High stress after box relaxation

Dear Users/developers,

I found some unexpected results on this simulation: I want to compute some shear properties of a crystal. As a first step I relax the atomic corrdinates together with the box volume/shape:

units real

dimension 3
processors * * *
boundary p p p
atom_style charge

read_data atoms.data

pair_style reax/c NULL
pair_coeff * * ffield.reax 2 3 4 5
fix 99 all qeq/reax 1 0.0 10.0 1.0e-6 reax/c
neighbor 2.0 bin
neigh_modify every 10 check no

min_style cg
thermo 10
thermo_style custom step etotal vol lx ly lz xy xz yz press pxx pyy pzz pxy pxz pyz
fix box1 all box/relax tri 0.0 vmax 0.0001 dilate all nreset 25
minimize 1.0e-9 1.0e-9 5000 50000

write_restart relax.rest

In the restarted run, does the initial stress (timestep 0) equal
the final minimized stress? Also, if you are running dynamics with
assigned velocities than that adds a kinetic component
to the new stress.