Problems encountered when using fix pimd/nvt

Hello all:
I’m attempting to run pimd/nvt with the stable version (2 Aug 2023) .

fix 1 all pimd/nvt method nmpimd fmass 1.0 temp 300 nhc 4

When I use fix nvt, the temperature stabilizes at 300K as expected.
When I do not expand the beads, the temperature can be maintained at around 310K.
However, when the input command specifying a 4-bead simulation, the temperatures decrease and and vary from each other. The temperature of the first bead and the third bead is shown below.

export OMPI_ALLOW_RUN_AS_ROOT=1 &&
export OMPI_ALLOW_RUN_AS_ROOT_CONFIRM=1 &&
mpirun -np 4 lmp_mpi -in in.lammps -p 4x1 -log log

2f846f74f47e385b59f87e73232852e
7cec49b7a3e33cc7e4a10f046fa7cc8

Am I missing something in these simulations?I don’t know where I went wrong.

Any advice is much appreciated,
Thanks!