The evolution of two-point space correlation function of QSOs is analyzed in the framework of the theory of the large scale structure formation. For given cosmological models the agreement between theoretical predictions and observational data on correlation function of QSOs is achieved by determination of scales of structures R(z), in which could be formed quasars at different redshifts z. It is shown that quasars could be formed in the nucleus of massive galaxies or groups of galaxies owing to merges at their central region. It depends on an amplitude of the power spectrum of the initial matter density fluctuations at galaxy-cluster scales.

We examine the dependence of the spatial two-point correlation function of quasars $\xi_{qq}(r,z)$ at different redshifts on the initial power spectrum in flat cosmological models. Quasars and other elements of the large-scale structure of the universe are supposed to form in the peaks of the scalar Gaussian field of density fluctuations of appropriate scales. Quasars are considered as a manifestation of short-term active processes at the centers of these fluctuations. We propose a method for calculating the correlation function $\xi_{qq}(r,z)$ and show its amplitude and slope to depend on the shape of the initial power spectrum and the scale R of the fluctuations in which quasars are formed. We demonstrate that in the CDM models with the initial power spectrum slope $n=0.7\div1$ it is possible to explain, by choosing appropriate values of R, how the amplitudes and correlation radii of $\xi_{qq}(r,z)$ may either increase or decrease with increasing redshift z. In particular, the correlation radii of $\xi_{qq}(r,z)$ grow from $6-10h^{-1}$ Mpc when R grows from $0.45 h^{-1}$ to $1.3h^{-1}$ Mpc. The H+CDM model at realistic values of R fails to account for the observational data according to which the $\xi_{qq}(r,z)$ amplitude decreases with increasing $z$.

The dependence of the number density of the bright QSOs at different redshifts ($n_{QSO}(z)$) on initial power spectrum is studied. It is assumed that QSO phenomenon is an early short term stage of evolution of massive galaxies with $M\geq 2\times 10^{11}h^{-1}M_{\odot}$. The duration of such QSO stage which is passed through by fraction $\alpha$ of galaxies is determined by means of minimization of the divergence of the theoretical number density of QSOs at different redshifts for specified initial spectrum from observable one \cite{sc91}. It is shown that the nearest number densities of QSOs at $0.7\le z\le 3.5$ to observable ones are obtained for the tilted CDM model ($\Omega_{b}=0.1$, $n=0.7$). The QSO stage lasts $\sim 7\times 10^{7}/\alpha$ years and begins soon after the moment of rise of the first counterflow in collisionless component and shock wave in gas. The possibility of the reconstruction of initial power spectrum on small scale on the base of the observable data on number density of QSOs at different $z$ is considered too. Such reconstructed spectrum in comparison with standard CDM has steep reducing of power at $k\ge 0.5 h Mpc^{-1}$.