The Mathematics of CMB flat sky approximation
Flat-sky approximation, Curved sky, Spin-weighted fields, CMB polarization, Fourier analysis, Mathematical conventions
The Mathematics of CMB flat sky approximation
Flat-sky approximation
The following are derived from,1 including its sign conventions.
Also, we follow the physicists’ convention that the same symbol with a different argument actually indicates a different function after a transform. For example, \(f \mathopen{} \left ( x \right ) \mathclose{} , f \mathopen{} \left ( k \right ) \mathclose{}\) .
Definitions
\[\begin{align*} \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} & \equiv Q \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} + \mathrm{i} U \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \\ \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \equiv E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{align*}\]
Note that \(Q , U\) are real fields, \(\mathcal{Q} , \mathcal{E} , E , B\) are complex fields.
From curved sky to flat sky
We replace
\[T \mathopen{} \left ( \mathbf{\hat{n}} \right ) \mathclose{} = \sum_{\ell , m} a_{\ell m} Y_{\ell m} \mathopen{} \left ( \mathbf{\hat{n}} \right ) \mathclose{} \rightarrow \int T \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell}\]
that is, using the angular frequency, non-unitary convention of Fourier transform,
\[\begin{cases} T \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} &\equiv \int T \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \mathrm{e}^{- \mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \symbf{\theta} \\ \mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} &\equiv \int \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \mathrm{e}^{- \mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \symbf{\theta} \end{cases}\]
\[\begin{equation} \begin{cases} T \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &= \frac{1}{\left( 2 \pi \right)^{2}} \int T \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \\ \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &= \frac{1}{\left( 2 \pi \right)^{2}} \int \mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \label{TQDef} \end{cases} \tag{38.1}\end{equation}\]
Note that \(\mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{}\) is defined above, without giving a Physical meaning.
\(\mathcal{Q}\) as spin-\(2\) field
As \(\mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{}\) is a spin-\(2\) field (vice versa for \(\bar{\mathcal{Q}} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{}\) ),
\[\begin{align} \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &= \frac{1}{\left( 2 \pi \right)^{2}} \int \left[ E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \right] \frac{-1}{\ell^{2}} \eth^{2} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \tag{38.2}\\ \bar{\mathcal{Q}} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &= \frac{1}{\left( 2 \pi \right)^{2}} \int \left[ E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} - \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \right] \frac{-1}{\ell^{2}} \bar{\eth}^{2} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \tag{38.3}\end{align}\]
In the small scale limit,
\[\begin{align*} \frac{-1}{\ell^{2}} \eth^{2} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} & \approx \mathrm{e}^{-2 \mathrm{i} \left( \phi - \phi_{\ell} \right)} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \\ \frac{-1}{\ell^{2}} \bar{\eth}^{2} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} & \approx \mathrm{e}^{2 \mathrm{i} \left( \phi - \phi_{\ell} \right)} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \end{align*}\]
That is,
\[\begin{align} \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &\approx \frac{1}{\left( 2 \pi \right)^{2}} \int \left[ E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \right] \mathrm{e}^{-2 \mathrm{i} \left( \phi - \phi_{\ell} \right)} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \label{EBDef} \tag{38.4}\\ \bar{\mathcal{Q}} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} &\approx \frac{1}{\left( 2 \pi \right)^{2}} \int \left[ E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} - \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \right] \mathrm{e}^{2 \mathrm{i} \left( \phi - \phi_{\ell} \right)} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \label{EBDefConj} \tag{38.5}\end{align}\]
\[\mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \approx \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \mathrm{e}^{-2 \mathrm{i} \left( \phi - \phi_{\ell} \right)}\]
Further approximation
Where \(\phi_{\ell}\) is the complex phase of \(\ell\) (when we interpret \(\ell \in \mathbb{R}^{2} \cong \mathbb{C}\).)
We can further set \(\phi = 0\) by rotating from \(\left( \symbf{\hat{\theta}} , \symbf{\hat{\phi}} \right)\) to \(\left( \mathbf{\hat{x}} , \mathbf{\hat{y}} \right)\) on the \(\mathbf{\hat{z}}\)-tangent plane. Hence,
\[\begin{equation} \begin{cases} \mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx \mathcal{E} \mathopen{} \left ( \ell \right ) \mathclose{} \mathrm{e}^{2 \mathrm{i} \phi_{\ell}} \\ \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx \mathcal{Q} \mathopen{} \left ( \ell \right ) \mathclose{} \mathrm{e}^{-2 \mathrm{i} \phi_{\ell}} \end{cases} \label{eq:QErelation} \tag{38.6}\end{equation}\]
Decomposing \(\mathcal{E}\) in terms of \(E , B\)
Starting from (38.5), take the c.c., and \(\mathbf{\ell} \mapsto - \mathbf{\ell}\), we have
\[\begin{equation} \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \approx \frac{1}{\left( 2 \pi \right)^{2}} \int \left[ \bar{E} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} \bar{B} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} \right] \mathrm{e}^{-2 \mathrm{i} \left( \phi - \phi_{\ell} \right)} \mathrm{e}^{\mathrm{i} \mathbf{\ell} \cdot \symbf{\theta}} \,\mathrm{d}^{2} \mathbf{\ell} \label{EBDefAlt} \tag{38.7}\end{equation}\]
\[\begin{cases} \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \equiv E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \approx \bar{E} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} + \mathrm{i} \bar{B} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} \\ \mathcal{E} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} & \approx E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} - \mathrm{i} B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{cases}\]
That is,
\[\begin{equation} \begin{cases} E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx \frac{1}{2} \left[ \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} + \bar{\mathcal{E}} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} \right] \\ B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx \frac{1}{2 \mathrm{i}} \left[ \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} - \bar{\mathcal{E}} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} \right] \end{cases} \label{eq:realEB} \tag{38.8}\end{equation}\]
Note that \(E, B\) have the following property accordingly,
\[\begin{align*} \bar{E} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} &= E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \\ \bar{B} \mathopen{} \left ( - \mathbf{\ell} \right ) \mathclose{} &= B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{align*}\]
That is, the field corresponding to \(E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} , B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{}\) , namely, \(E \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} , B \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{}\) , are real, as they fulfill the reality condition.
\[\begin{align*} \mathbb{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \equiv \begin{bmatrix} E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \\ B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{bmatrix} \\ \mathbb{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \equiv \begin{bmatrix} Q \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \\ U \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{bmatrix} \\ R \mathopen{} \left ( \phi \right ) \mathclose{} & \equiv \begin{bmatrix} \cos \phi & - \sin \phi \\ \sin \phi & \cos \phi \end{bmatrix} \\ \mathbb{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx R \mathopen{} \left ( -2 \phi_{\ell} \right ) \mathclose{} \mathbb{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \\ \mathbb{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} & \approx R \mathopen{} \left ( 2 \phi_{\ell} \right ) \mathclose{} \mathbb{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \end{align*}\]
which can be compared to (38.6).
Conclusion
That is, we can go between
\[\left( Q \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} , U \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \right) \leftrightarrow \mathcal{Q} \mathopen{} \left ( \symbf{\theta} \right ) \mathclose{} \leftrightarrow \mathcal{Q} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \leftrightarrow \mathcal{E} \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \leftrightarrow \left( E \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} , B \mathopen{} \left ( \mathbf{\ell} \right ) \mathclose{} \right)\]
Footnotes
Zaldarriaga and Seljak, “All-Sky Analysis of Polarization in the Microwave Background,” sec. V.↩︎
using the reality condition on real fields. See proof in the companion file
flat-sky-approximation.nb↩︎