Abstract
We test degree-scale cosmic microwave background (CMB) anisotropy for Gaussianity by studying the QMASK map, obtained from combining the QMAP and Saskatoon data. We compute seven morphological functions $M_i(\Delta T)$, including six Minkowski functionals and the number of regions $N_c$ at a hundred $\Delta T$ levels. We also introduce a new parameterization of the morphological functions $M_i(A)$ in terms of the total area $A$ of the excursion set. We show that the latter considerably decorrelates the morphological statistics. We compare these results with those from 1,000 Gaussian Monte Carlo maps with the same power spectrum and conclude that the QMASK map is neither a very typical nor a very exceptional realization of a Gaussian field. Roughly 20% of the 1,000 Gaussian Monte Carlo maps differ more than the QMASK map from the mean morphological parameters of the Gaussian fields.
Introduction
In the standard cosmological paradigm, the theory of inflation predicts that the initial temperature fluctuations in the cosmic microwave background (CMB) should follow a Gaussian distribution. Detecting "non-Gaussianity" would be a significant discovery, potentially pointing toward exotic physics such as cosmic strings or topological defects. However, testing for Gaussianity is statistically challenging. Many traditional tests are highly correlated, meaning that if a map appears anomalous in one metric, it likely appears anomalous in others, without providing independent confirmation.
This study evaluates the Gaussianity of the QMASK map—a combined dataset from the QMAP and Saskatoon experiments—using a robust morphological framework designed to minimize these correlations and reduce sensitivity to experimental noise.
Methodology: Minkowski Functionals
To characterize the geometry of the CMB fluctuations, we employ Minkowski functionals (MFs). For a two-dimensional map, these functionals describe the morphology of "excursion sets"—regions where the temperature exceeds a specific threshold. The primary functionals used are:
- Area: The total surface area above the threshold.
- Perimeter: The total length of the boundary contours.
- Genus: A topological measure representing the number of isolated "islands" minus the number of "holes."
To improve numerical accuracy, we utilize linear interpolation to identify sub-pixel contour points. This avoids the common pitfall of simply counting pixel edges, which can overestimate perimeters by a factor of $4/\pi$.
The Innovation of Area Parameterization
A central contribution of this work is the shift from temperature parameterization ($M_i(\Delta T)$) to area parameterization ($M_i(A)$). Traditionally, morphological statistics are plotted against temperature thresholds. However, we demonstrate that plotting these functionals against the total area ($A$) of the excursion set offers two distinct advantages:
- Decorrelation: Under standard temperature parameterization, 16 out of 21 pairs of statistics are almost perfectly correlated (coefficients > 0.9). Switching to area parameterization shatters this degeneracy, making the tests statistically independent.
- Robustness: By using cumulative probability (Area) as the independent variable, the analysis becomes less sensitive to the "tails" of the distribution—the extreme temperature values most susceptible to experimental noise.
Data and Comparative Analysis
The QMASK map covers approximately 648 square degrees near the North Celestial Pole with a resolution of 0.68°. To establish a baseline for Gaussian behavior, we generated 1,000 Gaussian Monte Carlo simulations. These "mock maps" were meticulously constructed to replicate the sky coverage, noise properties, and power spectrum of the actual QMASK data.
We also examined the "percolating region," defined as the single largest connected patch on the map. The Minkowski functionals of this specific region serve as sensitive indicators of phase transitions within the field's geometry.
Results and Discussion
The analysis confirms that the QMASK map is consistent with Gaussianity. When ranked against the 1,000 simulations, the QMASK data was neither perfectly average nor an extreme outlier.
Key findings include:
- Approximately 20% of the random Gaussian simulations exhibited a greater deviation from the mean than the actual QMASK data.
- The most statistically noteworthy feature was the number of isolated regions ($N_c$) when parameterized by area, which fell into the 82nd percentile. While high, this remains well within the expected variance of a Gaussian realization.
Conclusion
The QMASK map represents a typical, if slightly unusual, realization of a Gaussian field. The area parameterization technique introduced here provides a computationally efficient ($\propto N \ln N$) and robust method for testing Gaussianity. While the current map size (648 square degrees) limits the signal-to-noise ratio, this methodology is ideally suited for larger datasets from missions like the Microwave Anisotropy Probe (MAP). Future applications include setting quantitative limits on foreground contamination and searching for topological defects in high-resolution CMB surveys.
Summary updated