Determining cosmic microwave background structure from its peak distribution

We present a new method for time-efficient and accurate extraction of the power spectrum from future cosmic microwave background (CMB) maps based on properties of peaks and troughs of the Gaussian CMB sky. We construct a statistic describing their angular clustering - analogously to galaxies, the 2-point angular correlation function, $\xi_\nu(\theta)$. We show that for increasing peak threshold, $\nu$, the $\xi_\nu(\theta)$ is strongly amplified and becomes measurable for $\nu\geq$1 on angular scales $\leq 10^\circ$. Its amplitude at every scale depends uniquely on the CMB temperature correlation function, $C(\theta)$, and thus the measured $\xi_\nu$ can be uniquely inverted to obtain $C(\theta)$ and its Legendre transform, the power spectrum of the CMB field. Because in this method the CMB power spectrum is deduced from high peaks/troughs of the CMB field, the procedure takes only $[f(\nu)]^2N^2$ operations where $f(\nu)$ is the fraction of pixels with $|\delta T|\geq\nu$ standard deviations in the map of $N$ pixels and is e.g. 0.045 and 0.01 for $\nu$=2 and 2.5 respectively. We develop theoretical formalism for the method and show with detailed simulations, using MAP mission parameters, that this method allows to determine very accurately the CMB power spectrum from the upcoming CMB maps in only $\sim(10^{-4}-10^{-3})\times N^2$ operations.