Problem with geometric optimization of unit cell with DMol3

I am facing a problem optimizing the geometric structure of a unit cell of Ni2P using DMol3 in MS4.4. The energy fluctuates with the optimization steps so do the energy change , max displacement and max force.

Can anybody tell me whats going wrong here?

input file >>

# Task parameters

Calculate optimize

Opt_energy_convergence 2.0000e-005

Opt_gradient_convergence 4.0000e-003 A

Opt_displacement_convergence 5.0000e-003 A

Opt_iterations 1000

Opt_max_displacement 0.3000 A

Lattice_optimization on

Simple_geometry_opt

Symmetry on

Max_memory 10000

# Electronic parameters

Spin_polarization unrestricted

Charge 0

Basis dnd

Pseudopotential none

Functional gga(p91)

Aux_density octupole

Integration_grid medium

Occupation thermal 0.0050

Cutoff_Global 5.5000 angstrom

Scf_density_convergence 1.0000e-005

Scf_charge_mixing 0.2000

Scf_spin_mixing 0.5000

Scf_iterations 1000

Scf_diis 6 pulay

use_DIIS_restart off

 

# Kpoint definition file (intervals/offset):

Kpoints file 5 5 5 0.0000 0.0000 0.0000

Ni2P.kpoints

# Calculated properties