I was reading Sikivie's paper on Axion Cosmology: https://arxiv.org/abs/astro-ph/0610440 . It says the equation of motion for the axion field $a(x)$ in Friedmann–Lemaître–Robertson–Walker space-time is

$$D_\mu \partial^\mu a(x) + V_a′(a(x)) = (\partial_t^2 + 3\frac{\dot{R}}{R}\partial_t - \frac{1}{R^2}\nabla_x^2)a(x) + V_a′(a(x)) = 0$$ where $R(t)$ is the cosmological scale factor and $V$ is the potential.

Can someone tell me what $D_\mu$ is and how one might derive the right hand side of the above equation?


2 Answers 2


$D_\mu$ is just the covariant derivative acting on the vector $\partial^\mu a$. The middle expression is just expanding this derivative out in terms of the Christoffel symbols for the FLR metric. The actual eq of motion is the last equals sign.

  • $\begingroup$ I tried expanding out $D_\mu$ using the -+++ sign convention for the FLRW metric, i.e., $g_{00} = -1$ and $g_{ij} = \delta_{ij} R^2(t)$. But I got $D_\mu\partial^\mu a = (-\partial_t^2 + 3\frac{\dot{R}}{R}\partial_t + \frac{1}{R^2}\nabla_x^2)a$. Am I using the wrong sign convention or did I do something totally wrong? $\endgroup$
    – Matrix23
    Commented Oct 26, 2021 at 14:18
  • $\begingroup$ That I do not know, but it looks like a different metric convention since it's only signs that differ. $\endgroup$
    – mike stone
    Commented Oct 26, 2021 at 17:31

When applied to scalar functions, $Δ = D_μ ∂^μ$ is the Hodge-deRham operator associated with the metric $g$. Since $$\begin{align} Δa & = D_μ ∂^μ a = ∂_μ ∂^μ a + Γ_{νμ}^μ ∂^ν a = ∂_μ ∂^μ a + Γ_{μν}^ν ∂^μ a \\ Γ_{μν}^ν & = g^{νρ} \frac{1}{2} \left(∂_μg_{νρ} + ∂_νg_{ρμ} - ∂_ρg_{μν}\right) = \frac{1}{2} g^{νρ} ∂_μg_{νρ} = \frac{1}{2} \frac{∂_μg}{g} = \frac{∂_μ\sqrt{|g|}}{\sqrt{|g|}} \end{align}$$ then you may also write this operator as $$Δa = ∂_μ ∂^μ a + \frac{∂_μ\sqrt{|g|}}{\sqrt{|g|}} ∂^μ a = \frac{∂_μ\left(\sqrt{|g|}∂^μ a\right)}{\sqrt{|g|}} = \frac{1}{\sqrt{|g|}} ∂_μ\left(\sqrt{|g|}g^{μν}∂_νa\right).$$

If $$g_{00} = -1, \hspace 1em g_{i0} = 0 = g_{0j}, \hspace 1em g_{ij} = R^2 δ_{ij} \hspace 1em (i, j = 1, 2, 3),$$ where $R = R(t)$ is a function only of $t$, with $R(t) > 0$, and the coordinates are $\left(x^0, x^1, x^2, x^3\right) = \left(t, x, y, z\right)$, then $g = -R^6$, $\sqrt{|g|} = R^3$ and $$\sqrt{|g|}g^{00} = -R^3, \hspace 1em \sqrt{|g|}g^{i0} = 0 = \sqrt{|g|}g^{0j}, \hspace 1em \sqrt{|g|}g^{ij} = R δ^{ij} \hspace 1em (i, j = 1, 2, 3).$$ Therefore $$Δa = \frac{1}{R^3} \left(\frac{∂}{∂t}\left(-R^3\frac{∂a}{∂t}\right) + ∇·\left(R∇a\right)\right)$$ where $∇ = (∂/∂x, ∂/∂y, ∂/∂z)$. Thus, $$ Δa = -\frac{∂^2a}{∂t^2} - \frac{3}{R} \frac{∂R}{∂t}\frac{∂a}{∂t} + \frac{1}{R^2} ∇^2a. $$

If you rescale $g_{μν} → k g_{μν}$ by a constant $k$, then $\sqrt{|g|} → k^2 \sqrt{|g|}$ and $\sqrt{|g|} g^{μν} → k \sqrt{|g|} g^{μν}$. Therefore, $Δ → k^{-1} Δ$. The operator is scale-dependent ... which also means: sign-dependent. Choosing $k = -1$ resets the metric as $$g_{00} = 1, \hspace 1em g_{i0} = 0 = g_{0j}, \hspace 1em g_{ij} = -R^2 δ_{ij} \hspace 1em (i, j = 1, 2, 3),$$ and the Hodge-deRham operator as $$ Δa = \frac{∂^2a}{∂t^2} + \frac{3}{R} \frac{∂R}{∂t}\frac{∂a}{∂t} - \frac{1}{R^2} ∇^2a. $$ So, the authors are treating the metric as a proper-time metric, rather than a proper-length metric.

As a (propaganda) footnote, if you're writing the metric as a proper time metric then you could just as well treat the $R$ factor as being the actual $R$, not $R$ modulo any $c = 1$ assumption. Then, it is normalized as $R(\text{now}) = 1/c$, where $c = 299792458$ meters/second. That shows that this is secretly just an ordinary Minkowski metric ... but with a variable light speed $c(t) = 1/R(t)$ and line element $$ds^2 = dt^2 - \frac{dx^2 + dy^2 + dz^2}{c(t)^2}$$ where $s$ now denotes proper time; and that - with this understanding - the Hodge-deRham operator can just as well be written as: $$Δa = \frac{∂^2a}{∂t^2} - \frac{3}{c} \frac{∂c}{∂t}\frac{∂a}{∂t} - c^2 ∇^2a.$$ If you carefully apply this to the field $F_{μν}$, which is given in terms of the Maxwell vectors by $𝐄 = \left(F_{10}, F_{20}, F_{30}\right)$, and $𝐁 = \left(F_{23}, F_{31}, F_{12}\right)$, and to the response field, which is given by the tensor density $$𝔊^{μν} = ε_0 c \sqrt{|g|} g^{μρ} g^{νσ} F_{ρσ} + \frac{1}{2} θ ε^{μνρσ} F_{ρσ}$$ (where $ε_0$ denotes the permittivity of the vacuum) and in terms of the Maxwell letters by $𝐃 = \left(𝔊^{01}, 𝔊^{02}, 𝔊^{03}\right)$ and $𝐇 = \left(𝔊^{23}, 𝔊^{31}, 𝔊^{12}\right)$, then you'll get the following constitutive relations $$𝐃 = ε 𝐄 + θ 𝐁, \hspace 1em 𝐇 = \frac{𝐁}{μ} - θ 𝐄$$ with $$ε = ε_0 c R = ε_0 \frac{c}{c(t)}, \hspace 1em μ = μ_0 c R = μ_0 \frac{c}{c(t)}$$ where $μ_0 = 1/(ε_0 c^2)$ denotes the permeability of the vacuum. The axion $a$ is just $-θ$ up to an author-dependent positive constant factor and is just the chiral counterpart to permittivity, promoted to a field. The permittivity and permeability, themselves, are effectively fields too because of their metric-dependency. They embody the optics of warped space-time.

These fields satisfy the usual Maxwell Equations $∇·𝐃 = ρ$ for a charge density $ρ$ and $∇×𝐇 - ∂𝐃/∂t = 𝐉$ for a current density $𝐉$ and the corresponding electromagnetic wave speed is $$\sqrt{\frac{1}{εμ}} = \sqrt{\frac{c(t)^2}{c^2} \frac{1}{ε_0μ_0}} = c(t).$$ So $c(t)$ really, literally is the speed of light we're talking about there.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.