Why does the Zel'dovich approximation (ZA) work well to describe gravitational collapse in the universe? This problem is examined by focusing on its dependence on the dimensionality of the collapse. The ZA is known to be exact for a one-dimensional collapse. We show that the ZA becomes progressively more accurate in the order of three-, two-, and one-dimensional collapses. Furthermore, using models for spheroidal collapse, we show that the ZA remains accurate in all collapses, which become progressively lower dimensional with the passage of time. That is, the ZA is accurate because the essence of the gravitational collapse is incorporated in the ZA.

According to N-body numerical simulations, the radial pairwise velocities of galaxies have negative skewness in the quasi-nonlinear regime. To understand its origin, we calculate the probability distribution function of the radial pairwise velocity using the Zel'dovich approximation, i.e., an analytical approximation for gravitational clustering. The calculated probability distribution function is in good agreement with the result of N-body simulations. Thus the negative skewness originates in relative motions of galaxies in the clustering process that the infall dominates over the expansion.

Among several analytic approximations for the growth of density fluctuations in the expanding Universe, Zel'dovich approximation in Lagrangian coordinate scheme is known to be unusually accurate even in mildly non-linear regime. This approximation is very similar to the Pad\'e approximation in appearance. We first establish, however, that these two are actually different and independent approximations with each other by using a model of spheroidal mass collapse. Then we propose Pad\'e-prescribed Zel'dovich-type approximations and demonstrate, within this model, that they are much accurate than any other known nonlinear approximations.

Among various analytic approximations for the growth of density fluctuations in the expanding Universe, Zel'dovich approximation and its extensions in Lagrangian scheme are known to be accurate even in mildly non-linear regime. The aim of this paper is to investigate the reason why these Zel'dovich-type approximations work accurately beyond the linear regime from the following two points of view: (1) Dimensionality of the system and (2) the Lagrangian scheme on which the Zel'dovich approximation is grounded. In order to examine the dimensionality, we introduce a model with spheroidal mass distribution. In order to examine the Lagrangian scheme, we introduce the Pad\'e approximation in Eulerian scheme. We clarify which of these aspects supports the unusual accuracy of the Zel'dovich-type approximations. We also give an implication for more accurate approximation method beyond the Zel'dovich-type approximations.