at half-integer time steps. Spatially, ϕ\phi and π\pi reside on the lattice sites, while UiU_i and FiF_i are defined on the links between adjacent sites.

Prior to evolving the equations of motion, we must address the time dependence of the gauge coupling gg introduced by the PRS prescription. As evident from Eq. (24), gg enters directly into the definition of the link variable UiU_i, as well as the conjugate momenta FiF_i. Consequently, at the beginning of each time step, we first “align” the coupling parameter via:

Ui,new(τ,x)=−iσa2g(τ)g(τ−0)tr⁡[iσaUi,old(τ,x)].(26)U_{i,\mathrm{new}}(\tau,x)=-i\frac{\sigma^a}{2}\frac{g(\tau)}{g(\tau-0)}\operatorname{tr}[i\sigma^a U_{i,\mathrm{old}}(\tau,x)]. \tag*{(26)}
Fi,new(τ,x)=−iσa2g(τ)g(τ−0)tr⁡[iσaFi,old(τ,x)].(27)F_{i,\mathrm{new}}(\tau,x)=-i\frac{\sigma^a}{2}\frac{g(\tau)}{g(\tau-0)}\operatorname{tr}[i\sigma^a F_{i,\mathrm{old}}(\tau,x)]. \tag*{(27)}

The time evolution within the current time step is then carried out using Ui,newU_{i,\mathrm{new}} and Fi,new(τ,x)F_{i,\mathrm{new}}(\tau,x). For notational convenience, we drop the subscript ’new’ hereafter. Regarding the coupling parameter λ\lambda, because it enters exclusively through the potential term, no special treatment is required. Then, employing the leapfrog algorithm, we first update ϕ\phi and UiU_i using the following equations:

ϕ(τ+0,x)=ϕ(τ,x)+a−2π(τ,x)Δτ,(28)\phi(\tau+0,x)=\phi(\tau,x)+a^{-2}\pi(\tau,x)\Delta\tau, \tag*{(28)}
Ui(τ+0,x)=g(τ)ΔτΔxFi(τ,x)Ui(τ,x)(29)U_i(\tau+0,x)=g(\tau)\Delta\tau\Delta x F_i(\tau,x)U_i(\tau,x) \tag*{(29)}

The above repeated indices are not summed up. Subsequently, the conjugate momentum fields π\pi and FiF_i are updated via:

tr⁡[π(τ,x)σa]=cCNtr⁡[π(τ−0,x)σa]+Δτeff{tr⁡[a2Δx2∑i[Ui(τ,x)ϕ(τ,x+i)Ui†(τ,x)−2ϕ(τ,x)+Ui†(τ,x−i)ϕ(τ,x−i)Ui(τ,x−i)]σa]−a4∂V∂ϕ}.(30)\operatorname{tr}[\pi(\tau,x)\sigma^a]=c_{\mathrm{CN}}\operatorname{tr}[\pi(\tau-0,x)\sigma^a]+\Delta\tau_{\mathrm{eff}}\left\{\operatorname{tr}\left[\frac{a^2}{\Delta x^2}\sum_i\left[U_i(\tau,x)\phi(\tau,x+i)U_i^\dagger(\tau,x)-2\phi(\tau,x)+U_i^\dagger(\tau,x-i)\phi(\tau,x-i)U_i(\tau,x-i)\right]\sigma^a\right]-a^4\frac{\partial V}{\partial\phi}\right\}. \tag*{(30)}
tr⁡[iσaFi(τ,x)]=cCNtr⁡[iσaFi(τ−0,x)]+ΔτeffRe⁡{a2Δx2tr⁡[g(τ)Ui(τ,x)ϕ(τ,x+i)Ui†(τ,x)iσaϕ(τ,x)]−1g(τ)Δx3tr⁡∑jiσa[Ui(τ,x)Uj(τ,x+k)Ui†(τ,x+j)Uj†(τ,x)+Ui(τ,x)Uj†(τ,x−j+i)Ui†(τ,x−j)Uj(τ,x−j)]}.(31)\operatorname{tr}[i\sigma^a F_i(\tau,x)]=c_{\mathrm{CN}}\operatorname{tr}[i\sigma^a F_i(\tau-0,x)]+\Delta\tau_{\mathrm{eff}}\operatorname{Re}\left\{\frac{a^2}{\Delta x}2\operatorname{tr}\left[g(\tau)U_i(\tau,x)\phi(\tau,x+i)U_i^\dagger(\tau,x)i\sigma^a\phi(\tau,x)\right]-\frac{1}{g(\tau)\Delta x^3}\operatorname{tr}\sum_j i\sigma^a\left[U_i(\tau,x)U_j(\tau,x+k)U_i^\dagger(\tau,x+j)U_j^\dagger(\tau,x)+U_i(\tau,x)U_j^\dagger(\tau,x-j+i)U_i^\dagger(\tau,x-j)U_j(\tau,x-j)\right]\right\}. \tag*{(31)}

where

cCN=2−σΔτ2+σΔτ,Δτeff=2Δτ2+σΔτ,(32)c_{\mathrm{CN}}=\frac{2-\sigma\Delta\tau}{2+\sigma\Delta\tau},\qquad\Delta\tau_{\mathrm{eff}}=\frac{2\Delta\tau}{2+\sigma\Delta\tau}, \tag*{(32)}

are the Crank-Nicolson coefficient and effective time interval, respectively. While the phase transition is ongoing, we set the damping coefficient σ=1.5\sigma=1.5 to allow the system to transition smoothly from a bubble configuration to a monopole configuration. Once the phase transition completes, we set σ=0\sigma=0, at which point cCN=1c_{\mathrm{CN}}=1, Δτeff=Δτ\Delta\tau_{\mathrm{eff}}=\Delta\tau. This method ensures that the Gauss constraint,

Ga(τ,x)=1Δxtr⁡∑iiσa[Fi(τ,x)−Ui†(τ,x−i)Fi(τ,x−i)Ui(τ,x−i)]−2tr⁡[gπ(τ,x)iσaϕ(τ,x)],(33)G^a(\tau,x)=\frac{1}{\Delta x}\operatorname{tr}\sum_i i\sigma^a\left[F_i(\tau,x)-U_i^\dagger(\tau,x-i)F_i(\tau,x-i)U_i(\tau,x-i)\right]-2\operatorname{tr}[g\pi(\tau,x)i\sigma^a\phi(\tau,x)], \tag*{(33)}

remains well satisfied throughout the simulation.

Identify monopoles on lattice

After the SU (2) symmetry is broken, under unitary gauge, the diagonal term of the gauge field is exactly the gauge field corresponding to the remaining U(1) symmetry. By using the projection operator [39, 65, 66]

P±=12(12×2±σ3),(34)P_{\pm}=\frac{1}{2}\left(\mathbb{1}_{2\times2}\pm\sigma^3\right), \tag*{(34)}

we can obtain the U(1) link fields under unitary gauge

V~i,±(τ,x)=P±U~i(τ,x)P±.(35)\widetilde{V}_{i,\pm}(\tau,x)=P_{\pm}\widetilde{U}_i(\tau,x)P_{\pm}. \tag*{(35)}

Only the upper left (lower right) element of V~i,+\widetilde{V}_{i,+} (V~i,−\widetilde{V}_{i,-}) is non-zero. The theory (1) is invariant under gauge transformations

ϕ(τ,x)→Λ(τ,x)ϕ(τ,x)Λ†(τ,x),(36)\phi(\tau,x)\to\Lambda(\tau,x)\phi(\tau,x)\Lambda^\dagger(\tau,x), \tag*{(36)}
Ui(τ,x)→Λ(τ,x)Ui(τ,x)Λ†(τ,x+i).(37)U_i(\tau,x)\to\Lambda(\tau,x)U_i(\tau,x)\Lambda^\dagger(\tau,x+i). \tag*{(37)}

So, for any SU(2) triplet under any gauge, the following specification transformation matrix R(τ,x)R(\tau,x) can transform ϕ^\hat{\phi} at any point into σ3\sigma^3.

R(τ,x)∝i[σ3+ϕ^(τ,x)],(38)R(\tau,x)\propto i[\sigma^3+\hat{\phi}(\tau,x)], \tag*{(38)}

where ϕ^=2ϕ/tr⁡(ϕϕ)\hat{\phi}=\sqrt{2}\phi/\sqrt{\operatorname{tr}(\phi\phi)} is the unit SU(2) triplet field.

ϕ^→RϕR†=(σ3+ϕ^)ϕ^(σ3+ϕ^)=σ3ϕ^σ3+ϕ^ϕ^ϕ^+σ3ϕ^ϕ^+ϕ^ϕ^σ3=(−ϕ^1σ1−ϕ^2σ2+ϕ^3σ3)+(ϕ^1σ1+ϕ^2σ2+ϕ^3σ3)+2σ3=2(1+ϕ^3)σ3∝σ3,(39)\begin{aligned} \hat{\phi}&\to R\phi R^\dagger\\ &=(\sigma^3+\hat{\phi})\hat{\phi}(\sigma^3+\hat{\phi})\\ &=\sigma^3\hat{\phi}\sigma^3+\hat{\phi}\hat{\phi}\hat{\phi}+\sigma^3\hat{\phi}\hat{\phi}+\hat{\phi}\hat{\phi}\sigma^3\\ &=(-\hat{\phi}^1\sigma^1-\hat{\phi}^2\sigma^2+\hat{\phi}^3\sigma^3)+(\hat{\phi}^1\sigma^1+\hat{\phi}^2\sigma^2+\hat{\phi}^3\sigma^3)+2\sigma^3\\ &=2(1+\hat{\phi}^3)\sigma^3\\ &\propto\sigma^3, \tag*{(39)} \end{aligned}

here we use ϕ^ϕ^=1\hat{\phi}\hat{\phi}=\mathbb{1}. So, for SU(2) gauge field Ui(τ,x)U_i(\tau,x) under any gauge, we can obtain the remaining U(1) gauge field Vi(τ,x)V_i(\tau,x) under unitary gauge through P±P_{\pm} and R(τ,x)R(\tau,x):

V~i,±(τ,x)=P±R(τ,x)Ui(τ,x)R†(τ,x+i)P±∝14(1±σ3)[σ3+ϕ^(τ,x)]Ui(τ,x)[σ3+ϕ^(τ,x+i)](1±σ3)=14[σ3+ϕ^(τ,x)±1±σ3ϕ^(τ,x)]Ui(τ,x)[σ3+ϕ^(τ,x)±1±ϕ^(τ,x)σ3]=[σ3+ϕ^(τ,x)]12[1+ϕ^(τ,x)]Ui(τ,x)12[1+ϕ^(τ,x+i)][σ3+ϕ^(τ,x+i)]∝R(τ,x)Π±(τ,x)Ui(τ,x)Π±(τ,x+i)R†(τ,x+i)=R(τ,x)Vi,±(τ,x)R†(τ,x+i).(40)\begin{aligned} \widetilde{V}_{i,\pm}(\tau,x)&=P_{\pm}R(\tau,x)U_i(\tau,x)R^\dagger(\tau,x+i)P_{\pm}\\ &\propto\frac{1}{4}(\mathbb{1}\pm\sigma^3)[\sigma^3+\hat{\phi}(\tau,x)]U_i(\tau,x)[\sigma^3+\hat{\phi}(\tau,x+i)](\mathbb{1}\pm\sigma^3)\\ &=\frac{1}{4}[\sigma^3+\hat{\phi}(\tau,x)\pm\mathbb{1}\pm\sigma^3\hat{\phi}(\tau,x)]U_i(\tau,x)[\sigma^3+\hat{\phi}(\tau,x)\pm\mathbb{1}\pm\hat{\phi}(\tau,x)\sigma^3]\\ &=[\sigma^3+\hat{\phi}(\tau,x)]\frac{1}{2}[\mathbb{1}+\hat{\phi}(\tau,x)]U_i(\tau,x)\frac{1}{2}[\mathbb{1}+\hat{\phi}(\tau,x+i)][\sigma^3+\hat{\phi}(\tau,x+i)]\\ &\propto R(\tau,x)\Pi_{\pm}(\tau,x)U_i(\tau,x)\Pi_{\pm}(\tau,x+i)R^\dagger(\tau,x+i)\\ &=R(\tau,x)V_{i,\pm}(\tau,x)R^\dagger(\tau,x+i). \tag*{(40)} \end{aligned}

Now, we have the U(1) link fields under any gauge

Vi,±(τ,x)=Π±(τ,x)Ui(τ,x)Π±(τ,x+i),(41)V_{i,\pm}(\tau,x)=\Pi_{\pm}(\tau,x)U_i(\tau,x)\Pi_{\pm}(\tau,x+i), \tag*{(41)}

where the projectors are

Π±(τ,x)=12[12×2±ϕ^(τ,x)],Π±2=Π±.(42)\Pi_{\pm}(\tau,x)=\frac{1}{2}[\mathbb{1}_{2\times2}\pm\hat{\phi}(\tau,x)],\qquad\Pi_{\pm}^2=\Pi_{\pm}. \tag*{(42)}

With the link field, we can define a plaquette

Vij,±(τ,x)=Vi,±(τ,x)Vj,±(τ,x+i)Vi,±†(τ,x+j)Vj,±†(τ,x).(43)V_{ij,\pm}(\tau,x)=V_{i,\pm}(\tau,x)V_{j,\pm}(\tau,x+i)V^\dagger_{i,\pm}(\tau,x+j)V^\dagger_{j,\pm}(\tau,x). \tag*{(43)}

Here Vij,+(τ,x)V_{ij,+}(\tau,\boldsymbol{x}) is still a matrix, but only the upper left element is non-zero. The relationship between its trace and the U(1) gauge field strength Aij(τ,x)A_{ij}(\tau,\boldsymbol{x}) is

tr⁡Vij,+(τ,x)=exp⁡[igΔx2Aij(τ,x)/2].(44)\operatorname{tr} V_{ij,+}(\tau,\boldsymbol{x}) = \exp[ig\Delta x^2 A_{ij}(\tau,\boldsymbol{x})/2]. \tag*{(44)}

Therefore, the simplest way is to take the argument to obtain the U(1) gauge field strength, which is the magnetic field BiB_i,

Aij(τ,x)=2gΔx2arg⁡tr⁡Vij,+(τ,x),(45)A_{ij}(\tau,\boldsymbol{x}) = \frac{2}{g\Delta x^2} \arg\operatorname{tr} V_{ij,+}(\tau,\boldsymbol{x}), \tag*{(45)}
Bi(τ,x)=12ϵijkAjk(τ,x).(46)B_i(\tau,\boldsymbol{x}) = \frac{1}{2}\epsilon_{ijk}A_{jk}(\tau,\boldsymbol{x}). \tag*{(46)}

Finally, utilizing Gauss’s theorem ∇⋅B=ρM\nabla\cdot\boldsymbol{B} = \rho_M, we get the monopole density

1Δx∑i[Bi(τ,x+i)−Bi(τ,x)]=ρM,M‾(τ,x)=4πgnM,M‾(τ,x),(47)\frac{1}{\Delta x}\sum_i [B_i(\tau,\boldsymbol{x}+\boldsymbol{i})-B_i(\tau,\boldsymbol{x})] = \rho_{M,\overline{M}}(\tau,\boldsymbol{x}) = \frac{4\pi}{g}n_{M,\overline{M}}(\tau,\boldsymbol{x}), \tag*{(47)}

where g/4πg/4\pi is the magnetic charge of a ’t Hooft-Polyakov monopole [67]. n(τ,x)n(\tau,\boldsymbol{x}) represents the monopole (or antimonopole if it is negative) number density in a lattice site Δx3\Delta x^3. The total number of monopoles and antimonopoles in a lattice site is

NM(τ,x)=nM(τ,x)Δx3,nM>0,(48)N_M(\tau,\boldsymbol{x}) = n_M(\tau,\boldsymbol{x})\Delta x^3,\quad n_M > 0, \tag*{(48)}
NM‾(τ,x)=nM‾(τ,x)Δx3,nM‾<0,(49)N_{\overline{M}}(\tau,\boldsymbol{x}) = n_{\overline{M}}(\tau,\boldsymbol{x})\Delta x^3,\quad n_{\overline{M}} < 0, \tag*{(49)}

which are both integers. Sum up the number of monopoles and antimonopoles in each lattice point separately to obtain the total number of monopoles and antimonopoles, and the sum of the two is always 0 because there are no monopoles or antimonopoles in the lattice at the beginning of the simulation.

NM(τ)=∑xNM(τ,x),NM‾(τ)=∑xNM‾(τ,x),NM(τ)+NM‾(τ)≡0.(50)N_M(\tau) = \sum_{\boldsymbol{x}} N_M(\tau,\boldsymbol{x}),\quad N_{\overline{M}}(\tau) = \sum_{\boldsymbol{x}} N_{\overline{M}}(\tau,\boldsymbol{x}),\quad N_M(\tau)+N_{\overline{M}}(\tau)\equiv0. \tag*{(50)}

The Phase Transition Progress

In this Appendix, we briefly demonstrate the dynamics of the FOPT. At the onset of the simulation, we simultaneously nucleate NbN_b bubbles within the false-vacuum ϕaϕa=0\phi^a\phi^a=0 background, with the initial mean bubble separation given by R∗=(V/Nb)1/3R_*=(V/N_b)^{1/3}, as depicted in the top-left panel of Figure 6. Evolving under the equations of motion (A5–A8), the bubbles expand, collide, and merge with one another, as illustrated in the top-right panel of Figure 6. Once the entire volume relaxes into the true vacuum ϕaϕa≈v\phi^a\phi^a\approx v, the phase transition terminates, as shown in the bottom-left panel of Figure 6. Throughout the phase transition, the expansion of the universe is neglected. Concurrently, due to our choice of relatively large initial couplings λ0\lambda_0 and g0g_0, the physical size of the monopoles at this stage is smaller than a single lattice spacing Δx\Delta x, causing the scalar field ϕ\phi at the monopole center to remain close to the true vacuum. Following the completion of the phase transition, the “core growth” stage of the PRS method commences. In this stage, λ\lambda and gg decrease as the scale factor aa increases, allowing the monopole cores to broaden and span across multiple lattice sites. The bottom-right panel displays the distribution of ϕaϕa\phi^a\phi^a at the conclusion of the “core growth” stage, where the false vacuum remains trapped at the monopole cores while the rest of space is occupied by the true vacuum.

Figure 6. Two-dimensional spatial slices of the scalar field modulus squared ϕaϕa\phi^{a}\phi^{a} at four representative epochs for a total bubble number Nb=500N_b = 500: bubble nucleation (top left), during the phase transition (top right), at the completion of the phase transition (bottom left), and at the end of the “core growth” stage (bottom right). The color scale indicates the vacuum state, with blue denoting the false vacuum (ϕaϕa=0\phi^{a}\phi^{a} = 0) and red representing the true vacuum (ϕaϕa≈v\phi^{a}\phi^{a} \approx v). The bottom left panel presents the deviation of ϕaϕa\phi^{a}\phi^{a} from “+1”. In the bottom right panel, the cores of the formed monopoles remain trapped in the false vacuum state (ϕaϕa=0\phi^{a}\phi^{a} = 0), while the surrounding background has fully relaxed into the true vacuum.

Figure 7 illustrates the temporal evolution of ϕaϕa\phi^a\phi^a (Left panel) and the monopole number NMN_M (Right panel) throughout the phase transition. It should be emphasized again that the onset of the expansion is defined as τ=0\tau=0; consequently, the phase transition proceeds entirely in the regime τ<0\tau<0. Because of damping σ=1.5\sigma=1.5, the progression of the phase transition is slowed down; typically, the transition completes around τ∼R∗\tau\sim R_*. Following the nucleation of monopoles via bubble collisions, closely separated monopole–antimonopole pairs rapidly annihilate, leading to an initial reduction in NMN_M. However, because we adopt relatively large initial couplings λ\lambda and gg, the monopole core size remains smaller than a single lattice spacing Δx\Delta x. This severely restricts the free motion of the monopoles (lattice pinning), causing NMN_M to plateau near the end of the phase transition stage. It is not until the “core growth” stage broadens the monopole cores across multiple lattice sites that the monopoles regain mobility and resume annihilation.

Figure 7. Left: Time evolution of the squared modulus of the scalar field ϕ\phi for various bubble numbers NbN_b, characterizing the progression of the phase transition. Right: Time evolution of the monopole number NMN_M during the phase transition for different NbN_b. On the horizontal axis, R∗=(V/Nb)1/3R_* = (V/N_b)^{1/3} denotes the mean bubble separation. The expansion begins at τ=0\tau= 0, so the phase transition proceeds entirely in the regime τ<0\tau< 0.

The Time Evolution of the Power Spectrum

Figure 8 shows the time evolution of the magnetic field after the phase transition for various values of NbN_b. Aside from the evident gradual amplification of the magnetic field spectra, their peak positions shift left and right during expansion. This behavior originates from the distinct dependencies of the monopole core size wM=1/mHw_M = 1/m_H on the scale factor aa during the ’core growth’ stage versus the subsequent physical evolution stage. This effect will be seen even more clearly in the gravitational wave power spectra below.

Figure 8. Evolution of the magnetic field spectrum dρB/dln⁡k\mathrm{d}\rho_B/\mathrm{d}\ln k for different bubble numbers NbN_b. The color gradient from purple to yellow denotes the physical chronological evolution from earlier to later times. The orange and red solid lines highlight the spectrum at the end of the “core growth” stage and the whole simulation, respectively. The black solid, dashed, and dotted vertical lines mark the physical characteristic scales k/a=2π/R∗k/a = 2\pi/R_*, k/a=2π/dMk/a = 2\pi/d_M, and k/a=2π/wMk/a = 2\pi/w_M at the end of the simulation, respectively.

Figure 9 shows the time evolution of the magnetic field after the phase transition for various values of NbN_b. The effects of the PRS “core growth” stage and the expansion are considerably more pronounced in this figure than in the magnetic field spectrum of Figure 8. During the “core growth” stage, k/(amH)∝a2k/(a m_H) \propto a^2, causing the power spectrum curve to shift from left to right (from the purple curve to the orange curve). Once the “core growth” stage concludes, mHm_H ceases to vary while the universe continues to expand, giving k/(amH)∝a−1k/(a m_H) \propto a^{-1}; consequently, the power spectrum curve shifts from right to left (from the orange curve to the red curve).

Figure 9. Evolution of the gravitational wave power spectrum ΩGW\Omega_{\mathrm{GW}} for different bubble numbers NbN_b. The color gradient from purple to yellow denotes the chronological evolution from earlier to later times. The orange and red solid lines highlight the spectrum at the end of the “core growth” stage and the whole simulation, respectively. The black solid, dashed, and dotted vertical lines mark the physical characteristic scales k/a=2π/R∗k/a = 2\pi/R_*, k/a=2π/dMk/a = 2\pi/d_M, and k/a=2π/wMk/a = 2\pi/w_M at the end of the simulation, respectively.

References

  1. [1]G. ’t Hooft, Nucl. Phys. B 79, 276 (1974).
  2. [2]A. M. Polyakov, JETP Lett. 20, 194 (1974).
  3. [3]J. Preskill, Phys. Rev. Lett. 43, 1365 (1979).
  4. [4]A. H. Guth and E. J. Weinberg, Phys. Rev. D 23, 876 (1981).
  5. [5]A. D. Linde, Phys. Lett. B 108, 389 (1982).
  6. [6]A. H. Guth and S. H. H. Tye, Phys. Rev. Lett. 44, 631 (1980), [Erratum: Phys.Rev.Lett. 44, 963 (1980)].
  7. [7]S. R. Coleman, Phys. Rev. D 15, 2929 (1977), [Erratum: Phys.Rev.D 16, 1248 (1977)].
  8. [8]A. D. Linde, Nucl. Phys. B 216, 421 (1983), [Erratum: Nucl.Phys.B 223, 544 (1983)].
  9. [9]M. Hindmarsh, M. Lüben, J. Lumma, and M. Pauly, Sci-Post Physics Lecture Notes (2021), ISSN 2590-1990, URL http://dx.doi.org/10.21468/SciPostPhysLectNotes.24.
  10. [10]R. Caldwell et al., Gen. Rel. Grav. 54, 156 (2022), 2023.07926.
  11. [11]P. Athron, C. Balázs, A. Fowlie, L. Morris, and L. Wu, Progress in Particle and Nuclear Physics 135, 104094 (2024), ISSN 0146-6410, URL http://dx.doi.org/10.1016/j.ppnp.2023.104094.
  12. [12]T. W. B. Kibble, Journal of Physics A: Mathematical and General 9, 1387 (1976), URL https://doi.org/10.1088/0305-4470/9/8/029.
  13. [13]T. W. B. Kibble, Phys. Rept. 67, 183 (1980).
  14. [14]M. Hindmarsh, A.-C. Davis, and R. H. Brandenberger, Phys. Rev. D 49, 1944 (1994), hep-ph/9307203.
  15. [15]E. J. Copeland and P. M. Saffin, Phys. Rev. D 54, 6088 (1996), hep-ph/9604231.
  16. [16]T. W. B. Kibble and A. Vilenkin, Phys. Rev. D 52, 679 (1995), hep-ph/9501266.
  17. [17]C. J. Hogan, Phys. Rev. Lett. 51, 1488 (1983), URL https://link.aps.org/doi/10.1103/PhysRevLett.51.1488.DOI
  18. [18]Y. Di, J. Wang, R. Zhou, L. Bian, R.-G. Cai, and J. Liu, Physical Review Letters 126 (2021), ISSN 1079-7114, URL http://dx.doi.org/10.1103/PhysRevLett.126.251102.
  19. [19]K. Enqvist and P. Olesen, Phys. Lett. B 319, 178 (1993), hep-ph/9308270.
  20. [20]J. Yang and L. Bian, Physical Review D 106 (2022), ISSN 2470-0029, URL http://dx.doi.org/10.1103/PhysRevD.106.023510.
  21. [21]T. Vachaspati, Phys. Lett. B 265, 258 (1991).
  22. [22]G. Baym, D. Bodeker, and L. D. McLerran, Phys. Rev. D 53, 6246 (1996), hep-ph/9507429.
  23. [23]D. Grasso and A. Riotto, Phys. Lett. B 418, 258 (1998), hep-ph/9707265.
  24. [24]Y. Di, L. Bian, and R.-G. Cai (2024), 2409.16124.
  25. [25]Y. Di, L. Bian, and R.-G. Cai, Phys. Rev. D 113, 043529 (2026), 2508.07416.
  26. [26]N. Bhusal, S. Blasi, M. Cataldi, A. Chatrychyan, M. Gorghetto, and G. Servant, Phys. Rev. Lett. 136, 241803 (2026), URL https://link.aps.org/doi/10.1103/iv7q-vsbn.
  27. [27]A. Mazumdar and G. White, Reports on Progress in Physics 82, 076901 (2019), ISSN 1361-6633, URL http://dx.doi.org/10.1088/1361-6633/ab1f55.
  28. [28]C. Caprini et al., Journal of Cosmology and Astroparticle Physics 2016, 001–001 (2016), ISSN 1475-7516, URL http://dx.doi.org/10.1088/1475-7516/2016/04/001.
  29. [29]C. Caprini et al., Journal of Cosmology and Astroparticle Physics 2020, 024–024 (2020), ISSN 1475-7516, URL http://dx.doi.org/10.1088/1475-7516/2020/03/024.
  30. [30]R. Roshan and G. White, Rev. Mod. Phys. 97, 015001 (2025), 2401.04388.
  31. [31]H. Georgi and S. L. Glashow, Phys. Rev. Lett. 32, 438 (1974).
  32. [32]A. J. Creaco and N. Kalogeropoulos, Mod. Phys. Lett. A 23, 2053 (2008), 0805.2353.
  33. [33]M. Hindmarsh, A. Lopez-Eiguren, R. Seppä, and D. J. Weir (2025), 2511.14204.
  34. [34]A. Tranberg, S. Tähtinen, and D. J. Weir, JCAP 04, 012 (2018), 1706.02365.
  35. [35]V. Guada, D. Nemvešek, and M. Pintar, Computer Physics Communications 256, 107480 (2020), ISSN 0010-4655, URL http://dx.doi.org/10.1016/j.cpc.2020. 107480.
  36. [36]J. Crank and P. Nicolson, Adv Comput Math 6, 207–226 (1996).
  37. [37]W. H. Press, B. S. Ryden, and D. N. Spergel, Astrophys. J. 347, 590 (1989).
  38. [38]M. Bevis, M. Hindmarsh, M. Kunz, and J. Urrestilla, Physical Review D 82 (2010), ISSN 1550-2368, URL http://dx.doi.org/10.1103/PhysRevD.82.065004.
  39. [39]M. Hindmarsh, K. Rummukainen, and D. J. Weir, Physical Review D (2017), URL http://dx.doi.org/10.1103/physrevd.95.063520.
  40. [40]R.-G. Cai, S. Pi, and M. Sasaki, Phys. Rev. D 102, 083528 (2020), 1909.13728.
  41. [41]P. Forgács, N. Obadia, and S. Reuillon, Phys. Rev. D 71, 119902 (2005), URL https://link.aps.org/doi/10.1103/PhysRevD.71.119902.DOI
  42. [42]M. Ambrosio et al. (MACRO), Eur. Phys. J. C 25, 511 (2002), hep-ex/0207020.
  43. [43]E. N. Parker, Astrophys. J. 160, 383 (1970).
  44. [44]C. Zhang, S.-H. Zhang, B. Fu, J.-F. Zhang, and X. Zhang, Journal of High Energy Physics 2024 (2024), ISSN 1029-8479, URL http://dx.doi.org/10.1007/JHEP08(2024)220.
  45. [45]R. Durrer and A. Neronov, The Astronomy and Astrophysics Review 21 (2013), ISSN 1432-0754, URL http://dx.doi.org/10.1007/s00159-013-0062-7.
  46. [46]A. Brandenburg, R. Durrer, Y. Huang, T. Kahnishvili, S. Mandal, and S. Mukohyama, Physical Review D (2020), URL http://dx.doi.org/10.1103/physrevd.102.023536.
  47. [47]A. M. Taylor, I. Vovk, and A. Neronov, Astronomy & Astrophysics 529, A144 (2011), ISSN 1432-0746, URL http://dx.doi.org/10.1051/0004-6361/201116441.
  48. [48]M. Ackermann et al. (Fermi-LAT), Astrophys. J. Suppl. 237, 32 (2018), 1804.08035.
  49. [49]J. Yang, R. Zhou, and L. Bian, Phys. Lett. B 839, 137822 (2023), 2204.07540.
  50. [50]A. Ejlli, D. Ejlli, A. M. Cruise, G. Pisano, and H. Grote, The European Physical Journal C 79 (2019), ISSN 1434-6052, URL http://dx.doi.org/10.1140/epjc/s10052-019-7542-5.
  51. [51]V. Domcke and C. Garcia-Cely, Physical Review Letters 126 (2021), ISSN 1079-7114, URL http://dx.doi.org/10.1103/PhysRevLett.126.021104.
  52. [52]A. Ito, K. Kohri, and K. Nakayama, Progress of Theoretical and Experimental Physics 2024 (2024), ISSN 2050-3911, URL http://dx.doi.org/10.1093/ptep/ptae004.
  53. [53]H. Murayama and J. Shu, Physics Letters B 686, 162 (2010), ISSN 0370-2693, URL https://www.sciencedirect.com/science/article/pii/S0370269310002248.
  54. [54]J. P. Preskill, Phys. Rev. Lett. 43, 1365 (1979), URL https://link.aps.org/doi/10.1103/PhysRevLett.43.1365.DOI
  55. [55]F. Brümmer, G. Ferrante, T. Fischer, and M. Frigerio, Phys. Rev. D 113, L091701 (2026), URL https://link.aps.org/doi/10.1103/bttg-zfr3.DOI
  56. [56]F. Brümmer, G. Ferrante, T. Fischer, and M. Frigerio (2026), 2607.29492.
  57. [57]E. Witten, Phys. Lett. B 86, 283 (1979).
  58. [58]J. Fan, K. Fraser, M. Reece, and J. Stout, Phys. Rev. Lett. 127, 131602 (2021), 2105.09950.
  59. [59]M. Kawasaki, F. Takahashi, and M. Yamada, Phys. Lett. B 753, 677 (2016), 1511.05030.
  60. [60]Y. Nomura, S. Rajendran, and F. Sanches, Phys. Rev. Lett. 116, 141803 (2016), 1511.06347.
  61. [61]S. Nakagawa, F. Takahashi, and M. Yamada, Phys. Rev. Lett. 127, 181103 (2021), 2103.08153.
  62. [62]R. Sato, F. Takahashi, and M. Yamada, Phys. Rev. D 98, 043535 (2018), 1805.10533.
  63. [63]R. Zhou and L. Bian, Chin. Phys. C 49, 115110 (2025), 2501.09596.
  64. [64]R. Daido, S.-Y. Ho, and F. Takahashi, JHEP 01, 185 (2020), 1909.03627.
  65. [65]A. C. Davis, T. W. Kibble, A. Rajantie, and H. P. Shanahan, Journal of High Energy Physics 2000, 010 (2000), URL https://doi.org/10.1088/1126-6708/2000/11/010.
  66. [66]S. Edwards, D. Mehta, A. Rajantie, and L. von Smekal, Physical Review D 80 (2009), ISSN 1550-2368, URL http://dx.doi.org/10.1103/PhysRevD.80.065030.
  67. [67]G. Hooft, Nuclear Physics B 79, 276 (1974), ISSN 0550-3213, URL https://www.sciencedirect.com/science/article/pii/0550321374904866.

Paper details

vlnc:W_a2ztcme2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 12

downloads · 2

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents