We perform numerical simulations for the formation of the Galactic stellar halo , based on the currently favored cold dark matter ( CDM ) theory of galaxy formation .
Our numerical models , taking into account both dynamical and chemical evolution processes in a consistent manner , are aimed at explaining observed structure and kinematics of the stellar halo in the context of hierarchical galaxy formation .
The main results of the present simulations are summarized as follows .

( 1 ) Basic physical processes involved in the formation of the stellar halo , composed of metal-deficient stars with [ Fe/H ] \leq -1.0 , are described by both dissipative and dissipationless merging of subgalactic clumps and their resultant tidal disruption in the course of gravitational contraction of the Galaxy at high redshift ( z > 1 ) .

( 2 ) The simulated halo has the density profile similar to the observed power-law form of \rho ( r ) \sim r ^ { -3.5 } , and has also the similar metallicity distribution to the observations .
The halo virtually shows no radial gradient for stellar ages and only small gradient for metallicities .

( 3 ) The dual nature of the halo , i.e. , its inner flattened and outer spherical density distribution , is reproduced , at least qualitatively , by the present model .
The outer spherical halo is formed via essentially dissipationless merging of small subgalactic clumps , whereas the inner flattened one is formed via three different mechanisms , i.e. , dissipative merging between larger , more massive clumps , adiabatic contraction due to the growing Galactic disk , and gaseous accretion onto the equatorial plane .

( 4 ) For the simulated metal-poor stars with [ Fe/H ] \leq -1.0 , there is no strong correlation between metal abundances and orbital eccentricities , in good agreement with the recent observations .
Moreover , the observed fraction of the low-eccentricity stars is reproduced correctly for [ Fe/H ] \leq -1.6 and approximately for the intermediate abundance range of -1.6 < [ Fe/H ] \leq -1.0 .

( 5 ) The mean rotational velocity of the simulated halo , < V _ { \phi } > , is somewhat positive ( prograde ) at [ Fe/H ] < -2.2 and increases linearly with [ Fe/H ] at [ Fe/H ] > -2.2 .
The stars at smaller distance from the disk plane appear to show systematically larger < V _ { \phi } > .

Based on these results , we discuss how early processes of dissipationless and dissipative merging of subgalactic clumps can reproduce plausibly and consistently the recent observational results on the Galactic stellar halo .
We also present a possible scenario for the formation of the entire Galaxy structure , including bulge and disk components , in conjunction with the halo formation .