betaFeedback

WorksPhysics and Astronomy

A Poisson-gas approach to medium-induced gluon radiation

Abstract

The precision era of jet-quenching phenomenology requires a quantitative control of medium-induced gluon radiation. We consider the BDMPS-Z spectrum in the soft-gluon limit for a Yukawa (Gyulassy-Wang) potential. Writing the Yukawa dipole cross section as a Laplace superposition of Gaussians, we express the fully resummed spectrum as an average over Poisson-distributed configurations of Gaussian scattering centers, each of which is solved in closed form. This yields a master formula for the spectrum both differential in the transverse momentum k⊥k_\perp and integrated with a cut on k⊥k_\perp, valid for static and expanding media with arbitrary profiles. Its evaluation by Monte Carlo sampling of the configurations is efficient and trivially parallelizable. The method reproduces the first order in opacity and published results without kinematic cuts, while with a cut we find sizeable relative deviations from previous results. We argue that the same construction extends to the more complex building blocks of an in-medium parton cascade, such as finite energy fraction and emissions off color antennas, and provides a natural starting point for analytical approximations. The code is publicly available.

Introduction

The analysis of hard probes in high-energy heavy-ion collisions is one of the main experimental handles on the properties of the quark–gluon plasma created in those collisions. One of the central theoretical tools to describe hard probes in such an environment is the medium-induced gluon radiation spectrum. With more precise measurements and new observables becoming available, a precise evaluation of this spectrum has become essential.

In this paper we consider the BDMPS-Z formalism [1, 2, 3, 4] for gluon radiation off an eikonal parton in the soft limit. In this description the medium acts as a color background field with which the jet interacts. We neglect recoil effects and assume the background field to be classical and to obey Gaussian statistics. The medium is then characterized by the two-point function

⟨Aa−(x+,x)Ab−(y+,y)⟩=δabn(x+)δ(x+−y+)γ(x−y)(1)\left\langle A_a^-(x^+, \mathbf{x}) A_b^-(y^+, \mathbf{y}) \right\rangle= \delta_{ab} n(x^+) \delta(x^+ - y^+) \gamma(\mathbf{x} - \mathbf{y}) \tag*{(1)}

where x and y are transverse positions, n(s)n(s) is the density of scattering centers at path length ss, and γ\gamma is the Fourier transform of the elastic scattering potential, whose screening mass μ(s)\mu(s) may also depend on ss. In the soft limit the medium enters through an adjoint dipole, whose Gaussian average is e−12∫ds n(s)σ(r)\mathrm{e}^{-\frac{1}{2}\int ds\,n(s)\sigma(\mathbf{r})} with

σ(r)=2g2Nc[γ(0)−γ(r)]=∫d2q(2π)2V(q)(1−eiq⋅r)\sigma(\mathbf{r}) = 2g^2N_c[\gamma(0)-\gamma(\mathbf{r})] = \int\frac{d^2\mathbf{q}}{(2\pi)^2}V(\mathbf{q})\left(1-\mathrm{e}^{i\mathbf{q}\cdot\mathbf{r}}\right)

that is, V(q)=2g2Ncγ~(q)V(\mathbf{q}) = 2g^2N_c\widetilde{\gamma}(\mathbf{q}), with γ~\widetilde{\gamma} the Fourier transform of γ\gamma, is the elastic collision rate of Ref. [5].

Typically, the potential is taken to be of Yukawa type (Gyulassy–Wang model) [6]. Another option is the HTL potential [7], which shares the 1/q41/q^4 tail of the Yukawa potential and, when regularized with a magnetic mass, also admits the representation as a Poisson gas of Gaussian scattering centers used in this paper (Appendix B). In this paper we focus on the Yukawa potential and leave other cases for future studies. This tail is associated with rare but relevant hard kicks. It makes the dipole cross section behave as r2log⁡(1/μ2r2)r^2\log(1/\mu^2r^2) at small rr, and it has proven a hurdle that is difficult to overcome analytically.

Traditionally, two analytical limits have been used: GLV [8, 9, 10] and the harmonic oscillator (HO) [1, 11]. The GLV spectrum is the first order of an expansion in the opacity of the medium. The HO approximation keeps only the quadratic term of the dipole cross section, which amounts to a fixed width for every scattering, but it resums multiple scatterings. More recently, the improved opacity expansion (IOE) [12, 13, 14], which expands around the HO, and the resummed opacity expansion (ROE) [15], which extends the description to the Bethe–Heitler regime, have made significant progress.

From the numerical point of view, Andrés et al. [5] proposed to compute the spectrum by solving a PDE, following earlier work by Caron-Huot and Gale [16], and later extended this approach to expanding media with kinematic cuts [17]. Closer to the approach followed in this paper, Feal and Vázquez [18] used a Monte Carlo method. Both approaches resum multiple scatterings to all orders.

In this paper we also tackle the problem in full generality. We write the Gyulassy–Wang dipole cross section as a mixture of Gaussians, which allows us to interpret the medium as a Poisson gas of Gaussian scattering centers, each characterized by its position and its transverse-momentum width (see § 2). Analytical calculations lead us to a master formula that encompasses both the spectrum differential in k⊥k_\perp and the spectrum with a cut on k⊥k_\perp. This master formula is amenable to a standard Monte Carlo integration over configurations, which is efficient, highly parallelizable and able to handle arbitrary profiles of both μ(s)\mu(s) and n(s)n(s).

Unlike PDE-based approaches, each configuration of the medium is solved in closed form. This provides a clearer picture of the BDMPS-Z spectrum: every configuration is Gaussian and is solved like the harmonic oscillator, with a frequency given by a sum of impulses located at the scattering centers. Feal and Vázquez also use a Monte Carlo method, but they sample momentum kicks in discretized time, whereas we sample Poisson-distributed centers in continuous time. On the other hand, Feal and Vázquez include exact kinematics, large emission angles, a thermal gluon mass and the external legs. We leave the inclusion of this physics in our formalism for future work.

The method reproduces the first order in opacity and the published results without kinematic cuts. With a cut, we find sizeable relative deviations from previous results, both at small gluon energy fraction, where the cut strongly suppresses the spectrum, and at large energy fraction, where they reach 5–20 %. The code is publicly available at https://github.com/manoel-maker/poisson-gas-bdmps_z.

The paper is organized as follows. In Sec. II we set up the calculation: starting from known expressions for the BDMPS-Z spectrum [5, 10] and relying on our representation of the medium, we write the spectrum in a form from which the master formula, on which this paper is based, follows directly. In Sec. III we derive the master formula for the spectrum differential in k⊥k_\perp and with a cut, for arbitrary μ(s)\mu(s) and n(s)n(s), by integrating the transverse positions center by center; the equivalent route through Arnold’s propagator [19] is given in Appendix A. In Sec. IV we obtain the spectrum integrated over all k⊥k_\perp, again by two routes. In Sec. V we present the Monte Carlo method. Although the methodology is standard, we give a self-contained review: we present Mecke’s formula and discuss the importance sampling of the time between the emission and a marked scattering center. Section VI contains the validation and the comparison with published results, and Sec. VII our summary and outlook. Appendix B discusses the HTL potential with a magnetic mass.

Representation of the medium

Starting point

As detailed in the Introduction, we study the medium-induced gluon emission spectrum in the soft-gluon limit. Our starting point is the spectrum differential in kk, Eqs. (2.2)–(2.4) of Ref. [5], written in position space:

dIdω d2k=2αsCR(2π)2Re⁡∫0∞dt2∫0t2dt1∫d2z e−ik⋅ze−12∫t2∞ds n(s)σ(z)×∇y⋅∇zK(t2,z;t1,y)∣y=0−(vacuum),(2)\frac{\mathrm{d}I}{\mathrm{d}\omega\,\mathrm{d}^{2}k} = \frac{2\alpha_s C_R}{(2\pi)^2} \operatorname{Re} \int_{0}^{\infty} \mathrm{d}t_2 \int_{0}^{t_2} \mathrm{d}t_1 \int\mathrm{d}^{2}z\, e^{-i\boldsymbol{k}\cdot\boldsymbol{z}} e^{-\frac{1}{2}\int_{t_2}^{\infty}\mathrm{d}s\,n(s)\sigma(\boldsymbol{z})} \times\left.\boldsymbol{\nabla}_y\cdot\boldsymbol{\nabla}_z \mathcal{K}(t_2,\boldsymbol{z};t_1,\boldsymbol{y})\right|_{\boldsymbol{y}=0} - (\text{vacuum}), \tag*{(2)}

Figure 3.

K(t2,z;t1,y)=∫r(t1)=yr(t2)=zDr exp⁡{∫t1t2ds[iω2r˙ 2−v(r(s),s)]},(3)\mathcal{K}(t_2,\boldsymbol{z};t_1,\boldsymbol{y}) = \int_{\boldsymbol{r}(t_1)=\boldsymbol{y}}^{\boldsymbol{r}(t_2)=\boldsymbol{z}} \mathcal{D}\boldsymbol{r}\,\exp\left\{\int_{t_1}^{t_2}\mathrm{d}s\left[\frac{i\omega}{2}\dot{\boldsymbol{r}}^{\,2}-v(\boldsymbol{r}(s),s)\right]\right\}, \tag*{(3)}

with v=12nσv=\frac{1}{2}n\sigma, and “(vacuum)” denotes the same expression with n≡0n\equiv0. In (2) the medium enters through two factors. The first one is the kernel K\mathcal{K}, which describes the evolution of the dipole between t1t_1 and t2t_2. The second factor is the broadening factor e−12∫t2∞ds n(s)σ(z)e^{-\frac{1}{2}\int_{t_2}^{\infty}\mathrm{d}s\,n(s)\sigma(\boldsymbol{z})}. It has the form of a kernel without kinetic term: after the kinetic phases of the amplitude and of the conjugate amplitude cancel, the dipole size is frozen at z\boldsymbol{z}, and the factor describes the broadening of the emitted gluon. That the problem reduces to a single dipole is a consequence of the soft limit, in which the emitter is purely eikonal.

For the Yukawa potential V(q)=8πμ2/(q2+μ2)2V(q)=8\pi\mu^2/(q^2+\mu^2)^2 [5], with a local screening mass μ(s)\mu(s) [17], the dipole cross section σ(r)=∫d2q(2π)2V(q)(1−eiq⋅r)\sigma(r)=\int\frac{\mathrm{d}^2q}{(2\pi)^2}V(q)\left(1-e^{i\boldsymbol{q}\cdot\boldsymbol{r}}\right) gives

v(r,s)=12n(s)σ(r)=n(s)[1−μ(s)rK1(μ(s)r)].(4)v(r,s)=\frac{1}{2}n(s)\sigma(r)=n(s)\left[1-\mu(s)rK_1\left(\mu(s)r\right)\right]. \tag*{(4)}

The Yukawa dipole as a superposition of Gaussians

Since V(q)V(q) is a rational function of q2+μ2q^2+\mu^2, its inverse Laplace transform [20] is immediate,

1(q2+μ2)2=∫0∞dλ λe−λ(q2+μ2),(5)\frac{1}{(q^2+\mu^2)^2}=\int_{0}^{\infty}\mathrm{d}\lambda\,\lambda e^{-\lambda(q^2+\mu^2)}, \tag*{(5)}

and the integral over q\boldsymbol{q} in σ(r)\sigma(r) becomes Gaussian. This is, in essence, Schwinger’s proper-time parametrization [21], commonly used in quantum field theory. Performing the Gaussian integral and changing variables to τ=1/(4λ)\tau=1/(4\lambda) we obtain, for each ss,

1−μ(s)rK1(μ(s)r)=∫0∞dτ ρs(τ)(1−e−τr2),(6)1-\mu(s)rK_1\left(\mu(s)r\right)=\int_{0}^{\infty}\mathrm{d}\tau\,\rho_s(\tau)\left(1-e^{-\tau r^2}\right), \tag*{(6)}

with

ρs(τ)=μ(s)24τ2e−μ(s)2/(4τ).(7)\rho_s(\tau)=\frac{\mu(s)^2}{4\tau^2}e^{-\mu(s)^2/(4\tau)}. \tag*{(7)}

Each term e−τr2e^{-\tau r^2} is the dipole factor of a Gaussian scattering center, which transfers a transverse momentum distributed as e−q2/(4τ)e^{-q^2/(4\tau)}, with ⟨q2⟩=4τ\langle q^2\rangle=4\tau. We refer to τ\tau as the transverse-momentum width of the center. The screening mass only sets the scale of the widths: with ε=μ2/(4τ)\varepsilon=\mu^2/(4\tau), ρs(τ) dτ=e−εdε\rho_s(\tau)\,\mathrm{d}\tau=e^{-\varepsilon}\mathrm{d}\varepsilon.

Let us analyze the result. ρs\rho_s is a probability density, ∫0∞ρs dτ=1\int_0^\infty\rho_s\,\mathrm{d}\tau=1, but it has a heavy tail, ρs≃μ2/(4τ2)\rho_s\simeq\mu^2/(4\tau^2) for τ→∞\tau\to\infty, so that the moments ∫ρs τm dτ\int\rho_s\,\tau^m\,\mathrm{d}\tau diverge for m≥1m\ge1. This tail is the Coulomb tail 1/q41/q^4 of V(q)V(q), that is, the hard kicks. It will be relevant in the Monte Carlo of § 5: it is the reason why we integrate exactly over the width of a center whenever possible (§ 5 C), and why we introduce importance sampling at short times (§ 5 E).

The medium as a Poisson process

Beyond the apparent simplification of writing the potential as an average of Gaussians, the representation (6) allows us to interpret the interaction with the medium as a Poisson process. We first discuss an identity valid for Poisson processes in general, and then apply it to our problem.

Consider points (sj,τj)(s_j,\tau_j) in the plane, generated as follows: we divide the plane into cells ds dτ\mathrm{d}s\,\mathrm{d}\tau and, independently for each cell, place a point in it with probability n(s)ρs(τ) ds dτn(s)\rho_s(\tau)\,\mathrm{d}s\,\mathrm{d}\tau. This is a Poisson process in the plane with intensity n(s)ρs(τ)n(s)\rho_s(\tau). Equivalently, the positions sjs_j are distributed along [0,L][0,L] with density n(s)n(s), and each center draws its width τj\tau_j from ρsj\rho_{s_j}, independently of everything else (marking theorem [22 Theorem 5.6]). The number of centers is Poisson distributed with mean χ=∫0Ln ds\chi=\int_0^L n\,\mathrm{d}s, the opacity. We denote by E\mathbb{E} the expectation over this process.

Generating functional. For a bounded function f(s,τ)f(s,\tau),

E[∏jf(sj,τj)]=e−∫0Lds n(s)∫0∞dτ ρs(τ) [1−f(s,τ)].(8)\mathbb{E}\left[\prod_j f(s_j,\tau_j)\right] = e^{-\int_0^L \mathrm{d}s\,n(s)\int_0^\infty\mathrm{d}\tau\,\rho_s(\tau)\,[1-f(s,\tau)]}. \tag*{(8)}

Proof. The cells are independent, so the expectation of the product is the product of the expectations. In a cell the factor equals f(s,τ)f(s,\tau) with probability nρs ds dτn\rho_s\,\mathrm{d}s\,\mathrm{d}\tau and 1 otherwise; its expectation is 1−nρs ds dτ (1−f)1-n\rho_s\,\mathrm{d}s\,\mathrm{d}\tau\,(1-f). The product over cells tends to the exponential in Eq. (8). □ This is the probability generating functional of a Poisson process [22 Eq. (3.5) and Theorem 3.9]; the proof above is complete and does not require homogeneity.

In our case, in each slice ds\mathrm{d}s there is a scattering center with probability n(s) dsn(s)\,\mathrm{d}s, and each center carries a Gaussian potential whose width is drawn from ρs\rho_s; the Yukawa potential emerges from the average over the widths. Indeed, with f(s,τ)=e−τr(s)2f(s,\tau)=e^{-\tau r(s)^2}, Eqs. (8) and (6) give, for any path r(s)r(s) and any time interval JJ,

e−∫Jds v(r(s),s)=∑N=0∞e−χJχJNN!∏i=1N∫Jdsi n(si)χJ∫0∞dτi ρsi(τi)e−τir(si)2=E[∏j:sj∈Je−τjr(sj)2]=E[e−∫Jds∑jτjδ(s−sj)r(s)2].(9)\begin{aligned} e^{-\int_J \mathrm{d}s\,v(r(s),s)} = \sum_{N=0}^{\infty}e^{-\chi_J}\frac{\chi_J^N}{N!}\prod_{i=1}^{N}\int_J\frac{\mathrm{d}s_i\,n(s_i)}{\chi_J}\int_0^\infty\mathrm{d}\tau_i\,\rho_{s_i}(\tau_i)e^{-\tau_i r(s_i)^2} \\ &= \mathbb{E}\left[\prod_{j:s_j\in J}e^{-\tau_j r(s_j)^2}\right] = \mathbb{E}\left[e^{-\int_J\mathrm{d}s\sum_j\tau_j\delta(s-s_j)r(s)^2}\right]. \tag*{(9)} \end{aligned}

with χJ=∫Jn ds\chi_J=\int_J n\,\mathrm{d}s. For a fixed configuration the exponent is quadratic in r\mathbf{r}: a harmonic oscillator whose frequency acts only at the instants sjs_j.

Note that the number NN of centers in a configuration is not the order in the opacity expansion. Each power of v=n(1−∫ρse−τr2)v=n\left(1-\int\rho_s e^{-\tau r^2}\right) in that expansion mixes a real and a virtual term, whereas in Eq. (9) the virtual terms are resummed in the Poisson weight e−χe^{-\chi} of each configuration. This weight is the elastic Sudakov factor that is resummed in the resummed opacity expansion [15], which then expands in the number of real scatterings. Here, instead, each configuration, with any number of centers, is solved in closed form, and the average over configurations sums all of them.

Kernel and broadening as a single functional

Consider the complete path on [t1,∞)[t_1,\infty), equal to r(s)r(s) between t1t_1 and t2t_2 and equal to z\mathbf{z} (frozen) after t2t_2. The product of kernel and broadening factor in Eq. (2) then contains

e−∫t1t2v(r(s)) dse−∫t2∞v(z) ds=e−∫t1∞v(r(s)) ds,(10)e^{-\int_{t_1}^{t_2}v(r(s))\,\mathrm{d}s}e^{-\int_{t_2}^{\infty}v(\mathbf{z})\,\mathrm{d}s} =e^{-\int_{t_1}^{\infty}v(r(s))\,\mathrm{d}s}, \tag*{(10)}

that is, the functional (9) evaluated on a single path. For a single configuration: the centers with t1<sj<t2t_1<s_j<t_2 enter the kernel, with the varying path; the centers with sj>t2s_j>t_2 enter the broadening factor with r=z\mathbf{r}=\mathbf{z} fixed, ∏sj>t2e−τjz2=e−T(t2)z2\prod_{s_j>t_2}e^{-\tau_j\mathbf{z}^2}=e^{-T(t_2)\mathbf{z}^2}, with T(t2)=∑sj>t2τjT(t_2)=\sum_{s_j>t_2}\tau_j; and the centers with sj<t1s_j<t_1 do not contribute. Exchanging E\mathbb{E} with the path integral,

e−12∫t2∞nσ(z) ∇y⋅∇zK=E[e−T(t2)z2 ∇y⋅∇zKc(t2,z;t1,y)],(11)e^{-\frac{1}{2}\int_{t_2}^{\infty}n\sigma(z)\,\nabla_{\mathbf{y}}\cdot\nabla_{\mathbf{z}}\mathcal{K}} = \mathbb{E}\left[e^{-T(t_2)\mathbf{z}^2}\,\nabla_{\mathbf{y}}\cdot\nabla_{\mathbf{z}}\mathcal{K}_c(t_2,\mathbf{z};t_1,\mathbf{y})\right], \tag*{(11)}
Kc=∫Drexp⁡{∫t1t2ds[iω2r˙2−∑jτjδ(s−sj)r2]}.\mathcal{K}_c=\int\mathcal{D}\mathbf{r}\exp\left\{\int_{t_1}^{t_2}\mathrm{d}s\left[\frac{i\omega}{2}\dot{\mathbf{r}}^2-\sum_j\tau_j\delta(s-s_j)\mathbf{r}^2\right]\right\}.

Configuration by configuration, everything is Gaussian: an oscillator with an impulsive frequency.

Since centers in disjoint intervals are independent, for fixed t2t_2 the expectation would also factorize, E[Kce−Tz2]=E[Kc]E[e−Tz2]\mathbb{E}[\mathcal{K}_c e^{-T\mathbf{z}^2}]=\mathbb{E}[\mathcal{K}_c]\mathbb{E}[e^{-T\mathbf{z}^2}]. The advantage of using a single configuration is a different one: configuration by configuration everything is Gaussian in z\mathbf{z}, and when t2t_2 moves a center passes continuously from one factor to the other (Sec. III C), which allows the integral over t2t_2 to be done in closed form.

Master formula: spectrum differential in k and with a cut

By Eqs. (2) and (11), for a fixed configuration we need

Dc(t1,t2;k)≡∫d2z e−ik⋅ze−T(t2)z2∇y⋅∇zKc(t2,z;t1,y)∣y=0.(12)\mathcal{D}_c(t_1,t_2;\mathbf{k})\equiv\int\mathrm{d}^2z\,e^{-i\mathbf{k}\cdot\mathbf{z}}e^{-T(t_2)\mathbf{z}^2} \left.\nabla_{\mathbf{y}}\cdot\nabla_{\mathbf{z}}\mathcal{K}_c(t_2,\mathbf{z};t_1,\mathbf{y})\right|_{\mathbf{y}=0}. \tag*{(12)}

Everything in this section refers to a fixed configuration, with arbitrary n(s)n(s) and μ(s)\mu(s), which enter only through the sjs_j and the τj\tau_j; the expectation E\mathbb{E} is taken at the end.

Center-by-center transverse integrals

For a fixed configuration, the integrand of Eq. (12) is a product of three types of factors: free Fresnel kernels between consecutive instants, ω2πiΔexp⁡{iω(r′−r)22Δ}\frac{\omega}{2\pi i\Delta}\exp\left\{\frac{i\omega(\mathbf{r}'-\mathbf{r})^2}{2\Delta}\right\}; at each center t1<sj<t2t_1<s_j<t_2, the factor e−τjrj2e^{-\tau_j r_j^2}, with rj\mathbf{r}_j the transverse position at sjs_j; and, at t2t_2, the broadening factor e−T(t2)z2e^{-T(t_2)z^2} together with the phase e−ik⋅ze^{-i\mathbf{k}\cdot\mathbf{z}}. We integrate over the positions rj\mathbf{r}_j one at a time, in time order.

Accumulated kernel. After integrating over all positions before an instant tt, the result as a function of the position r\mathbf{r} at tt is

N(t)exp⁡{iω2[p(t)r2−2q(t)r⋅y]+O(y2)},(13)\mathcal{N}(t)\exp\left\{\frac{i\omega}{2}\left[p(t)r^2-2q(t)\mathbf{r}\cdot\mathbf{y}\right]+O(y^2)\right\}, \tag*{(13)}

where the term in y2y^2 is not needed, since only ∇y\nabla_{\mathbf{y}} at y=0\mathbf{y}=0 enters. The family (13) is closed under the two operations that appear, with the rules of Table 1. A free segment is the two-dimensional Gaussian integral

stepppqqN\mathcal{N}
first segment, length Δ\Delta from (t1,y)(t_1,\mathbf{y})1/Δ1/\Delta1/Δ1/\Deltaω/(2πiΔ)\omega/(2\pi i\Delta)
free segment, length Δ\Deltap1+pΔ\frac{p}{1+p\Delta}q1+pΔ\frac{q}{1+p\Delta}N1+pΔ\frac{\mathcal{N}}{1+p\Delta}
center jjp+ajp+a_jqqN\mathcal{N}

Table 1. TABLE I. Rules for the accumulated kernel (13).

∫d2r ω2πiΔeiω2Δ(r′−r)2eiω2[pr2−2qr⋅y];\int d^2r\,\frac{\omega}{2\pi i\Delta}e^{\frac{i\omega}{2\Delta}(\mathbf{r}'-\mathbf{r})^2}e^{\frac{i\omega}{2}[pr^2-2q\mathbf{r}\cdot\mathbf{y}]};

with ζ≡1/p\zeta\equiv1/p, its rule for pp is simply ζ→ζ+Δ\zeta\to\zeta+\Delta. A center multiplies by e−τjr2=exp⁡{iω2ajr2}e^{-\tau_jr^2}=\exp\left\{\frac{i\omega}{2}a_jr^2\right\}, with

aj=2iτjω.(14)a_j=\frac{2i\tau_j}{\omega}. \tag*{(14)}

The Fresnel integrals with complex parameters have a precise distributional meaning. Here, moreover, they are absolutely convergent in almost all cases: each intermediate position rj\mathbf{r}_j sits at a center and carries the real damping e−τjrj2e^{-\tau_jr_j^2}, and the final position z\mathbf{z} carries e−Tz2e^{-Tz^2} with T>0T>0 as soon as there is a center after t2t_2. Only for T=0T=0 is the integral over z\mathbf{z} a pure Fresnel integral, to be understood in the distributional sense.

The function SS. We define

S≡1q,S˙≡pq,so thatN=ω2πiS.(15)S\equiv\frac{1}{q},\qquad\dot{S}\equiv\frac{p}{q},\qquad\text{so that}\qquad\mathcal{N}=\frac{\omega}{2\pi iS}. \tag*{(15)}

Between centers ζ=1/p=S/S˙\zeta=1/p=S/\dot{S} grows as ζ+Δ\zeta+\Delta, so that SS is linear in tt with constant slope S˙\dot{S}. At a center, p→p+ajp\to p+a_j with qq fixed means

S˙j+=S˙j−+ajSj,Sj≡S(sj;t1),(16)\dot{S}_j^{+}=\dot{S}_j^{-}+a_jS_j,\qquad S_j\equiv S(s_j;t_1), \tag*{(16)}

with SS continuous. Without medium, S=t2−t1S=t_2-t_1; centers with sj<t1s_j<t_1 do not enter. Expanding the recursion, SS is a multilinear polynomial in the aja_j,

S(t2;t1)=∑U∏i∈Uai⋅(su1−t1)(su2−su1)⋯(t2−suk),(17)S(t_2;t_1)=\sum_U\prod_{i\in U}a_i\cdot(s_{u_1}-t_1)(s_{u_2}-s_{u_1})\cdots(t_2-s_{u_k}), \tag*{(17)}

where the sum runs over the subsets U={u1<⋯<uk}U=\{u_1<\cdots<u_k\} of the centers in (t1,t2)(t_1,t_2), the term U=∅U=\varnothing being t2−t1t_2-t_1. This is the Gelfand–Yaglom structure [23] that reappears in Sec. IV B.

Integrals over z\mathbf{z} and over ∣k∣<K|\mathbf{k}|<K

Broadening as one more kick. Multiplying the kernel at t2t_2 by e−T(t2)z2e^{-T(t_2)z^2} amounts, in Table 1, to p→p+2iT/ωp\to p+2iT/\omega: the accumulated effect of all centers after t2t_2, applied at once, with no free segments between them. The total coefficient of −z2-z^2 in the exponent is then

ϑ=T−η=iω2(p+2iTω),η≡iωp2=iωS˙2S.(18)\vartheta=T-\eta=\frac{i\omega}{2}\left(p+\frac{2iT}{\omega}\right),\qquad\eta\equiv\frac{i\omega p}{2}=\frac{i\omega\dot{S}}{2S}. \tag*{(18)}

When t2t_2 crosses a center, its kick passes from “after t2t_2” to “before t2t_2” without changing the sum of kicks applied at that instant: ϑ\vartheta is continuous by construction. This is the transparent form of the statement that kernel and broadening are a single functional (see Sec. III D).

Positivity of Re⁡ϑ\operatorname{Re}\vartheta. From Eq. (18), Re⁡ϑ=T+ω2Im⁡S˙/S\operatorname{Re}\vartheta=T+\frac{\omega}{2}\operatorname{Im}\dot{S}/S. On the first segment p=1/Δp=1/\Delta is real. At each center Im⁡S˙\operatorname{Im}\dot{S} increases by 2τj/ω>02\tau_j/\omega>0. On a free segment ζ→ζ+Δ\zeta\to\zeta+\Delta with Δ\Delta real preserves the sign of Im⁡ζ\operatorname{Im}\zeta, and Im⁡S˙=−Im⁡ζ/∣S∣2\operatorname{Im}\dot{S}=-\operatorname{Im}\zeta/|S|^2, so Im⁡p≥0\operatorname{Im}p\ge0 is preserved. Hence Re⁡ϑ≥T≥0\operatorname{Re}\vartheta\ge T\ge0, with strict inequality as soon as there is any center after t1t_1.

Derivative at y=0\mathbf{y}=0. From (13) and (15),

∇yKc∣y=0=−iωqzNeηz2=−ω22πS2zeηz2.\nabla_{\mathbf{y}}\mathcal{K}_c\big|_{\mathbf{y}=0}=-i\omega qz\mathcal{N}e^{\eta z^2}=-\frac{\omega^2}{2\pi S^2}\mathbf{z}e^{\eta z^2}.

Applying ∇z\nabla_{\mathbf{z}}, with ∇z⋅z=2\nabla_{\mathbf{z}}\cdot\mathbf{z}=2 in two dimensions,

∇y⋅∇zKc∣y=0=−ω2πS2(1+ηz2)eηz2.(19)\nabla_{\mathbf{y}}\cdot\nabla_{\mathbf{z}}\mathcal{K}_c\big|_{\mathbf{y}=0}=-\frac{\omega^2}{\pi S^2}(1+\eta z^2)e^{\eta z^2}. \tag*{(19)}

By Eq. (19), the integrand of Eq. (12) is e−ik⋅z(1+ηz2)e−ϑz2e^{-i\mathbf{k}\cdot\mathbf{z}}(1+\eta z^2)e^{-\vartheta z^2}. Since Re⁡ϑ>0\operatorname{Re}\vartheta>0, these are two Gaussian integrals in two dimensions,

∫d2z e−ik⋅z−ϑz2=πϑe−k2/4ϑ,\int d^2z\,e^{-i\mathbf{k}\cdot\mathbf{z}-\vartheta z^2}=\frac{\pi}{\vartheta}e^{-k^2/4\vartheta},
∫d2z z2e−ik⋅z−ϑz2=πϑ2(1−k24ϑ)e−k2/4ϑ.\int d^2z\,z^2e^{-i\mathbf{k}\cdot\mathbf{z}-\vartheta z^2}=\frac{\pi}{\vartheta^2}\left(1-\frac{k^2}{4\vartheta}\right)e^{-k^2/4\vartheta}.

Thus, per configuration and for fixed t1,t2t_1,t_2,

Dc(t1,t2;k)=−ω2S2ϑe−k2/4ϑ[1+ηϑ(1−k24ϑ)].(20)\mathcal{D}_c(t_1,t_2;\mathbf{k})=-\frac{\omega^2}{S^2\vartheta}e^{-k^2/4\vartheta}\left[1+\frac{\eta}{\vartheta}\left(1-\frac{k^2}{4\vartheta}\right)\right]. \tag*{(20)}

For the integral over ∣k∣<K|\mathbf{k}|<K we use d2k=πd(k2)d^2k=\pi d(k^2) and Q=e−K2/4ϑQ=e^{-K^2/4\vartheta}. The terms in η/ϑ\eta/\vartheta without KK cancel, and

∫∣k∣<Kd2k πϑe−k2/4ϑ[1+ηϑ(1−k24ϑ)]=4π2[1−Q(1−ηK24ϑ2)].(21)\int_{|\mathbf{k}|<K}d^2k\,\frac{\pi}{\vartheta}e^{-k^2/4\vartheta}\left[1+\frac{\eta}{\vartheta}\left(1-\frac{k^2}{4\vartheta}\right)\right] =4\pi^2\left[1-Q\left(1-\frac{\eta K^2}{4\vartheta^2}\right)\right]. \tag*{(21)}

With the prefactor of (2) and IK≡π2αsCRωdIdω∣k<K\mathcal{I}_K\equiv\frac{\pi}{2\alpha_s C_R}\omega\frac{dI}{d\omega}\big|_{k<K},

IK=Re⁡E[∫0Ldt1∫t1∞dt2(−h(t2))]−(vacuum),(22)\mathcal{I}_K=\operatorname{Re}\mathbb{E}\left[\int_0^Ldt_1\int_{t_1}^{\infty}dt_2\left(-h(t_2)\right)\right]-(\text{vacuum}), \tag*{(22)}
h(t2)≡1S2[1−Q+ηQK24ϑ2].h(t_2)\equiv\frac{1}{S^2}\left[1-Q+\frac{\eta QK^2}{4\vartheta^2}\right].

Integral over t2t_2

For fixed t1t_1 and configuration, SS, S˙\dot{S}, TT, η\eta, ϑ\vartheta and QQ are functions of t2t_2.

(i) Primitive between centers. Between two centers S˙\dot{S} and TT are constant. Let

P(t2)≡Q−1SS˙.(23)P(t_2)\equiv\frac{Q-1}{S\dot{S}}. \tag*{(23)}

Then ∂t2(1/SS˙)=−1/S2\partial_{t_2}(1/S\dot{S})=-1/S^2; η˙=−ηS˙/S\dot{\eta}=-\eta\dot{S}/S, hence ϑ˙=−η˙=ηS˙/S\dot{\vartheta}=-\dot{\eta}=\eta\dot{S}/S; and Q˙=QK24ϑ2ϑ˙\dot{Q}=Q\frac{K^2}{4\vartheta^2}\dot{\vartheta}. Adding up,

∂t2P=1−QS2+Q˙SS˙=h(t2).(24)\partial_{t_2}P=\frac{1-Q}{S^2}+\frac{\dot{Q}}{S\dot{S}}=h(t_2). \tag*{(24)}

(ii) Continuity at the centers. SS and ϑ\vartheta are continuous (Sec. III A), hence so is QQ; only S˙\dot{S} jumps. At t2=sjt_2=s_j,

ϑj=Tj+−iωS˙j+2Sj=Tj−−iωS˙j−2Sj,(25)\vartheta_j=T_j^+-\frac{i\omega\dot{S}_j^+}{2S_j}=T_j^--\frac{i\omega\dot{S}_j^-}{2S_j}, \tag*{(25)}

with Tj+=∑i>jτiT_j^+=\sum_{i>j}\tau_i and Tj−=τj+Tj+T_j^-=\tau_j+T_j^+: the jump of TT exactly compensates that of S˙\dot{S}. The value of TT at t2=sjt_2=s_j itself is ambiguous, but it is a set of measure zero, the piecewise sum only uses the one-sided limits, and the combination that enters is continuous.

(iii) Piecewise sum. ∫t1∞h dt2=∑segments[P(end)−P(start)]\int_{t_1}^{\infty}h\,dt_2=\sum_{\mathrm{segments}}[P(\mathrm{end})-P(\mathrm{start})], with three types of contribution. (a) Endpoint t2→t1+t_2\to t_1^+. S→0S\to0, S˙=1\dot{S}=1, η≃iω/(2S)→∞\eta\simeq i\omega/(2S)\to\infty, ϑ≃−η\vartheta\simeq-\eta, K2/4ϑ→0K^2/4\vartheta\to0 and P→K2/(2iω)P\to K^2/(2i\omega): a purely imaginary constant, independent of the configuration. It is exactly the vacuum contribution, which cancels in the subtraction. (b) Endpoint t2→∞t_2\to\infty. SS grows linearly and ∣Q∣≤1|Q|\le1, so P→0P\to0. (c) Centers sj>t1s_j>t_1. With SS and QQ continuous and the jump (16),

P(sj−)−P(sj+)=Qj−1Sj(1S˙j−−1S˙j+)=(Qj−1)ajS˙j−S˙j+.P(s_j^-)-P(s_j^+)=\frac{Q_j-1}{S_j}\left(\frac{1}{\dot{S}_j^-}-\frac{1}{\dot{S}_j^+}\right)=\frac{(Q_j-1)a_j}{\dot{S}_j^-\dot{S}_j^+}.

with Qj=e−K2/4ϑjQ_j=e^{-K^2/4\vartheta_j}. For t1>Lt_1>L there are no later centers and the contribution vanishes. We arrive at the master formula, valid for arbitrary n(s)n(s) and μ(s)\mu(s):

IK=Re⁡E[∫0Ldt1∑j:sj>t1aj(1−e−K2/4ϑj)S˙j−S˙j+].(26)\mathcal{I}_K=\operatorname{Re}\mathbb{E}\left[\int_0^Ldt_1\sum_{j:s_j>t_1}\frac{a_j(1-e^{-K^2/4\vartheta_j})}{\dot{S}_j^-\dot{S}_j^+}\right]. \tag*{(26)}

with ϑj\vartheta_j given by (25), Sj=S(sj;t1)S_j=S(s_j;t_1), S˙j∓\dot{S}_j^\mp the slopes just before and after center jj, aj=2iτj/ωa_j=2i\tau_j/\omega, and E\mathbb{E} the expectation over the Poisson process of § 2 C. Differentiating with respect to K2K^2,

ωdIdω dk2=2αsCRπ×Re⁡E[∫0Ldt1∑j:sj>t1aj4ϑjS˙j−S˙j+e−k2/4ϑj],(27)\omega\frac{dI}{d\omega\,dk^2}=\frac{2\alpha_s C_R}{\pi}\times\operatorname{Re}\mathbb{E}\left[\int_0^Ldt_1\sum_{j:s_j>t_1}\frac{a_j}{4\vartheta_j\dot{S}_j^-\dot{S}_j^+}e^{-k^2/4\vartheta_j}\right], \tag*{(27)}

a mixture of complex Gaussians in kk.

Reading. The gluon is emitted at t1t_1; the centers between t1t_1 and sjs_j build SS through the jumps (16); then comes a “special” center jj; the centers after jj only count through Tj+T_j^+ inside ϑj\vartheta_j, that is, through the broadening of the already emitted gluon. The profile of the medium only enters through where the centers fall (nn) and through their widths (μ\mu).

Spectrum integrated over all k

Integrating (2) over all kk gives ∫d2k e−ik⋅z=(2π)2δ(2)(z)\int d^2k\,e^{-i\mathbf{k}\cdot\mathbf{z}}=(2\pi)^2\delta^{(2)}(\mathbf{z}), and since σ(0)=0\sigma(0)=0 the broadening factor disappears. With I≡π2αsCRωdIdω\mathcal{I}\equiv\frac{\pi}{2\alpha_s C_R}\omega\frac{dI}{d\omega} and (19) at z=0z=0,

I=Re⁡E[−∫0∞dt1∫t1∞dt2(1S(t2;t1)2−1(t2−t1)2)],(28)\mathcal{I}=\operatorname{Re}\mathbb{E}\left[-\int_0^\infty dt_1\int_{t_1}^{\infty}dt_2\left(\frac{1}{S(t_2;t_1)^2}-\frac{1}{(t_2-t_1)^2}\right)\right], \tag*{(28)}

where the vacuum has S0=t2−t1S_0=t_2-t_1. This is the structure of Eqs. (3.7)–(3.8) of Arnold [19]. It is also the K→∞K\to\infty limit of the master formula: with Qj→0Q_j\to0, Eq. (26) gives

I=Re⁡E[∫0Ldt1∑j:sj>t1ajS˙j−S˙j+].(29)\mathcal{I}=\operatorname{Re}\mathbb{E}\left[\int_0^Ldt_1\sum_{j:s_j>t_1}\frac{a_j}{\dot{S}_j^-\dot{S}_j^+}\right]. \tag*{(29)}

The next two subsections bring this spectrum to the form I=Eln⁡∣Φ∣\mathcal{I}=\mathbb{E}\ln|\Phi|.

Arnold’s route: the Wronskian identity

Integral over t2t_2 [19 Eqs. (3.9)–(3.15)]. SS and C=−∂t1SC=-\partial_{t_1}S solve the same equation in t2t_2, with constant Wronskian CS˙−S∂t2C=1C\dot{S}-S\partial_{t_2}C=1, evaluated at t2=t1t_2=t_1, where S=0S=0 and S˙=C=1\dot{S}=C=1. Dividing by S2S^2, ∂t2(C/S)=−1/S2\partial_{t_2}(C/S)=-1/S^2, and for the vacuum C0/S0=1/(t2−t1)C_0/S_0 = 1/(t_2-t_1). At t2→t1+t_2 \to t_1^+, before the first center, S=t2−t1S=t_2-t_1 and C=1C=1 exactly, so the lower limit does not contribute. For t2≥Lt_2 \ge L, SS is a straight line, S=S(L;t1)+φ(t1)(t2−L)S=S(L;t_1)+\varphi(t_1)(t_2-L) with φ(t1)≡S˙(t2;t1)∣t2≥L\varphi(t_1)\equiv\dot{S}(t_2;t_1)|_{t_2\ge L}, and C/S→−φ′(t1)/φ(t1)C/S\to-\varphi'(t_1)/\varphi(t_1). Hence

∫t1∞dt2(1S2−1(t2−t1)2)=φ′(t1)φ(t1).(30)\int_{t_1}^{\infty}dt_2\left(\frac{1}{S^2}-\frac{1}{(t_2-t_1)^2}\right)=\frac{\varphi'(t_1)}{\varphi(t_1)}. \tag*{(30)}

Integral over t1t_1 [19 Eqs. (3.19)–(3.20)]. For t1≥Lt_1\ge L there are no centers after t1t_1 and φ=1\varphi=1. Then −∫0∞dt1 φ′/φ=ln⁡φ(0)-\int_0^\infty dt_1\,\varphi'/\varphi=\ln\varphi(0), and with Φ≡φ(0)=f′(L)\Phi\equiv\varphi(0)=f'(L), f(s)≡S(s;0)f(s)\equiv S(s;0),

I=Re⁡E[ln⁡Φ]=E[ln⁡∣Φ∣],Φ=f′(L).(31)\mathcal{I}=\operatorname{Re}\mathbb{E}\left[\ln\Phi\right]=\mathbb{E}\left[\ln|\Phi|\right],\qquad\Phi=f'(L). \tag*{(31)}

The real part removes the branch ambiguity of the logarithm.

Link with the master formula. Doing the same t2t_2 integral with the primitive of Sec. III C at K→∞K\to\infty, where P→−1/(SS˙)P\to-1/(S\dot{S}), the left-hand side of Eq. (30) equals −∑sj>t1aj/(S˙j−S˙j+)-\sum_{s_j>t_1}a_j/(\dot{S}_j^-\dot{S}_j^+). Therefore

∑j:sj>t1ajS˙j−S˙j+=−φ′(t1)φ(t1).(32)\sum_{j:s_j>t_1}\frac{a_j}{\dot{S}_j^-\dot{S}_j^+}=-\frac{\varphi'(t_1)}{\varphi(t_1)}. \tag*{(32)}

and Eqs. (29) and (31) coincide. Arnold’s paper [19] treats precisely an arbitrary q^(t)\hat{q}(t), i.e., expanding media, in the HO approximation; here the same identity is applied configuration by configuration.

Gelfand–Yaglom route: Φ\Phi as a multilinear determinant

Lemma. Φ=f′(L)=det⁡(1+AG)\Phi=f'(L)=\det(1+AG), with A=diag⁡(a1,…,aN)A=\operatorname{diag}(a_1,\ldots,a_N) and Gij=min⁡(si,sj)G_{ij}=\min(s_i,s_j).

Proof. Multilinearity. By the rules of Table 1, or with the transfer matrices (free segment (f,f′)↦(f+Δf′,f′)(f,f')\mapsto(f+\Delta f',f'); center jj, (f,f′)↦(f,f′+ajf)(f,f')\mapsto(f,f'+a_jf)), each center acts linearly in its aja_j, so that f′(L)f'(L) is multilinear, f′(L)=∑UcU∏i∈Uaif'(L)=\sum_U c_U\prod_{i\in U}a_i, with the sum over subsets UU of centers. Coefficients. Let U={u1<⋯<uk}U=\{u_1<\cdots<u_k\}. The coefficient cUc_U can be computed with only the centers in UU. After the last one, f′(L)=f′(uk−)+akf(uk)f'(L)=f'(u_k^-)+a_kf(u_k), and f(uk)=f(uk−1)+(uk−uk−1)f′(uk−1−)f(u_k)=f(u_{k-1})+(u_k-u_{k-1})f'(u_{k-1}^-). Since f(uk−1)f(u_{k-1}) does not depend on ak−1a_{k-1}, the coefficient of a1⋯ak−1a_1\cdots a_{k-1} in f(uk)f(u_k) is (uk−uk−1)(u_k-u_{k-1}) times that of f′(uk−1−)f'(u_{k-1}^-), which is the problem with k−1k-1 centers. With c{u1}=u1c_{\{u_1\}}=u_1, cU=u1(u2−u1)⋯(uk−uk−1)c_U=u_1(u_2-u_1)\cdots(u_k-u_{k-1}). Determinant. Since AA is diagonal, det⁡(1+AG)=∑U∏i∈Uaidet⁡GU\det(1+AG)=\sum_U\prod_{i\in U}a_i\det G_U, with GUG_U the principal minors. Subtracting from each row of GUG_U the previous one gives a triangular matrix with diagonal u1,u2−u1,…u_1,u_2-u_1,\ldots, so det⁡GU=cU\det G_U=c_U. □

This is the analogue, for the final slope, of the multilinear polynomial (17). In practice Φ\Phi is computed with the Möbius recursion: with ζ=f/f′\zeta=f/f' (the 1/p1/p of Sec. III A with t1=0t_1=0), ζ→ζ+Δ\zeta\to\zeta+\Delta between centers and f′→f′(1+ajζ)f'\to f'(1+a_j\zeta) at each center, so that ζ→ζ/(1+ajζ)\zeta\to\zeta/(1+a_j\zeta)

Φ=∏j=1N(1+ajζj−),ζ(0)=0,(33)\Phi=\prod_{j=1}^{N}(1+a_j\zeta_j^-),\qquad\zeta(0)=0, \tag*{(33)}

where ζj−\zeta_j^- is the value just before center jj.

One center: closed-form average over τ\tau

Each factor of Eq. (33) only depends on the width of its own center and on the earlier centers. For the center at ss, aζ=2iτζ/ω=β/ϵa\zeta=2i\tau\zeta/\omega=\beta/\epsilon, with

β=iμ(s)2ζ2ω(34)\beta=\frac{i\mu(s)^2\zeta}{2\omega} \tag*{(34)}

and ϵ∼Exp⁡(1)\epsilon\sim\operatorname{Exp}(1) (Sec. II B). The average over the width is then elementary,

∫0∞dϵ e−ϵln⁡(1+βϵ)=V(β)≡ln⁡β+γE+eβE1(β),(35)\int_0^\infty d\epsilon\,e^{-\epsilon}\ln\left(1+\frac{\beta}{\epsilon}\right)=V(\beta)\equiv\ln\beta+\gamma_E+e^\beta E_1(\beta), \tag*{(35)}

since ∫0∞e−ϵln⁡(ϵ+β) dϵ=ln⁡β+eβE1(β)\int_0^\infty e^{-\epsilon}\ln(\epsilon+\beta)\,d\epsilon=\ln\beta+e^\beta E_1(\beta) and ∫0∞e−ϵln⁡ϵ dϵ=−γE\int_0^\infty e^{-\epsilon}\ln\epsilon\,d\epsilon=-\gamma_E. This is the piece used by the Monte Carlo of the integrated spectrum (Sec. V C). With a single center (ζ=s)(\zeta=s) it gives the first order in opacity, ∫0Lds n(s)Re⁡V(iμ(s)2s/2ω)\int_0^L ds\,n(s)\operatorname{Re}V(i\mu(s)^2s/2\omega), which for a static medium is the GLV result, Eq. (B.4) of Ref. [5] in the limit Rˉ→∞\bar{R}\to\infty (no kinematic cut; see Sec. V B).

Monte Carlo

Equations (26), (27) and (31) are averages over configurations of the Poisson process. The Monte Carlo method is the standard one [24]: independent samples X1,…,XnX_1,\ldots,X_n of an estimator Xˉ\bar{X} whose expectation is the quantity sought, and the result Xˉ±sX/n\bar{X}\pm s_X/\sqrt{n}. What is not trivial is the choice of XX. We use four variance-reduction techniques, all of general use: (i) a marked center (Mecke’s formula), which replaces a sum over centers by an integral over the position of one “special” center; (ii) exact integration of one variable, the width of the marked center, whenever possible; (iii) importance sampling of the time between the emission and the marked center; and (iv) a control variate, the estimator without cut, whose mean is the integrated spectrum. Everything is written for arbitrary n(s)n(s) and μ(s)\mu(s); the static medium is a particular case.

Mecke’s formula

For a function h(x;C)h(x;\mathcal{C}) of a point x=(s,τ)x=(s,\tau) and of the configuration C\mathcal{C} of the Poisson process of Sec. II C,

E[∑xj∈Ch(xj;C)]=∫0Lds n(s)∫0∞dτ ρs(τ) E[h((s,τ);C∪{(s,τ)})].(36)\mathbb{E}\left[\sum_{x_j\in\mathcal{C}}h(x_j;\mathcal{C})\right]=\int_0^L\mathrm{d}s\,n(s)\int_0^\infty\mathrm{d}\tau\,\rho_s(\tau)\,\mathbb{E}\left[h((s,\tau);\mathcal{C}\cup\{(s,\tau)\})\right]. \tag*{(36)}

This is Mecke’s formula [25], [22 Theorem 4.1]. With the cells of Sec. II C it follows in a few lines: the cell at (s,τ)(s,\tau) contributes to the sum only if it contains a center, which happens with probability nρsds dτn\rho_s\mathrm{d}s\,\mathrm{d}\tau; by the independence between cells, the rest of the configuration has the same law as if that cell did not exist, and an infinitesimal cell does not change that law. □

In practice one samples the position ss of the marked center with density nn, and only those centers that influence its term. Each sample contributes a single term instead of a sum, and centers that do not influence it need not be sampled.

Dimensionless variables and sampling in the opacity coordinate

We measure times in units of LL and transverse momenta in units of a reference screening mass μref\mu_{\rm ref}:

s^=sL,χ(s^)=n(s^L)L,m2(s^)=μ(s^L)2μref2,\hat{s}=\frac{s}{L},\qquad\chi(\hat{s})=n(\hat{s}L)L,\qquad m_2(\hat{s})=\frac{\mu(\hat{s}L)^2}{\mu_{\rm ref}^2},
gref=iμref2L2ω,αj=ajL=grefm2(s^j)εj,Θj=∑i>jm2(s^i)εi,κ=Kμref,(37)g_{\rm ref}=\frac{i\mu_{\rm ref}^2L}{2\omega},\qquad\alpha_j=a_jL=\frac{g_{\rm ref}m_2(\hat{s}_j)}{\varepsilon_j},\qquad\Theta_j=\sum_{i>j}\frac{m_2(\hat{s}_i)}{\varepsilon_i},\qquad\kappa=\frac{K}{\mu_{\rm ref}}, \tag*{(37)}

with εj∼Exp⁡(1)\varepsilon_j\sim\operatorname{Exp}(1) (Sec. II B). Then K2/4ϑj=κ2/DjK^2/4\vartheta_j=\kappa^2/D_j with

Dj=Θj+S˙j+grefS^j,S^j=SjL.(38)D_j=\Theta_j+\frac{\dot{S}_j^+}{g_{\rm ref}\hat{S}_j},\qquad\hat{S}_j=\frac{S_j}{L}. \tag*{(38)}

For a static medium χ\chi is the opacity, m2=1m_2=1 and gref=iγg_{\rm ref}=i\gamma with γ=μ2L/(2ω)=1/x\gamma=\mu^2L/(2\omega)=1/x, where, as in Ref. [5], x=ω/ωˉcx=\omega/\bar{\omega}_c and ωˉc=μ2L/2\bar{\omega}_c=\mu^2L/2. The kinematic cut K=ωK=\omega then corresponds to κ2=Rˉx2/2\kappa^2=\bar{R}x^2/2, with Rˉ=ωˉcL=μ2L2/2\bar{R}=\bar{\omega}_cL=\mu^2L^2/2; Rˉ→∞\bar{R}\to\infty means no cut.

Sampling. With the opacity coordinate X(s^)=∫0s^ds^′ χX(\hat{s})=\int_0^{\hat{s}}\mathrm{d}\hat{s}'\,\chi, the probability of a center in a slice is χ ds^=dX\chi\,\mathrm{d}\hat{s}=\mathrm{d}X: in the variable XX the process is homogeneous with unit rate on [0,χ][0,\chi], where χ≡X(1)\chi\equiv X(1) is the total opacity. We draw points uniformly in XX, their number being Poisson distributed with mean equal to the length of the interval in XX, and map them back with s^=X−1(X)\hat{s}=X^{-1}(X). Stretches without medium (n=0n=0) receive no centers. In Mecke’s formula, ∫n ds=∫dX\int n\,\mathrm{d}s=\int\mathrm{d}X.

Case 1: spectrum integrated over kk

By Eq. (33), ln⁡Φ=∑jln⁡(1+ajζj−)\ln\Phi=\sum_j\ln(1+a_j\zeta_j^-), where ζj−\zeta_j^- only depends on the centers before sjs_j. By Eqs. (36) and (35), with the width of the marked center integrated exactly,

I=∫0χdX∗ E[Re⁡V(grefm2(s^∗)ζ^−(s^∗))],(39)\mathcal{I}=\int_0^\chi\mathrm{d}X_*\ \mathbb{E}\left[\operatorname{Re}V\left(g_{\rm ref}m_2(\hat{s}_*)\hat{\zeta}^-(\hat{s}_*)\right)\right], \tag*{(39)}

with s^∗=X−1(X∗)\hat{s}_*=X^{-1}(X_*), and where ζ^−(s^∗)=ζ−/L\hat{\zeta}^-(\hat{s}_*)=\zeta^-/L just before s^∗\hat{s}_* is generated by the centers in (0,s^∗)(0,\hat{s}_*). One sample consists of: (a) X∗X_* uniform in [0,χ][0,\chi] (stratified in the code) and s^∗=X−1(X∗)\hat{s}_*=X^{-1}(X_*); (b) Poisson(X∗)(X_*) centers in (0,s^∗)(0,\hat{s}_*), uniform in XX, each with εi∼Exp⁡(1)\varepsilon_i\sim\operatorname{Exp}(1); (c) the Möbius recursion ζ^→ζ^+Δ\hat{\zeta}\to\hat{\zeta}+\Delta between centers and ζ^→ζ^/(1+αiζ^)\hat{\zeta}\to\hat{\zeta}/(1+\alpha_i\hat{\zeta}) at each center, from ζ^(0)=0\hat{\zeta}(0)=0 up to s^∗\hat{s}_*; and (d) X=χ Re⁡V(grefm2(s^∗)ζ^)X=\chi\,\operatorname{Re}V(g_{\rm ref}m_2(\hat{s}_*)\hat{\zeta}). Centers after s^∗\hat{s}_* are not sampled. The exact integration over the width of the marked center removes the largest source of variance, the Coulomb tail (ε→0\varepsilon\to0). As a further control variate we use the same estimator without earlier centers, i.e., the first order in opacity, whose mean ∫01ds^ χ Re⁡V(grefm2s^)\int_0^1\mathrm{d}\hat{s}\,\chi\,\operatorname{Re}V(g_{\rm ref}m_2\hat{s}) is computed by quadrature.

Cases 2 and 3: with a cut and differential

In Eq. (26) there remain the integral over t1t_1 and the sum over the marked center jj. Applying Eq. (36) to jj and passing to units of LL, ajdt1 n(sj)dsj=αjχ(s^j)dt^1ds^ja_j\mathrm{d}t_1\,n(s_j)\mathrm{d}s_j=\alpha_j\chi(\hat{s}_j)\mathrm{d}\hat{t}_1\mathrm{d}\hat{s}_j, so that

IK=Re⁡∫01dt^1∫t^11ds^j χ(s^j)E[αjS˙j−S˙j+(1−e−κ2/Dj)],S˙j+=S˙j−+αjS˙j.(40)\mathcal{I}_K=\operatorname{Re}\int_0^1\mathrm{d}\hat{t}_1\int_{\hat{t}_1}^1\mathrm{d}\hat{s}_j\,\chi(\hat{s}_j)\mathbb{E}\left[\frac{\alpha_j}{\dot{S}_j^-\dot{S}_j^+}\left(1-e^{-\kappa^2/D_j}\right)\right],\qquad\dot{S}_j^+=\dot{S}_j^-+\alpha_j\dot{S}_j. \tag*{(40)}

With u=s^j−t^1u=\hat{s}_j-\hat{t}_1, ∫01dt^1∫t^11ds^j=∫01du∫01−udt^1\int_0^1\mathrm{d}\hat{t}_1\int_{\hat{t}_1}^1\mathrm{d}\hat{s}_j=\int_0^1\mathrm{d}u\int_0^{1-u}\mathrm{d}\hat{t}_1. One sample consists of: (a) uu log-uniform in [umin⁡,1][u_{\min},1] and t^1\hat{t}_1 uniform in [0,1−u][0,1-u], with s^j=t^1+u\hat{s}_j=\hat{t}_1+u and weight w=χ(s^j)(1−u)uln⁡(1/umin⁡)w=\chi(\hat{s}_j)(1-u)u\ln(1/u_{\min}); (b) the centers between the emission and the marked center, Poisson(X(s^j)−X(t^1))(X(\hat{s}_j)-X(\hat{t}_1)), uniform in XX, with their εi\varepsilon_i; starting from (S^,S^˙)=(0,1)(\hat{S},\dot{\hat{S}})=(0,1) at t^1\hat{t}_1, they are propagated with the free stretches and the jumps (16), which gives S^j\hat{S}_j, S˙j−\dot{S}_j^- and, with the εj\varepsilon_j of the marked center, S˙j+\dot{S}_j^+; (c) the centers after the marked center, Poisson(χ−X(s^j))(\chi-X(\hat{s}_j)), uniform in XX, which only enter through Θj\Theta_j; (d) the centers before the emission, which do not contribute and are not sampled; and (e) the estimators

XK=Re⁡[wαj(1−e−κ2/Dj)S˙j−S˙j+],X_K=\operatorname{Re}\left[w\frac{\alpha_j\left(1-e^{-\kappa^2/D_j}\right)}{\dot{S}_j^-\dot{S}_j^+}\right],
Xκ2′=Re⁡[wαjS˙j−S˙j+e−κ2/DjDj].(41)X'_{\kappa^2}=\operatorname{Re}\left[w\frac{\alpha_j}{\dot{S}_j^-\dot{S}_j^+}\frac{e^{-\kappa^2/D_j}}{D_j}\right]. \tag*{(41)}

with EXK=IK\mathbb{E}X_K=\mathcal{I}_K and EXκ2′=dIK/dκ2\mathbb{E}X'_{\kappa^2}=d\mathcal{I}_K/d\kappa^2, so that

ωdIdω dκ2=2αsCRπEXκ2′.\omega\frac{d\mathcal{I}}{d\omega\,d\kappa^2}=\frac{2\alpha_s C_R}{\pi}\mathbb{E}X'_{\kappa^2}.

Here the width of the marked center cannot be integrated in closed form, since it enters S˙j+\dot{S}^{+}_{j} and, through it, DjD_j inside the exponential; it is sampled.

Log-uniform sampling in uu and truncation error

If εj\varepsilon_j is small, ∣αj∣|\alpha_j| is large and, without intermediate centers, αj/(S˙−−S˙+)=αj/(1+αju)∼1/u\alpha_j/(\dot{S}^{-}-\dot{S}^{+})=\alpha_j/(1+\alpha_j u)\sim1/u over a wide range of uu. With uu uniform the estimator would be unbounded; with the weight uu of the log-uniform sampling it is bounded. The log-uniform sampling needs a lower limit, and the region u<umin⁡u<u_{\min} is omitted, with

umin⁡=10−91+κmax⁡2∣gref∣max⁡m2.(42)u_{\min}=\frac{10^{-9}}{1+\kappa_{\max}^2|g_{\mathrm{ref}}|\max m_2}. \tag*{(42)}

This is a bias, not a statistical error, and we quantify it.

Physical argument. The region u≲umin⁡u\lesssim u_{\min} only contributes when ∣αj∣u≳1|\alpha_j|u\gtrsim1, i.e., for a tiny εj\varepsilon_j: a center with a huge width τj=μ2/4εj\tau_j=\mu^2/4\varepsilon_j, which gives a transverse kick q2∼4τjq^2\sim4\tau_j. The condition ∣αj∣u∼1|\alpha_j|u\sim1 states that uLuL is the formation time of a gluon with that momentum, uL∼2ω/q2uL\sim2\omega/q^2. At the boundary k2∼q2∼2ω/(umin⁡L)k^2\sim q^2\sim2\omega/(u_{\min}L), that is, k2/ω2∼2/(ωLumin⁡)k^2/\omega^2\sim2/(\omega L u_{\min}), which is of order 10910^9 for the parameters below. Such a kick is so hard that the gluon comes out with k≫ωk\gg\omega, outside the kinematic cut k≤ωk\leq\omega and outside the soft-gluon approximation in which the whole formula is written.

Size. At first order in opacity (a single center, the marked one) the omitted part can be computed exactly by quadrature. For a static medium with the kinematic cut k≤ωk\leq\omega, the omitted fraction ranges from −10−15-10^{-15} to −10−22-10^{-22} for x=ω/ωˉcx=\omega/\bar{\omega}_c between 0.01 and 10 and Rˉ=103\bar{R}=10^3–4⋅1044\cdot10^4. Without cut (γ=1\gamma=1, umin⁡=10−9u_{\min}=10^{-9}) it is 3.0⋅10−93.0\cdot10^{-9}, in agreement with the analytic estimate Re⁡V(iγumin⁡)≃π2γumin⁡\operatorname{Re}V(i\gamma u_{\min})\simeq\frac{\pi}{2}\gamma u_{\min} divided by the first order. With several centers the order of magnitude should be the same, since the omitted region is dominated by the hard kicks of a single center.

Control variate, all kk at once, and real part

Control variate. In the same sample we compute Y=Re⁡[wαj/(S˙j−S˙j+)]Y=\operatorname{Re}[w\alpha_j/(\dot{S}^{-}_j\dot{S}^{+}_j)], the estimator without cut, whose mean is the integrated spectrum I\mathcal{I}. We take I\mathcal{I} from an independent run of the estimator of § 5 C, I^±δI^\hat{\mathcal{I}}\pm\delta\hat{\mathcal{I}}, and return Xˉ−c(Yˉ−I^)\bar{X}-c(\bar{Y}-\hat{\mathcal{I}}) with c=Cov⁡(X,Y)/Var⁡(Y)c=\operatorname{Cov}(X,Y)/\operatorname{Var}(Y). The variance is reduced by a factor 1−corr⁡(X,Y)21-\operatorname{corr}(X,Y)^2, and the error of I^\hat{\mathcal{I}} is propagated, c2δI2c^2\delta\mathcal{I}^2. Since the case-1 estimator is much more precise than Yˉ\bar{Y}, the control variate is effective at little cost.

All kk at once. κ2\kappa^2 only enters the last factor of Eq. (41), so every configuration serves a whole grid of κ2\kappa^2, and the covariance between different κ2\kappa^2 is obtained as well.

Real part, sample by sample. Re and E\mathbb{E} are linear and commute, so we take the real part of each sample before averaging. This has two advantages: the variance of Re⁡X\operatorname{Re}X does not exceed that of XX, and the imaginary vacuum constant from τ2→τ1+\tau_2\to\tau_1^+ (Sec. III C) disappears in each sample. What cannot be done is to take the real part inside the computation of a configuration, since Re does not commute with products, quotients or logarithms. The oscillator is genuinely complex, αj=2iτj/ω\alpha_j=2i\tau_j/\omega. S^\hat{S} and S˙\dot{S} are propagated as complex numbers, and their phases are the LPM interference.

Physical variables and conventions for expanding media

Time is measured from the production point, and there may be vacuum before the medium; reversing the profile in time gives a different result. Following [17], we use ω=ωˉc\omega=\bar{\omega}_c with ωˉc=∫dt tq^\bar{\omega}_c=\int dt\,t\hat{q}, so that gref=(i/2I1χ)g_{\mathrm{ref}}=(\mathrm{i}/2I_1\chi) with I1=∫01ds^ s^χm2I_1=\int_0^1d\hat{s}\,\hat{s}\chi m_2, and the kinematic cut K=ωK=\omega corresponds to κ2=23Rˉx2I22/(xI2)\kappa^2=\frac{2}{3}\bar{R}x^2I_2^2/(xI_2) with I2=∫01ds^ s^2χm2I_2=\int_0^1d\hat{s}\,\hat{s}^2\chi m_2 (static medium: Rˉx2/2\bar{R}x^2/2).

Cost

The cost per sample does not depend on the profile, which only changes the Poisson intensity and m2m_2. On a laptop (two physical cores), 10610^6 samples of the estimator of § 5 take about 3 s and give a relative error of (2(2–4)⋅10−44)\cdot10^{-4}; 10610^6 samples of the estimators (41), for a grid of six values of κ2\kappa^2, take 3–5 s, with relative errors of about 1% for the differential spectrum. The tasks are independent, with a single final reduction; with two processes the speed-up is 1.6.

Validation and comparison with published results

All numbers in this section are obtained with the publicly available code, with the statistical errors shown. The code and the scripts that produce the figures were written with the assistance of an AI tool (see Acknowledgments) and validated as described in this section. The curves of Refs. [5, 17] are read from the vector graphics of the PDF files, which gives the curves with the precision of the drawing; as a check of this reading and of the normalization, the published first-order (GLV) curves of Ref. [5] are reproduced to 0.9996–1.0002 with αs=0.3\alpha_s=0.3 and CR=4/3C_R=4/3.

Internal checks

The closed-form t2t_2 integral of Sec. III C agrees with a direct quadrature of the integrand to 10−1110^{-11}; the multilinear polynomial (17) and the determinant of Sec. IV B agree with the recursions to machine precision; and the integrated spectrum obtained from the estimator of § 5 C agrees with the K→∞K \to\infty limit of the estimator of § 5 D, which is a different estimator of the same quantity. The integral over κ2\kappa^2 of the differential estimator reproduces the estimator with a cut sample by sample. The bias from the truncation umin⁡u_{\min} was discussed in § 5 E.

First order in opacity

Table II compares the first order in opacity of the master formula, computed by quadrature, with the closed expressions (B.3) and (B.4) of Ref. [5] for the differential spectrum and for the spectrum with the kinematic cut. Without cut, the first order of Sec. IV C reproduces the GLV result to twelve digits.

xxκ2\kappa^2master formulaRef. [5]
cut, Rˉ=50\bar{R}=500.050.06250.05328910.0532891
1250.29438960.2943896
56250.12019060.1201826
cut, Rˉ=1000\bar{R}=10000.051.250.78625870.7862587
15000.50259570.5025957
5125000.13615010.1361425
differential0.030.031.23589261.2358926
0.310.10969060.1096906
1100.01781580.0178158

Table II. First order in opacity (static medium, γ=1/x\gamma=1/x): master formula at N=1N=1 (quadrature) versus Eqs. (B.4) (cut, κ2=Rˉx2/2\kappa^2=\bar{R}x^2/2) and (B.3) (differential) of Ref. [5], per unit of 2αsCRx/π2\alpha_s C_R x/\pi.

Comparison with the results of Andrés et al.

The master formula (26) has the same structure as Eq. (3.3) of Ref. [5]: the integral over t2t_2 is done analytically, one scattering is singled out, the kernel runs from the emission to that scattering and the broadening from it onwards, and the integrand vanishes outside the medium. Through Mecke’s formula (§ 5 A), the sum over jj becomes the integral ∫ds n(s)∫dτ ρs\int ds\,n(s)\int d\tau\,\rho_s, the analogue of the insertion ∫ds n(s) σ\int ds\,n(s)\,\sigma in that equation.

Static medium. Figure 1 compares the spectrum for a static medium with χ=5\chi=5, without cut and with the kinematic cut k≤ωk\leq\omega for several values of Rˉ\bar{R}, with Fig. 3 of Ref. [5]. Without cut, their curve lies 0.1–0.4 % below our result.

Figure 1. Static medium, χ=5\chi=5: spectrum without cut and with the kinematic cut k≤ωk\leq\omega for several Rˉ\bar{R}. Lines: Fig. 3 of Ref. [5]; points: this work. Lower panel: relative difference, Ref. [5]/this work −1-1, in %. Published points below the frame of the published panel (ω dI/dω<10−3\omega\,dI/d\omega<10^{-3}) are not compared. The figure was produced with scripts written with the assistance of an AI tool (see Sec. VI and Acknowledgments).

our result. With the cut, their curves lie 0.7–1.2 % below our results for x≥1x\geq1, while at small xx they lie above, by an amount that grows as xx and Rˉ\bar{R} decrease: up to 8.6 % for Rˉ=1000\bar{R}=1000 and 16 % for Rˉ=116\bar{R}=116 in the range shown. The statistical errors are below 0.35 %.

Differential spectrum. Figure 2 compares the spectrum differential in κ2\kappa^2 with the curves of Figs. 1 and 2 of Ref. [5], for χ=1\chi=1 and χ=5\chi=5, at 56 points each. Their values lie below ours by 1.05% on average (rms 1.1%) for χ=1\chi=1 and by 1.4% (rms 1.6%, maximum 4.8%) for χ=5\chi=5, with a mean statistical error of 0.5%.

Figure 2. Spectrum differential in κ2\kappa^2: relative difference between the curves of Figs. 1 (χ=1\chi= 1) and 2 (χ=5\chi= 5) of Ref. [5] and this work, Ref. [5]/this work − 1 in %, at the published points; the color gives κ2\kappa^2. The figure was produced with scripts written with the assistance of an AI tool (see Sec. VI and Acknowledgments).

Expanding media. Figure 3 compares our results with Fig. 3 of Ref. [17] for χ=5\chi=5: a static medium, a power-law profile, and the hydrodynamic profile read from the inset of that figure, with time measured from thydrot_{\rm hydro}; with this choice the equivalent static medium has Rˉ=1029\bar{R}=1029, close to the value Rˉ=1000\bar{R}=1000 quoted there. Following that reference, the power-law profile is n(t)=n0/(t+t0)αn(t)=n_0/(t+t_0)^\alpha, μ2(t)=μ02/(t+t0)2α\mu^2(t)=\mu_0^2/(t+t_0)^{2\alpha}, with α=0.5\alpha=0.5 and t0/L=0.1t_0/L=0.1, normalized to χ=5\chi=5, and the hydrodynamic one is n=k1Tn=k_1T, μ2=k2T2\mu^2=k_2T^2 with k1=0.5k_1=0.5 and k2=45k_2=45 along the temperature profile T(t)T(t) of the inset, which gives χ=4.97\chi=4.97. Without cut, their curves lie 0–1.6% below our results. With the kinematic cut, at small xx their curves for the expanding media lie well above ours: by 46% (power law) and 41% (hydrodynamic profile) at x=0.008x = 0.008, decreasing to 16 % and 13 % at x=0.02x = 0.02 and below 2 % for x≥0.1x \geq0.1; for the static medium the difference at x=0.008x = 0.008 is 3 %. At large xx their curves with cut fall below ours, by 4–9 % at x=5x = 5–10 and by about 20 % at x=30x = 30 for the expanding media (6 % and 18 % at x=10x = 10 and 30 for the static one). The statistical errors are below 0.25 %.

Summary and Outlook

We have computed the BDMPS-Z spectrum in the soft-gluon limit for a Yukawa potential, to all orders in opacity, by writing the Yukawa dipole cross section as a Laplace superposition of Gaussians. This turns the medium into a Poisson gas of Gaussian scattering centers, each configuration of which is a harmonic oscillator with an impulsive frequency, solved in closed form. The result is a master formula for the spectrum differential in k⊥k_\perp and with a cut on k⊥k_\perp, valid for arbitrary profiles n(s)n(s) and μ(s)\mu(s), whose K→∞K \to\infty limit is the average of the logarithm of a Gelfand–Yaglom determinant. Its Monte Carlo evaluation is efficient, gives the whole k⊥k_\perp dependence from a single run, and is trivially parallelizable. It reproduces the first order in opacity and the published results without cut, while with a kinematic cut we find sizeable relative deviations from the published results, both at small gluon energy fraction, where the cut strongly suppresses the spectrum, and at large energy fraction, where they reach 5–20 %.

We expect the construction not to rely on the soft limit in an essential way: what it requires is that, configuration by configuration, the problem is Gaussian. We expect it to extend to the more complex building blocks of an in-medium parton cascade, such as finite energy fraction and emissions off color antennas (for the latter, see Ref. [26] for a recent numerical treatment). The master formula also provides a natural starting point for analytical approximations, which we will address in future work. Finally, the independence of the samples makes the method well suited to massively parallel hardware.

Data Availability

The code used for all numerical results of this article is openly available at https://github.com/ manoel-maker/poisson-gas-bdmps_z. The scripts that produce the figures and the curves read from Refs. [5, 17] are available from the author upon reasonable request.

Acknowledgments

The author used Claude (Anthropic; several model versions, through claude.ai and Claude Code, between June and November 2026) as a research assistant. The representation of the Yukawa dipole cross section as a Laplace superposition of Gaussians, the central derivations, the choice of what to compute and to include, and the conclusions are the author’s; the Monte Carlo strategy was developed in the work with Claude. Claude was used to check derivations, including independent numerical checks; to write and validate the Monte Carlo code and the scripts that produce the figures (see Sec. VI); to search for and verify references; and to draft the English text of parts of the manuscript from the author’s notes. The abstract and § 1 and § 2 follow the author’s own drafts. The author checked the derivations by hand, reviewed the code and the text, and takes full responsibility for the content.

Static, power-law and hydrodynamic media with $\chi=5$, without cut and with the kinematic cut $k\le\omega$

Figure 3 (PDF p. 11). Static, power-law and hydrodynamic media with χ=5\chi=5, without cut (left) and with the kinematic cut k≤ωk\le\omega and Rˉ=1000\bar{R}=1000 (right). Lines: Fig. 3 of Ref. [17]; points: this work. Lower panels: relative difference, Ref. [17]/this work −1-1, in %. The figure was produced with scripts written with the assistance of an AI tool (see Sec. VI and Acknowledgments).

Continuous route: Jacobi equation and Arnold’s propagator

The exponent of KcK_c in Eq. (11) is ∫[iω2r˙2−U(s)r2]\int[\frac{i\omega}{2}\dot{r}^{2}-U(s)r^{2}] with U=∑jτjδ(s−sj)U=\sum_j\tau_j\delta(s-s_j). The classical equation is r¨=∑jajδ(s−sj)r\ddot{r}=\sum_j a_j\delta(s-s_j)r, with aja_j as in Eq. (14); in Arnold’s notation [19], c¨=−ω02(t)c\ddot{c}=-\omega_0^2(t)c with ω02(t)=−∑jajδ(t−sj)\omega_0^2(t)=-\sum_j a_j\delta(t-s_j). The Jacobi solution S(t2;t1)S(t_2;t_1) satisfies

∂t22S=∑jajδ(t2−sj)S(43)\partial_{t_2}^{2}S=\sum_j a_j\delta(t_2-s_j)S \tag*{(43)}

with S(t1;t1)=0S(t_1;t_1)=0 and ∂t2S∣t2=t1=1\partial_{t_2}S|_{t_2=t_1}=1. Between two centers SS is linear in t2t_2, and at each center sj>t1s_j>t_1 it is continuous while its slope jumps as in Eq. (16). The products of δ\delta with the solution are well defined without any prescription: SS is continuous at sjs_j, the discontinuity is in S˙\dot{S}, which never multiplies a δ\delta, and integrating Eq. (43) over [sj−ϵ,sj+ϵ][s_j-\epsilon,s_j+\epsilon] gives Eq. (16).

Since the exponent is quadratic, the path integral is the exponential of the classical action times the Van Vleck prefactor.

In two transverse dimensions,

Kc(t2,z;t1,y)=ω2πiSexp⁡{iω2S[S˙z2+Cy2−2z⋅y]}(44)K_c(t_2,z;t_1,y)=\frac{\omega}{2\pi i S}\exp\left\{\frac{i\omega}{2S}\left[\dot{S}z^2+Cy^2-2z\cdot y\right]\right\} \tag*{(44)}

with S=S(t2;t1)S=S(t_2;t_1) and C=−∂t1SC=-\partial_{t_1}S. For y=0y=0 the classical path is r(s)=zS(s;t1)/S(t2;t1)r(s)=zS(s;t_1)/S(t_2;t_1) and the classical action is ω2z2S˙/S\frac{\omega}{2}z^2\dot{S}/S. Comparing with Eq. (13) at t=t2t=t_2 gives q=1/Sq=1/S, p=S/S˙p=S/\dot{S} and N=ω/(2πiS)\mathcal{N}=\omega/(2\pi iS), which is Eq. (15): both routes define the same SS. Then

∇yKc∣y=0=−ω22πS2zeiωS˙z2/(2S)\nabla_yK_c|_{y=0}=-\frac{\omega^2}{2\pi S^2}z e^{i\omega\dot{S}z^2/(2S)}

which is Eq. (3.3) of Arnold [19] with M=ωM=\omega, B=zB=z, and Eq. (19) follows. In this route the continuity of ϑ\vartheta is checked directly: when t2t_2 crosses sjs_j, TT decreases by τj\tau_j and S˙\dot{S} increases by ajSja_jS_j, so that η\eta changes by iω2aj=−τj\frac{i\omega}{2}a_j=-\tau_j and ϑ=T−η\vartheta=T-\eta does not change.

HTL potential with a magnetic mass

The HTL potential [7] is, up to a constant prefactor CC,

V(q)=CmD2q2(q2+mD2)=C[1q2−1q2+mD2](45)V(q)=C\frac{m_D^2}{q^2(q^2+m_D^2)}=C\left[\frac{1}{q^2}-\frac{1}{q^2+m_D^2}\right] \tag*{(45)}

The term 1/q21/q^2, from the unscreened magnetostatic gluons, makes the total elastic cross section logarithmically divergent in the infrared, so that the Poisson intensity of § 2 C would be infinite. A magnetic mass mMm_M regularizes it,

VM(q)=CmD2−mM2(q2+mM2)(q2+mD2)=C[1q2+mM2−1q2+mD2],(46)V_M(q) = C \frac{m_D^2-m_M^2}{(q^2+m_M^2)(q^2+m_D^2)} = C\left[\frac{1}{q^2+m_M^2}-\frac{1}{q^2+m_D^2}\right], \tag*{(46)}

which has the form of the effective cross section of Ref. [27] (their Eq. (9), with μM=mM\mu_M=m_M, μE=mD\mu_E=m_D). Writing 1/(q2+m2)=∫0∞dλ e−λ(q2+m2)1/(q^2+m^2)=\int_0^\infty d\lambda\,e^{-\lambda(q^2+m^2)},

VM(q)=C∫0∞dλ e−λq2[e−λmM2−e−λmD2],(47)V_M(q) = C\int_0^\infty d\lambda\,e^{-\lambda q^2}\left[e^{-\lambda m_M^2}-e^{-\lambda m_D^2}\right], \tag*{(47)}

a superposition of Gaussians whose weight is positive for every λ\lambda if and only if mM<mDm_M<m_D. Performing the Gaussian integral over q\mathbf{q} and changing variables to τ=1/(4λ)\tau=1/(4\lambda),

σ(r)=C4π∫0∞dττ[e−mM2/4τ−e−mD2/4τ](1−e−τr2)=C2π[ln⁡mDmM−K0(mMr)+K0(mDr)].(48)\begin{aligned} \sigma(r) = \frac{C}{4\pi}\int_0^\infty\frac{d\tau}{\tau}\left[e^{-m_M^2/4\tau}-e^{-m_D^2/4\tau}\right]\left(1-e^{-\tau r^2}\right) \\ &= \frac{C}{2\pi}\left[\ln\frac{m_D}{m_M}-K_0(m_Mr)+K_0(m_Dr)\right]. \tag*{(48)} \end{aligned}

The total cross section σ(∞)=C4πln⁡(mD2/mM2)\sigma(\infty)=\frac{C}{4\pi}\ln(m_D^2/m_M^2) is finite, and v=12nσ=nM∫dτ ρM(τ)(1−e−τr2)v=\frac{1}{2}n\sigma=n_M\int d\tau\,\rho_M(\tau)\left(1-e^{-\tau r^2}\right) with nM=12nσ(∞)n_M=\frac{1}{2}n\sigma(\infty) and

ρM(τ)=e−mM2/4τ−e−mD2/4ττln⁡(mD2/mM2).(49)\rho_M(\tau)=\frac{e^{-m_M^2/4\tau}-e^{-m_D^2/4\tau}}{\tau\ln(m_D^2/m_M^2)}. \tag*{(49)}

This is a probability density, positive if and only if mM<mDm_M<m_D, with the same Coulomb tail as in the Yukawa case, ρM≃(mD2−mM2)/[4τ2ln⁡(mD2/mM2)]\rho_M\simeq(m_D^2-m_M^2)/[4\tau^2\ln(m_D^2/m_M^2)] for τ→∞\tau\to\infty. Everything in § 2 C–V then applies with n→nMn\to n_M and ρs→ρM\rho_s\to\rho_M: only the sampling of the widths changes. The condition μM<μE\mu_M<\mu_E imposed in Ref. [27] is precisely the positivity of this Poisson gas. For mM→0m_M\to0 the intensity diverges logarithmically: the unregularized HTL potential is still a superposition of Gaussians with positive weights, but not a Poisson gas with finite intensity. The magnetic mass is of order g2Tg^2T and nonperturbative; Ref. [27] takes 0.4<μM/μE<0.60.4<\mu_M/\mu_E<0.6 from nonperturbative estimates.

References

  1. [1]R. Baier, Y. L. Dokshitzer, A. H. Mueller, S. Peigné, and D. Schiff, Radiative energy loss of high-energy quarks and gluons in a finite volume quark-gluon plasma, Nucl. Phys. B 483, 291 (1997), hep-ph/9607355.DOI
  2. [2]R. Baier, Y. L. Dokshitzer, A. H. Mueller, S. Peigné, and D. Schiff, Radiative energy loss and $p_t$ broadening of high-energy partons in nuclei, Nucl. Phys. B 484, 265 (1997), hep-ph/9608322.DOI
  3. [3]B. G. Zakharov, Fully quantum treatment of the Landau-Pomeranchuk-Migdal effect in QED and QCD, JETP Lett. 63, 952 (1996), hep-ph/9607404.DOI
  4. [4]B. G. Zakharov, Radiative energy loss of high-energy quarks in finite size nuclear matter and quark gluon plasma, JETP Lett. 65, 615 (1997), hep-ph/9704255.DOI
  5. [5]C. Andres, L. Apolinário, and F. Dominguez, Medium-induced gluon radiation with full resummation of multiple scatterings for realistic parton-medium interactions, JHEP 07 (2020), 114, 2002.01517.DOI
  6. [6]M. Gyulassy and X.-N. Wang, Multiple collisions and induced gluon bremsstrahlung in QCD, Nucl. Phys. B 420, 583 (1994), nucl-th/9306003.DOI
  7. [7]P. Aurenche, F. Gelis, and H. Zaraket, A simple sum rule for the thermal gluon spectral function and applications, JHEP 05 (2002), 043, hep-ph/0204146.
  8. [8]M. Gyulassy, P. Levai, and I. Vitev, Reaction operator approach to non-Abelian energy loss, Nucl. Phys. B 594, 371 (2001), nucl-th/0006010.DOI
  9. [9]M. Gyulassy, P. Levai, and I. Vitev, Non-Abelian energy loss at finite opacity, Phys. Rev. Lett. 85, 5535 (2000), nucl-th/0005032.DOI
  10. [10]U. A. Wiedemann, Gluon radiation of hard quarks in a nuclear environment: opacity expansion, Nucl. Phys. B 588, 303 (2000), hep-ph/0005129.DOI
  11. [11]C. A. Salgado and U. A. Wiedemann, Calculating quenching weights, Phys. Rev. D 68, 014008 (2003), hep-ph/0302184.DOI
  12. [12]Y. Mehtar-Tani, Gluon bremsstrahlung in finite media beyond multiple soft scattering approximation, JHEP 07 (2019), 057, 1903.00506.
  13. [13]Y. Mehtar-Tani and K. Tywoniuk, Improved opacity expansion for medium-induced parton splitting, JHEP 06 (2020), 187, 1910.02032.
  14. [14]J. Barata and Y. Mehtar-Tani, Improved opacity expansion at NLO for medium induced gluon radiation, JHEP 10 (2020), 176, 2004.02323.DOI
  15. [15]J. H. Isaksen, A. Takacs, and K. Tywoniuk, A unified picture of medium-induced radiation, JHEP 02 (2023), 156, 2206.02811.DOI
  16. [16]S. Caron-Huot and C. Gale, Finite-size effects on the radiative energy loss of a fast parton in hot and dense strongly interacting matter, Phys. Rev. C 82, 064902 (2010), 1006.2379.
  17. [17]C. Andres, L. Apolinário, F. Dominguez, and M. Gonzalez Martinez, In-medium gluon radiation spectrum with all-order resummation of multiple scatterings in longitudinally evolving media, JHEP 11 (2024), 025, 2307.06226.DOI
  18. [18]X. Feal and R. Vazquez, Intensity of gluon bremsstrahlung in a finite plasma, Phys. Rev. D 98, 074029 (2018), 1811.01591.DOI
  19. [19]P. B. Arnold, Simple formula for high-energy gluon bremsstrahlung in a finite, expanding medium, Phys. Rev. D 79, 065025 (2009), 0808.2767.DOI
  20. [20]D. V. Widder, The Laplace Transform, Princeton Mathematical Series, Vol. 6 (Princeton University Press, Princeton, 1941).
  21. [21]J. Schwinger, On gauge invariance and vacuum polarization, Phys. Rev. 82, 664 (1951).
  22. [22]G. Last and M. Penrose, Lectures on the Poisson Process, IMS Textbooks (Cambridge University Press, Cambridge, 2017).DOI
  23. [23]I. M. Gel’fand and A. M. Yaglom, Integration in functional spaces and its applications in quantum physics, J. Math. Phys. 1, 48 (1960).DOI
  24. [24]F. James, Monte Carlo theory and practice, Rep. Prog. Phys. 43, 1145 (1980).
  25. [25]J. Mecke, Stationäre zufällige Maße auf lokal-kompakten Abelschen Gruppen, Z. Wahrsch. verw. Geb. 9, 36 (1967).
  26. [26]C. Andres, L. Apolinário, N. Armesto, A. Cordeiro, F. Domínguez, P. Guerrero-Rodríguez, and J. G. Mil-hano, Gluon radiation from a QCD antenna with realistic parton-medium interactions, arXiv:2607.00629 [hep-ph] (2026).
  27. [27]M. Djordjevic and M. Djordjevic, Generalization of radiative jet energy loss to non-zero magnetic mass, Phys. Lett. B 709, 229 (2012), 1105.4359.DOI

Paper details

vlnc:W_ugj5cyd2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 6

downloads · 0

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

  • Physics and Astronomy

Claude

View statement
Looking for related papers…
Contents