atoms drift around when set frozen

Dear all,

In a simulation cell I have 2 groups of atoms, fixed and mobile, defined using 'region' commands. The fixed group is set frozen using 'setforce 0.0 0.0 0.0', with initial velocity not set. The mobile group is given an initial velocity and controlled by npt. When I checked the trajectory I found the fixed atoms drifted slightly from their initial positions. Is this something to do with the npt settings of the mobile group which causes the change in simulation cell volume? Thanks.

The relevant part of the input file is:
region in_blk block 37.0 77.0 INF INF INF INF side in units box
region out_blk block 37.0 77.0 INF INF INF INF side out units box
group fixblk region in_blk
group freeblk region out_blk
velocity freeblk create 300.0 4928459 rot yes dist gaussian
variable tmp index 3 4 5
variable temperature equal v_tmp*100.0*6368.0/9792.0
label looptemp
     fix ffree freeblk npt temp \{temperature\} {temperature} 0.1 iso 0.0 0.0 1.0
     fix ffix fixblk setforce 0.0 0.0 0.0
     run 200000
next tmp
jump in.eam looptemp


Fix npt dilates the box which rescales all atoms,
unless you tell it not to, via dilate partial.