信号功率谱Sh 、分数能量密度谱Ω、特征应变hc
随机背景功率谱
对于随机背景信号 P 395
各向同性随机背景
h a b ( t , r ⃗ ) = ∑ P = + , × ∫ d f ∫ d 2 n ^ h ~ P ( f , n ^ ) e a b P ( n ^ ) e − 2 π i f ( t − n ^ ⋅ r ⃗ c ) h_{ab}(t,\vec{r}) = \sum_{P=+,\times} \int df \int d^2 \hat{n} \tilde{h}_P(f, \hat{n}) e^P_{ab}(\hat{n}) e^{-2\pi i f(t-\hat{n}\cdot \frac{\vec{r}}{c})}
h a b ( t , r ) = P = + , × ∑ ∫ d f ∫ d 2 n ^ h ~ P ( f , n ^ ) e a b P ( n ^ ) e − 2 π i f ( t − n ^ ⋅ c r )
⟨ h ~ a b ( f , n ^ ) h ~ a b ( f ′ , n ^ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) 4 π δ ( f − f ′ ) 2 S a b ( f ) \langle \tilde{h}_{ab}(f, \hat{n})~\tilde{h}_{ab}(f',\hat{n}) \rangle = \frac{\delta^2(\hat{n}, \hat{n}')}{4\pi} \frac{\delta(f-f')}{2} S_{ab}(f)
⟨ h ~ a b ( f , n ^ ) h ~ a b ( f ′ , n ^ ) ⟩ = 4 π δ 2 ( n ^ , n ^ ′ ) 2 δ ( f − f ′ ) S a b ( f )
基于随机背景信号的平稳(及各向同性无偏振)假设,上述结果可由从维纳—辛钦(Winner-Khitchine)定理得到。方便起见,这里h h h 不带下指标,可对应任意指标下的引力波取值。
⟨ h ~ ∗ ( f ) h ~ ( f ′ ) ⟩ = ∬ ⟨ h ∗ ( t ) h ( t ′ ) ⟩ e − i 2 π f t e i 2 π f ′ t ′ d t d t ′ = ∫ e − i 2 π ( f − f ′ ) t ′ d t ′ ∫ r ( τ ) e i 2 π f τ d τ = δ ( f − f ′ ) 1 2 S h ( f ) \begin{aligned}
\langle \tilde{h}^*(f) \tilde{h}(f') \rangle &= \iint \langle h^*(t) h(t')\rangle e^{-i 2\pi f t} e^{i 2\pi f' t'} dt dt' \\
&= \int e^{-i 2\pi (f-f') t'} dt' \int r (\tau) e^{i 2\pi f \tau}d\tau\\
&= \delta(f-f') \frac{1}{2} S_h(f)
\end{aligned} ⟨ h ~ ∗ ( f ) h ~ ( f ′ ) ⟩ = ∬ ⟨ h ∗ ( t ) h ( t ′ ) ⟩ e − i 2 π f t e i 2 π f ′ t ′ d t d t ′ = ∫ e − i 2 π ( f − f ′ ) t ′ d t ′ ∫ r ( τ ) e i 2 π f τ d τ = δ ( f − f ′ ) 2 1 S h ( f )
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) 4 π δ P P ′ δ ( f − f ′ ) 2 S P ( f ) \langle \tilde{h}_P(f, \hat{n})~\tilde{h}_{P'}(f',\hat{n}) \rangle = \frac{\delta^2(\hat{n}, \hat{n}')}{4\pi} \delta_{PP'} \frac{\delta(f-f')}{2} S_P(f)
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = 4 π δ 2 ( n ^ , n ^ ′ ) δ P P ′ 2 δ ( f − f ′ ) S P ( f )
不同方位相互独立
不同极化模式相互独立:the assumption is valid if and only if one works in a polarization basis that diagonalizes the kinetic matrix of the theory.
稳态
S + ( f ) = S × ( f ) = S h ( f ) S_{+}(f) = S_{\times}(f) = S_h(f) S + ( f ) = S × ( f ) = S h ( f )
S T ( f ) = S + ( f ) + S × ( f ) S_T(f) = S_{+}(f) + S_{\times}(f) S T ( f ) = S + ( f ) + S × ( f )
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) 4 π δ P P ′ 2 δ ( f − f ′ ) 2 S T ( f ) \langle \tilde{h}_P(f, \hat{n})~\tilde{h}_{P'}(f',\hat{n}) \rangle = \frac{\delta^2(\hat{n}, \hat{n}')}{4\pi} \frac{\delta_{PP'}}{2} \frac{\delta(f-f')}{2} S_{T}(f)
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = 4 π δ 2 ( n ^ , n ^ ′ ) 2 δ P P ′ 2 δ ( f − f ′ ) S T ( f )
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) 4 π δ P P ′ δ ( f − f ′ ) 2 S h ( f ) \boxed{\langle \tilde{h}_P(f, \hat{n})~\tilde{h}_{P'}(f',\hat{n}) \rangle = \frac{\delta^2(\hat{n}, \hat{n}')}{4\pi} \delta_{PP'} \frac{\delta(f-f')}{2} S_h(f)}
⟨ h ~ P ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ) ⟩ = 4 π δ 2 ( n ^ , n ^ ′ ) δ P P ′ 2 δ ( f − f ′ ) S h ( f )
分数能量密度谱
除功率谱之外S h ( f ) S_h(f) S h ( f ) ,随机背景另一种常用的描述是分数能量密度谱Ω ( f ) \Omega(f) Ω ( f ) ,代表单位对数频率间隔内的引力波能量密度占比:
Ω ( f ) ≡ 1 ρ c c 2 d ρ G W d ln f , ρ c = 3 H 0 2 8 π G \Omega(f) \equiv \frac{1}{\rho_c c^2}\frac{d\rho_{\tiny GW}}{d \ln f}, ~~~~ \rho_c = \frac{3 H_0^2}{8\pi G}
Ω ( f ) ≡ ρ c c 2 1 d ln f d ρ G W , ρ c = 8 π G 3 H 0 2
为建立功率 谱与分数能量 密度谱的关系,需要考虑引力波的有效能动张量。在广义相对论中,近似有:
t μ ν = c 4 32 π G ⟨ ⟨ ∂ μ h α β ∂ ν h α β ⟩ ⟩ t_{\mu\nu} = \frac{c^4}{32\pi G} \left\langle \left\langle \partial_\mu h_{\alpha\beta} \partial_\nu h^{\alpha\beta} \right\rangle \right\rangle
t μ ν = 3 2 π G c 4 ⟨ ⟨ ∂ μ h α β ∂ ν h α β ⟩ ⟩
这里双尖括号为时空平均,其中引力波能量密度所对应的t 00 t^{00} t 0 0 在TT规范下为:
ρ G W = t 00 = c 2 32 π G ⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩ \rho_{\rm GW} = t^{00} = \frac{c^2}{32\pi G} \big\langle \dot{h}_{ab}(t)~\dot{h}^{ab}(t) \big\rangle
ρ G W = t 0 0 = 3 2 π G c 2 ⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩
这里尖括号可理解为时间平均或空间平均,也对应于系综平均。考虑到分数能量密度谱的定义,有:
⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩ = 12 H 0 2 ∫ Ω ( f ) d ln f \big\langle \dot{h}_{ab}(t)~\dot{h}^{ab}(t) \big\rangle = 12 H_0^2 \int \Omega(f) ~ d \ln f
⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩ = 1 2 H 0 2 ∫ Ω ( f ) d ln f
另一方面,考虑到随机背景引力波h a b ( t , r ⃗ ) h_{ab}(t, \vec{r}) h a b ( t , r ) 以及S h ( f ) S_h(f) S h ( f ) 的定义式:
⟨ h a b ( t ) h a b ( t ) ⟩ = 4 ∫ S h ( f ) d f ⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩ = 4 ∫ ( 2 π f ) 2 S h ( f ) d f \begin{aligned}
\big\langle h_{ab}(t)~h^{ab}(t) \big\rangle &= 4 \int S_h(f) df\\
\big\langle \dot{h}_{ab}(t)~\dot{h}^{ab}(t) \big\rangle &= 4 \int (2\pi f)^2 S_h(f) df
\end{aligned} ⟨ h a b ( t ) h a b ( t ) ⟩ ⟨ h ˙ a b ( t ) h ˙ a b ( t ) ⟩ = 4 ∫ S h ( f ) d f = 4 ∫ ( 2 π f ) 2 S h ( f ) d f
这里的4是因为(TT规范下)h a b h a b h_{ab}h^{ab} h a b h a b 是4个分量求和,考虑到各向同性无偏振假设,所有分量强度相同,对应整体乘以4。最终:
Ω ( f ) = 4 π 2 3 H 0 2 ∣ f ∣ 3 S h ( f ) \boxed{\Omega(f) = \frac{4\pi^2}{3H_0^2} |f|^3 S_h(f) }
Ω ( f ) = 3 H 0 2 4 π 2 ∣ f ∣ 3 S h ( f )
注意,上述关系建立在引力波的有效能动张量t μ ν t_{\mu\nu} t μ ν 之上,当前广义相对论下的t μ ν t_{\mu\nu} t μ ν 表达式由Isaacson最先得到(1968),被称为Isaacson’s formula。对超越广义相对论的理论,有效能动张量本身需要调整,最终Ω ( f ) \Omega(f) Ω ( f ) 与S h ( f ) S_h(f) S h ( f ) 的关系也将是理论依赖的,具体可参考 Isi & Stein (2018) 。即便t μ ν t_{\mu\nu} t μ ν 保持不变,仍需注意Ω ( f ) \Omega(f) Ω ( f ) 是对所有分量求和,而S h ( f ) S_h(f) S h ( f ) 仅对应单个分量,当有额外极化模式时,上述关系也将改变。
此外,注意不少文章中S h ( f ) S_h(f) S h ( f ) 被定义为两种极化模式之和 ,对应上面的S T ( f ) S_T(f) S T ( f ) ,尤其是地面引力波相关文章多采用这个约定,如 Abbott et al. (2018) ,综述文章 Thrane & Romano (2013) 及 Romano & Cornish (2017) 也都是这种约定。这种约定下:
Ω ( f ) = 2 π 2 3 H 0 2 ∣ f ∣ 3 S T ( f ) \Omega(f) = \frac{2\pi^2}{3H_0^2} |f|^3S_T(f)
Ω ( f ) = 3 H 0 2 2 π 2 ∣ f ∣ 3 S T ( f )
但空间引力波多采用之前约定,如 Babak et al. (2021) 、Auclair et al. (2023) 。
PTA也是这种约定???
其它常见功率谱
上面的S h S_h S h 是最常见的表述,但引力波随机背景的早期经典 Allen & Romano (1997) 中定义了另外两种功率谱,不做全天积分的功率谱H ( f ) H(f) H ( f ) 和探测信号的功率谱S ( f ) S(f) S ( f ) (文中均采用双边谱),也会出现在后续一些文章中,需要注意区分
不对偏振求和、不做全天积分的(双边)功率谱H ( f ) H(f) H ( f ) :⟨ h ~ P ∗ ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ′ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) δ P P ′ δ ( f − f ′ ) H P ( f ) \langle \tilde{h}^*_P(f,\hat{n}) \tilde{h}_{P'}(f',\hat{n}')\rangle = \delta^2(\hat{n},\hat{n}') \delta_{PP'} \delta(f-f') H_P(f)
⟨ h ~ P ∗ ( f , n ^ ) h ~ P ′ ( f ′ , n ^ ′ ) ⟩ = δ 2 ( n ^ , n ^ ′ ) δ P P ′ δ ( f − f ′ ) H P ( f )
H ( f ) = 1 4 π 1 2 S h ( f ) , Ω ( f ) = 32 π 3 3 H 0 2 ∣ f ∣ 3 H ( f ) H(f) = \frac{1}{4\pi} ~\frac{1}{2} S_h(f), ~~~~ \Omega(f) = \frac{32\pi^3}{3H_0^2} |f|^3 H(f)
H ( f ) = 4 π 1 2 1 S h ( f ) , Ω ( f ) = 3 H 0 2 3 2 π 3 ∣ f ∣ 3 H ( f )
Seto 相关文章中所使用的转换经常是这个公式。
探测信号的双边互功率谱S i j ( f ) S_{ij}(f) S i j ( f ) 及自功率谱S ( f ) S(f) S ( f ) ⟨ s ~ i ∗ ( f ) s ~ j ( f ′ ) ⟩ = δ ( f − f ′ ) S i j ( f ) \langle \tilde{s}^*_i(f) \tilde{s}_j(f')\rangle = \delta(f-f') S_{ij}(f)
⟨ s ~ i ∗ ( f ) s ~ j ( f ′ ) ⟩ = δ ( f − f ′ ) S i j ( f )
S i j ( f ) = ∑ P Γ i j P 1 2 S P ( f ) = 4 π ∑ P Γ i j P H P ( f ) S_{ij}(f) = \sum_P\Gamma^P_{ij} ~ \frac{1}{2}S_P(f) = 4\pi \sum_P \Gamma^P_{ij} H_P(f)
S i j ( f ) = P ∑ Γ i j P 2 1 S P ( f ) = 4 π P ∑ Γ i j P H P ( f )
其中Γ i j P \Gamma^P_{ij} Γ i j P 为探测器间的重叠约化函数(Overlap Reduction Function, ORF),后面具体介绍。当各极化强度相同,可定义总体ORF Γ i j = ∑ P Γ i j P = ⟨ R ⟩ γ i j \Gamma_{ij} = \sum_P\Gamma^P_{ij} = \langle\mathcal{R}\rangle \gamma_{ij} Γ i j = ∑ P Γ i j P = ⟨ R ⟩ γ i j S i j ( f ) = ⟨ R ⟩ γ i j 1 2 S h ( f ) = 4 π ⟨ R ⟩ γ i j H ( f ) , Ω ( f ) = 1 ⟨ R ⟩ 8 π 2 3 H 0 2 ∣ f ∣ 3 S i j ( f ) γ i j ( f ) \small S_{ij}(f) = \langle\mathcal{R}\rangle \gamma_{ij} ~ \frac{1}{2}S_h(f) = 4\pi\langle\mathcal{R}\rangle \gamma_{ij} H(f), ~~~~ \Omega(f) = \frac{1}{\langle\mathcal{R}\rangle}\frac{8\pi^2}{3H_0^2} |f|^3 \frac{S_{ij}(f)}{\gamma_{ij}(f)}
S i j ( f ) = ⟨ R ⟩ γ i j 2 1 S h ( f ) = 4 π ⟨ R ⟩ γ i j H ( f ) , Ω ( f ) = ⟨ R ⟩ 1 3 H 0 2 8 π 2 ∣ f ∣ 3 γ i j ( f ) S i j ( f )
这里R \mathcal{R} R 是探测器的响应函数,对于直角迈克尔逊其全天平均值⟨ R ⟩ = 2 / 5 \langle\mathcal{R}\rangle=2/5 ⟨ R ⟩ = 2 / 5 ;γ i j ( f ) \gamma_{ij}(f) γ i j ( f ) 为经过归一化的探测器重叠约化函数,对相同位置指向的探测器对,其取值为1。对地基探测器而言,探测信号双边谱S ( f ) = ⟨ R ⟩ 1 2 S h ( f ) = 1 5 S h ( f ) = 8 π 5 H ( f ) , Ω ( f ) = 20 π 2 3 H 0 2 ∣ f ∣ 3 S ( f ) S(f) = \langle\mathcal{R}\rangle ~ \frac{1}{2} S_h(f) = \frac{1}{5} S_h(f) = \frac{8\pi}{5} H(f), ~~~~ \Omega(f) = \frac{20\pi^2}{3H_0^2} |f|^3 S(f)
S ( f ) = ⟨ R ⟩ 2 1 S h ( f ) = 5 1 S h ( f ) = 5 8 π H ( f ) , Ω ( f ) = 3 H 0 2 2 0 π 2 ∣ f ∣ 3 S ( f )
显然这里探测信号的功率谱S ( f ) S(f) S ( f ) 是探测器依赖的,对于地基、空间以及PTA等各有不同,具体可参考 Mingarelli et al. (2019) 。
注意:Ω ( f ) \Omega(f) Ω ( f ) 默认为单边,而上述H ( f ) , S i j ( f ) , S ( f ) H(f), S_{ij}(f), S(f) H ( f ) , S i j ( f ) , S ( f ) ,与 Allen & Romano 文中保持一致,均为双边。采用更常用的单边谱有:
H ( f ) = 1 4 π S h ( f ) , Ω ( f ) = 16 π 3 3 H 0 2 ∣ f ∣ 3 H ( f ) H(f) = \frac{1}{4\pi} ~S_h(f), ~~~~ \Omega(f) = \frac{16\pi^3}{3H_0^2} |f|^3 H(f)
H ( f ) = 4 π 1 S h ( f ) , Ω ( f ) = 3 H 0 2 1 6 π 3 ∣ f ∣ 3 H ( f )
S ( f ) = 2 5 S h ( f ) , Ω ( f ) = 10 π 2 3 H 0 2 ∣ f ∣ 3 S ( f ) S(f) = \frac{2}{5} S_h(f), ~~~~~~~ \Omega(f) = \frac{10\pi^2}{3H_0^2} |f|^3 S(f)
S ( f ) = 5 2 S h ( f ) , Ω ( f ) = 3 H 0 2 1 0 π 2 ∣ f ∣ 3 S ( f )
特征应变振幅谱
除了S h ( f ) S_h(f) S h ( f ) 及Ω ( f ) \Omega(f) Ω ( f ) ,还有一个常见的物理量是特征应变强度谱h c ( f ) h_c(f) h c ( f ) ,但其定义并不明确,常见的有两种表述
h c 2 ( f ) d ln f = 8 f ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ d f v . s . h c 2 ( f ) d ln f = 2 S h ( f ) d f h^2_c(f) d\ln f = 8f \langle|\tilde{h}(f)|^2\rangle df ~~~{\rm v.s.}~~~ h^2_c(f) d\ln f = 2S_h(f) df
h c 2 ( f ) d ln f = 8 f ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ d f v . s . h c 2 ( f ) d ln f = 2 S h ( f ) d f
前者可参考 Flanagan & Hughes (1997) 及 Finn & Thorne (2000) ,后者可参考 Maggiore (2000) 及 Thrane & Romano (2013) 。虽然大致都理解为单位对数频率 间隔内的引力波等效振幅,但前者随积分时间增加而增加,是累积振幅,后者则不随积分时间改变。 300 Hundred Years of Gravitation (1989) 中 Kip Thorne 还给了另一种对频率积分后的h c h_c h c 定义(P369),不太常见,这里不做介绍。
两种定义的主要区别在于:对并和等持续时间相对短暂的信号,通常保持探测灵敏度曲线固定,调整信号的累积强度;对于连续引力波或随机背景等持续信号,则通常保持信号强度不变,调整探测器灵敏度曲线。因此,前一定义常用于旋进并和信号,后一定义则常出现在PTA对引力波随机背景的探测。
这里重点关注前一种,其定义源自信噪比公式(Finn & Thorne, 2000) :
⟨ ρ 2 ⟩ = 2 ∫ 0 ∞ ⟨ ∣ s ~ ( f ) ∣ 2 ⟩ 1 2 N ( f ) d f ≡ ∫ [ h c ( f ) h n ( f ) ] 2 d ln f \langle\rho^2\rangle = 2\int_0^\infty \frac{\left\langle|\tilde{s}(f)|^2\right\rangle}{\frac{1}{2}N(f)} df \equiv \int \left[\frac{h_c(f)}{h_n(f)}\right]^2 d\ln f
⟨ ρ 2 ⟩ = 2 ∫ 0 ∞ 2 1 N ( f ) ⟨ ∣ s ~ ( f ) ∣ 2 ⟩ d f ≡ ∫ [ h n ( f ) h c ( f ) ] 2 d ln f
这种形式下,对数对数图中,SNR大致正比于ln h c ( f ) \ln h_c(f) ln h c ( f ) 与ln h n ( f ) \ln h_n(f) ln h n ( f ) 曲线间面积。其中噪声部分:
h n ( f ) ≡ 2 f N ( f ) ⟨ R ⟩ = f N ( f ) ⟨ F + 2 ⟩ h_n(f) \equiv \sqrt{\frac{2 f N(f)}{\langle\mathcal{R}\rangle}} = \sqrt{\frac{f N(f)}{\langle F^2_+ \rangle}}
h n ( f ) ≡ ⟨ R ⟩ 2 f N ( f ) = ⟨ F + 2 ⟩ f N ( f )
对地基探测器⟨ R ⟩ ∼ 2 / 5 , h n ∼ 5 f N ( f ) \langle\mathcal{R}\rangle\sim 2/5, h_n \sim \sqrt{5 f N(f)} ⟨ R ⟩ ∼ 2 / 5 , h n ∼ 5 f N ( f ) 。对应的信号部分:
⟨ ∣ s ~ ( f ) ∣ 2 ⟩ ∼ ⟨ F + 2 ⟩ ⟨ ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ⟩ ∼ ⟨ F + 2 + F × 2 ⟩ ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ \left\langle|\tilde{s}(f)|^2\right\rangle \sim \langle F^2_+ \rangle \left\langle|\tilde{h}_+(f)|^2 + |\tilde{h}_\times(f)|^2 \right\rangle \sim \left\langle F^2_+ + F^2_\times \right\rangle \langle|\tilde{h}(f)|^2\rangle
⟨ ∣ s ~ ( f ) ∣ 2 ⟩ ∼ ⟨ F + 2 ⟩ ⟨ ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ⟩ ∼ ⟨ F + 2 + F × 2 ⟩ ⟨ ∣ h ~ ( f ) ∣ 2 ⟩
而根据稳相近似 :
⟨ h ~ ( f ) ⟩ = 1 2 f ˙ − 1 / 2 h o [ t ( f ) ] , h o [ t ( f ) ] ≡ ⟨ h + 2 ( t , n ^ ) + h × 2 ( t , n ^ ) ⟩ \langle \tilde{h}(f)\rangle = \frac{1}{2} \dot{f}^{-1/2} h_o[t(f)], ~~~ h_o[t(f)] \equiv \sqrt{\langle h^2_+(t, \hat{n}) + h^2_\times(t, \hat{n})\rangle}
⟨ h ~ ( f ) ⟩ = 2 1 f ˙ − 1 / 2 h o [ t ( f ) ] , h o [ t ( f ) ] ≡ ⟨ h + 2 ( t , n ^ ) + h × 2 ( t , n ^ ) ⟩
h + , h × h_+, h_\times h + , h × 为震荡信号,平方平均时会出一个1 / 2 1/2 1 / 2 ,因此h o [ t ( f ) ] h_o[t(f)] h o [ t ( f ) ] 大致就对应引力波信号角平均的振幅。最终,特征应变强度h c ( f ) h_c(f) h c ( f ) 表达式为:
h c ( f ) = 8 f 2 ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ = 2 f 2 f ˙ h o [ t ( f ) ] ∼ N c y c l e s h o [ t ( f ) ] h_c(f) = \sqrt{ 8f^2 \langle|\tilde{h}(f)|^2\rangle} = \sqrt{\frac{2f^2}{\dot{f}}}h_o[t(f)] \sim \sqrt{N_{\rm cycles}}~ h_o[t(f)]
h c ( f ) = 8 f 2 ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ = f ˙ 2 f 2 h o [ t ( f ) ] ∼ N c y c l e s h o [ t ( f ) ]
这里h ~ ( f ) \tilde{h}(f) h ~ ( f ) 为单个极化模式的(平均)振幅,如果定义为两种极化模式平方求和开根号⟨ ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ⟩ \small \sqrt{\langle|\tilde{h}_+(f)|^2 + |\tilde{h}_\times(f)|^2 \rangle} ⟨ ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ⟩ ,则特征应变h c 2 ( f ) = 4 f 2 ∣ h ~ ( f ) ∣ 2 h^2_c(f) = 4 f^2 |\tilde{h}(f)|^2 h c 2 ( f ) = 4 f 2 ∣ h ~ ( f ) ∣ 2 ,与 Moore et al. (2015) 中一致。注意,时域的h ( t ) h(t) h ( t ) 无量纲,对应频域h ~ ( f ) , S h ( f ) \tilde{h}(f), S_h(f) h ~ ( f ) , S h ( f ) 均为时间量纲,而这里定义的特征应力振幅h c ( f ) , h n ( f ) h_c(f), h_n(f) h c ( f ) , h n ( f ) 则重新变为无量纲。
常用谱总结对比
综合最常用的三种表述,对比如下:
功率谱密度:S h ( f ) S_h(f) S h ( f ) 单位频率 间隔内的引力波功率(默认为单边谱)S h ( f ) ∼ 4 π 2 T o b s ∣ h ~ ( f ) ∣ 2 , H ( f ) ∼ 2 T o b s ∣ h ~ ( f ) ∣ 2 S_h(f)\sim 4\pi\frac{2}{T_{\rm obs}}\left|\tilde{h}(f)\right|^2,~~~ H(f) \sim \frac{2}{T_{\rm obs}}\left|\tilde{h}(f)\right|^2
S h ( f ) ∼ 4 π T o b s 2 ∣ ∣ ∣ ∣ h ~ ( f ) ∣ ∣ ∣ ∣ 2 , H ( f ) ∼ T o b s 2 ∣ ∣ ∣ ∣ h ~ ( f ) ∣ ∣ ∣ ∣ 2
分数能量密度谱:Ω ( f ) \Omega(f) Ω ( f ) 单位对数频率 间隔内的引力波能量密度占比Ω ( f ) = 4 π 2 3 H 0 2 ∣ f ∣ 3 S h ( f ) \Omega(f) = \frac{4\pi^2}{3H_0^2} |f|^3 S_h(f)
Ω ( f ) = 3 H 0 2 4 π 2 ∣ f ∣ 3 S h ( f )
特征应变强度谱:h c ( f ) h_c(f) h c ( f ) 单位对数频率 间隔内的引力波(累积)等效振幅h c 2 ( f ) = 8 f 2 ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ v . s . h c 2 ( f ) = 2 f S h ( f ) h^2_c(f) = 8f^2 \langle|\tilde{h}(f)|^2\rangle ~~~{\rm v.s.}~~~ h^2_c(f) = 2 f S_h(f)
h c 2 ( f ) = 8 f 2 ⟨ ∣ h ~ ( f ) ∣ 2 ⟩ v . s . h c 2 ( f ) = 2 f S h ( f )
注意,Moore et al. (2015) 中有各种表述的综合对比,是相对常引用的一篇文章,但其内容有点混乱:其一,文中引力波探测信号s ( t ) = F + h + + F × h × s(t) = F_+h_+ + F_\times h_\times s ( t ) = F + h + + F × h × 与引力波信号混用,从而探测信号的功率谱与引力波功率谱混淆不明,转换时未考虑响应函数;其二,强行将h c h_c h c 的两种定义划等号,但如前所述,两种定义并不兼容,使用时需格外小心。
从匹配滤波灵敏度角度理解,并和信号的S h , Ω S_h, \Omega S h , Ω 只具有形式上的意义,物理上与随机背景相关的量并没有关系!但此时绘制在同一张灵敏度曲线图上的随机背景该如何表示?T f S h ( f ) \sqrt{T f S_h(f)} T f S h ( f ) ?
∫ 0 ∞ 4 f 2 ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) 2 f N ( f ) / ⟨ R ⟩ d ln f ≡ ∫ 0 ∞ [ h c ( f ) h n ( f ) ] 2 d ln f = ∫ 0 ∞ 4 f ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) 2 N ( f ) / ⟨ R ⟩ d ln f ≡ ∫ 0 ∞ S h ( f ) S n ( f ) d ln f = ∫ 0 ∞ 2 π 2 ∣ f ∣ 3 3 H 0 2 4 f ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) 2 π 2 ∣ f ∣ 3 3 H 0 2 2 N ( f ) / ⟨ R ⟩ d ln f ≡ ∫ 0 ∞ Ω h ( f ) Ω n ( f ) d ln f \begin{aligned}
&\int_0^\infty \frac{4 f^2(|\tilde{h}_+(f)|^2+|\tilde{h}_\times(f)|^2)}{2 f N(f)/\langle\mathcal{R}\rangle} d\ln f \equiv \int_0^\infty \left[\frac{h_c(f)}{h_n(f)}\right]^2 d\ln f\\
&= \int_0^\infty \frac{4 f(|\tilde{h}_+(f)|^2+|\tilde{h}_\times(f)|^2)}{2 N(f)/\langle\mathcal{R}\rangle} d\ln f \equiv \int_0^\infty \frac{S_h(f)}{S_n(f)} d\ln f\\
&= \int_0^\infty \frac{ \frac{2\pi^2|f|^3}{3H_0^2} ~ 4 f(|\tilde{h}_+(f)|^2+|\tilde{h}_\times(f)|^2)}{\frac{2\pi^2|f|^3}{3H_0^2} ~ 2 N(f)/\langle\mathcal{R}\rangle} d\ln f \equiv \int_0^\infty \frac{\Omega_h(f)}{\Omega_n(f)} d\ln f\\
\end{aligned} ∫ 0 ∞ 2 f N ( f ) / ⟨ R ⟩ 4 f 2 ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) d ln f ≡ ∫ 0 ∞ [ h n ( f ) h c ( f ) ] 2 d ln f = ∫ 0 ∞ 2 N ( f ) / ⟨ R ⟩ 4 f ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) d ln f ≡ ∫ 0 ∞ S n ( f ) S h ( f ) d ln f = ∫ 0 ∞ 3 H 0 2 2 π 2 ∣ f ∣ 3 2 N ( f ) / ⟨ R ⟩ 3 H 0 2 2 π 2 ∣ f ∣ 3 4 f ( ∣ h ~ + ( f ) ∣ 2 + ∣ h ~ × ( f ) ∣ 2 ) d ln f ≡ ∫ 0 ∞ Ω n ( f ) Ω h ( f ) d ln f