Recently a procedure by generalized density matrix (GDM) is proposed [1] for calculating a collective/bosonic Hamiltonian microscopically from the shell-model Hamiltonian. In this work we examine the validity of the method by comparing the GDM results with that of the exact shellmodel diagonalization in a number of models. It is shown that the GDM method reproduces the low-lying collective states quite well, both for energies and transition rates, across the whole region going from vibrational to γ-unstable and deformed nuclei. It is a long-standing problem in nuclear physics to understand how macroscopic collective motion arises from microscopic single-particle motion. The shell-model (configuration interaction) successfully reproduces various collective behaviors by diagonalizing the nucleon Hamiltonian in a huge Slater-determinant basis. However, the dimension of the basis makes it impractical except for the cases with only a few valence nucleons. On the other hand, phenomenological bosonic approaches are often successful in fitting the experimental data (first of all the geometric Bohr Hamiltonian [2,3] and the interacting boson model [4]). This shows that, out of the huge Slater-determinant space, there exists a few degrees of freedom with a bosonic nature, which are usually enough in describing the collective states. Serious efforts were devoted to deriving those parameters of the bosonic Hamiltonian from the underlying shell-model Hamiltonian. However, the complete theory is still missing.Recently we proposed [1] a procedure based on the generalized density matrix (GDM) that was originally formulated in Refs. [5][6][7][8]. This procedure is rather simple, clean, and consistent. In compact form, there are only two equations, (14) and (23) in Ref. [1]. Results from the lowest orders give the well-known Hatree-Fock (HF) equations and random phase approximation (RPA). Higher orders fix the anharmonic terms in the collective/bosonic Hamiltonian. The aim of this work is to demonstrate the validity of the GDM method, by comparing its results with that of the exact shell-model diagonalization.In this work for simplicity we restrict ourselves to systems without rotational symmetry. The GDM formulation with angular-momentum vector coupling has been considered in Ref [9]. The single particle (s.p.) space in this work is drawn schematically in Fig. 1. There are two degenerate s.p. levels with energies e = ±1/2. The Fermi surface is in between, thus the lower levels are completely * Electronic address: jial@nscl.msu.edu filled and upper levels are empty. Each s.p. level has a quantum number m that is a half integer. Degenerate time-reversal pair has m with different sign, m1 = −m 1 . For fermions, |1 = −|1 , and we choose the phases such thatWe assume a two-body Hamiltonian,where f 12 = δ 12 e 1 , e 1 are the HF s.p. energies shown in Fig. 1. The density matrix ρ 12 = δ 12 n 1 , where the occupation number n 1 = 1(0) for the lower(upper) s.p. levels. N [a † 1 a † 2 a 3 a 4 ] is the normal-ordering form of operators. T...