Placeholder text. This section is laid out to show typography, math and citations; the prose is written in the content phase.
The two-point function of the primordial curvature perturbation $\zeta$ fixes the amplitude and tilt of the fluctuations but says little about the interactions that produced them. Those interactions live in the higher-point functions. The leading one is the bispectrum,
\[\langle \zeta_{\mathbf k_1}\zeta_{\mathbf k_2}\zeta_{\mathbf k_3}\rangle = (2\pi)^3\,\delta_D(\mathbf k_1+\mathbf k_2+\mathbf k_3)\,B_\zeta(k_1,k_2,k_3),\]whose dependence on the triangle formed by the three momenta is the shape. The figure at the top of the page shows the dimensionless shape function $S \propto (k_1k_2k_3)^2 B_\zeta$ for the standard templates.
We use the local ansatz $\Phi = \phi + f_{\rm NL}^{\rm local}\left(\phi^2 - \langle\phi^2\rangle\right)$ for Bardeen’s potential in the matter era, and quote $f_{\rm NL}$ in the CMB normalisation [@komatsu2001]. Single-field slow-roll models satisfy the squeezed-limit consistency relation [@maldacena2003].
The atlas below crosses four kinds of work (theory, simulation, method, observation) with the place the signal is generated or observed. Each cell is a section with its own revision date; the change log records every revision and every paper added.
The most recent broad reviews of the field date from 2010–2015 [@chen2010; @desjacques2010; @renauxpetel2015], with community white papers since.