Ryan Boyden edited section_Stastical_Comparisons_label_comparisons__.tex  over 8 years ago

Commit id: b581632b8921088b0532a63987d36cae7bf2e4fd

deletions | additions      

       

We follow the analysis of Koch et al. (2015) to compute PDFs of kurtosis and skewness. For all positions in an integrated intensity map, we compute each moment within a small, confined circular region. We, similar to Koch et al. (2015), choose a circular radius of 5 pixels. The kurtosis and skewness PDFs are histograms of these respective moment values, and are shown in figure ??? (necessary have this elaboration? Diffucult to word without directly plagiarizing).   For both runs, the most probable range of kurtosis values are close to zero. However, run W1T2t0.2 has higher probabilities at positive values, and is 15 percent less likely to have a kurtosis of zero than that of run W2T2t0. The Skewness PDFs indicate a similar trend, in which run W1T2T0.2 is 10 percent less likely to have a skewness of zero. On small scales, the integrated intensity distributions of our two  runs behave similar to a Gaussian, though run W1T2t0.2 has a greater tendency to diverge from this behavior than run W2T2t0.2 does. \subsections{Principal Component Analysis}  (The Principal Component Analysis (PCA) is one of the most sensitive statistics in our fiducial comparison.) We present our runs' velocity channel covariance matrices in Figure ???. After launching winds, the covariances for velocity channels between -2 and 2 km/s (the channels of strongest intensity???) increase in a particular behavior: features at the center of run W2T2t0's covariance matrix are augmented and further separated (as shown in run W1T2t0.2' covariance matrix). Covariances outlining the features also increase.   We argue that the observed difference implies a change in gas temperature distribution (may need eigenvalues to first suggest change in principal components, so this isn't a jump, assuming my argument).  \subsection{Fourier Statistics}