21 The POLARBEAR analysis pipelines
POLARBEAR, Mapmaking, Pseudo-spectra and MASTER, Simulations
21 The POLARBEAR analysis pipelines
Because of this null-test setup and the unblinding criteria, it means that we need to be able to run the pipeline for all these null-tests combinations and to rerun it repeatedly until we can pass the null-tests. We need to figure out a way to fail fast, speculate often on how to mitigate the systematics, and repeat.
21.1 Naïve Mapmaking
That is why the naïve mapmaker1 is the mapmaker of choice. Since it is just binning the data, per TOD we can perform binning all at once with different null-splits. This makes the computational cost of having more null-splits almost free. (In fact, in one paper with the small-patch dataset,2 both the naïve mapmaker and an unbiased mapmaker are used and they demonstrated they can only manage the full null-tests suite on the naïve mapmaker pipeline because of computational constraints, and the results from both pipelines are consistent.)
Per Constant Elevation Scan (CES), which is about an hour of TOD observation3, mapmaking can be performed independently.
Mathematically, we can make it abstract and write
\[\begin{equation} m^{i} = \mathcal{M} \mathopen{} \left \{ d^{i} \right \} \mathclose{} , i = 1 , \dots , n_{\text{CES}} , \label{eq:mapmaking_illustration} \tag{21.1}\end{equation} \]
where \(d\) represents the TOD, \(m\) represents the map, \(i\) represents the \(i\)-th CES, and \(n_\text{CES} = 3391\) represents the total number of CES. \(\mathcal{M}\) represents the mapmaking procedure that turns TOD into maps. While this procedure involves a lot of operations including filtering and TOD processing, the key feature here is that it is a mathematically well-defined operator.
Note that here many indices are suppressed, explicitly, we have \(m^{i , X , n}_{p}\), where \(X\) represents the different \(I, Q, U\) components, \(n\) represents different null-splits, such as \(1A\) for the first half of the 1st null-split, \(3B\) for the second half of the 3rd null-split, etc. \(p\) represents the pixel index of the map, which is inherently a \(2\)-dimensional object. As we take the flat-sky approximation here, we can understand \(p\) as enumerating over \((1, 1), (1, 2), \ldots, (2, 1), (2, 2), \ldots\).
Finally, note that \(m\) here is a random variable. Mathematically,
\[m^{i , X , n}_{p} \sim \mathcal{N} \mathopen{} \left ( \mu^{i , X , n}_{p}, 1 / w^{i , X , n}_{p} \right ) \mathclose{} .\]
Uncorrelated Gaussian white noise in the TOD is assumed in order for this to be strictly true.4 Here \(μ\) is what people usually called the map, where \(w\) is statistically called precision—the inverse of the variance, and is usually called the weights among physicists.
21.2 Coadd
After that, another pipeline called coadd simply adds individual CES maps together, weighted by inverse-variance.
Mathematically, first we should mention the variables that we propagate in the mapmaking and coadd pipelines to be \(M , w , \text{ where } M \equiv w \mu\).
And then we write down the inverse-variance weighting scheme,
\[\begin{align} w^{I} & = \sum_{i \in I} w^{i} \label{eq:coaddByInvVar1} \tag{21.2}\\ \mu^{I} & = \frac{\sum_{i \in I} w^{i} \mu^{i}}{\sum_{i \in I} w^{i}} \nonumber \\ \Rightarrow M^{I} & = \sum_{i \in I} M^{i} \label{eq:coaddByInvVar2} \tag{21.3}\end{align}\]
Here the \(I\) index represents a map-bundle—a collection of CES maps that together form a complete map of the patch we observed. Each map bundle amounts to about 10 days’ worth of observation, and we have 38 map-bundles in total.
This explains the verb “coadd” and why we keep track of the combination of \(M, w\), where \(μ = \frac{M}{w}\) is the quantity of interest. It also explains why naïve mapmaking hugely simplifies the pipelines computationally. To put it in programming terms, the naïve mapmaker can be implemented using the MapReduce paradigm where the mapmaking-pipeline maps each TOD per CES into a map, and the coadd-pipeline reduces these maps into map bundles.
Note again that many indices are suppressed and can be expanded to \(m^{I , X , n}_{p}\). Note that per map bundle & null-split, this is embarrassingly parallel.
21.3 Cross-pseudo-power-spectra
The cross-pseudo-power-spectra, often called pseudo-spectra for brevity, between the null-split map bundles will be computed by the pseudo-spectra pipeline.5
Schematically,
\[\tilde{C}^{X^{\prime} Y^{\prime} , n}_{b} = \mathcal{P} \mathopen{} \left \{ m^{I , X , n}_{p}, m^{J , Y , n}_{p} \right \} \mathclose{} ,\]
where \(\mathcal{P}\) is a mathematically well-defined operator6 that turns map-bundles into cross-pseudo-power-spectra. Note that \(X' = T, E, B\) while \(X = I, Q, U\) where \(T \equiv I\). For the purpose here, we would note only that \(\mathcal{P}\) involves Fourier transform, a procedure to “rotate” \(Q, U\) components into \(E, B\), as well as a procedure to obtain a pure B-mode estimator that suppresses E-to-B leakage.7 Computationally, we note that per null-split, the spectra can be computed independently.
21.4 Monte Carlo (MC) simulations
And the whole process will be repeated, once with the real data, and \(n_\text{sim}\) times each with a random simulation8.
Since all these are embarrassingly parallel, it should be any HPC (High-Performance Computing) programmer’s dream.
And because of the simplicity of the pipeline, it is a policy in POLARBEAR to implement all these in pure Python.
21.5 Splitting the pipeline into two
All these worked fabulously in small-patch. But when processing the large-patch dataset, it failed spectacularly. Simply put, the software had not been expecting nor tested against the larger map size9.
In the early days of large-patch analysis, it was decided to make maps with much larger pixels at \({10'}\) which is good enough for \(\ell \lesssim 600\). This decision split the pipeline into two, the low-\(\ell\) pipeline for \(50 \leq \ell \leq 600\) targeting primordial B-mode, and the high-\(\ell\) pipeline for \(600 \leq \ell \leq 3000\), targeting lensing B-mode, E-mode, and the lensing potential.
In addition to “solving” the computational problem in the interim, splitting such a large \(\ell\)-range into two enables each pipeline to have their own optimization and mitigation strategies, such as filtering, data-selection, etc.
Footnotes
The POLARBEAR Collaboration et al., “A Measurement of the Cosmic Microwave Background B-Mode Polarization Power Spectrum at Subdegree Scales from Two Years of POLARBEAR Data”.↩︎
Constant Elevation Scan refers to the fact that those hour-long observations are scanning the sky at constant elevation.↩︎
In the limit that this assumption is exact, naïve mapmaker is the optimal mapmaker (See chapter 11.) This implies that we need filtering to turn \(1/f\)-noise into approximately white noise, and deviation from this assumption results in less-than-optimal but usable maps, where usability is subject to other criteria including null-test passing and de-biasing.↩︎
This is called pseudo because it is so far biased and will be corrected for in later steps.↩︎
See The POLARBEAR Collaboration et al., “A Measurement of the Cosmic Microwave Background B-Mode Polarization Power Spectrum at Sub-Degree Scales with POLARBEAR”, eq. 16.↩︎
Smith, “Pseudo- C ℓ Estimators Which Do Not Mix E and B Modes”.↩︎
Large-patch is about 2 orders of magnitude larger than small-patch. When maps are made at the same resolution of \({2'}\), small-patch would be hundreds of pixels wide, and large-patch would be thousands of pixels wide.↩︎