Theoretical description of weak lensing critical points for power spectra in the mildly nonlinear regime

In cosmic web analysis, complementary to traditional cosmological probes, the extrema (e.g., peaks and voids) two-point correlation functions (2PCFs) are of particular interest for the study of both astrophysical phenomena and cosmological structure formation. However, most previous studies constructed those statistics via N-body simulations without a robust theoretical derivation from first principles. A strong motivation exists for analytically describing the 2PCFs of these local extrema, taking into account the nonlinear gravitational evolution in the late Universe. In this paper, we derive analytical formulas for the power spectra and 2PCFs of two-dimensional (2D) critical points, including peaks (maxima), voids (minima), and saddle points, in mildly non-Gaussian weak gravitational lensing fields. We apply a perturbative bias expansion to model the clustering of 2D critical points. A generalized Gram-Charlier A series expansion is used to describe the probability density functional of the cosmic density field. We successfully derive the power spectrum of weak lensing critical points up to the next-to-next-to-leading order in gravitational perturbation theory, where trispectrum configurations of the weak lensing field have to be included. We numerically evaluate those power spectra up to the next-to-leading order, which corresponds to the inclusion of bispectrum configurations, and transform them to the corresponding 2PCFs. An exact Monte Carlo integration is performed assuming a Gaussian distributed density field to validate our theoretical predictions. Overall, we find similar properties in two dimensions compared to the clustering of three-dimensional critical points previously measured from N-body simulations. Contrary to standard lensing power spectra analysis, we find distinct baryon acoustic oscillation features in the lensing peak 2PCFs due to the gradient and curvature constraints, and we quantify that non-Gaussianity makes for <a:math xmlns:a="http://www.w3.org/1998/Math/MathML" display="inline"><a:mo>∼</a:mo><a:mn>10</a:mn><a:mo>%</a:mo></a:math> of the signal at quasilinear scales, which could be important for current stage-IV surveys.