What are the differences between PIMD results and experiments(or PIMC) results?

Hello! I am trying to use ‘fix pimd all pimd method pimd temp 0.1’ to simulation the HCP-solid-He4 at ~0.1K. And it’s all know that the He4 is qunatum crystal, only quantum simulation method such as path-integral Monte Carlo(PIMC) can accurately reproduced the lattice constant (Molar volume 21cm^3 at about ~26 bars and 0.1K ) and other property in the experiments measurements.

As a kind of quantum simulation method, I think PIMD can also simulate solid he4 at low temperatures. However, when the experimentally measured structure is relaxed under the NVT ensemble, He4 becomes a liquid phase and the temperature is no longer maintained near 0.1K. And the solid phase only can stable at ~15.055cm^3.(For comparison, the molar mass will reduce to ~9cm^3 when minimized energy without PIMD method)
To sum up, I wonder what causes the difference between PIMD results in LAMMPS and experimental measurements and PIMC results? Is there a way to eliminate this error?

Thanks in advance.Any reply would be greatly appreciated.

My lammps version is lammps-31Mar17.
The potential I use is Aziz’s potential which has been successfully applied in the PIMC calculations for solid-He4 with Molar volume at 21cm^3.

And I simulate the solid-He4 with 64 atom.
the input file is as follow:

variable i loop 1
variable ibead uloop 30 pad
variable seed index   He.lmp
variable dump_relax index He.dump
variable potential index  aziz-He1995 aziz-He1987 aziz-He1979
variable runstep equal 200000
variable timesp equal 0.0001
variable logstep equal 5000
variable ngb equal 2  #define neighbor
variable T equal 0.1    #define simulation temperature
variable S equal 0.01   #define strain rate

units metal 
boundary  p p p 
atom_style atomic 
neighbor ${ngb} bin 
atom_modify map array
read_data ${seed}
pair_style table linear 10000
pair_coeff * * ${potential}.table HeHe_Aziz 

thermo 1000
thermo_style custom step lx ly lz press pxx pyy pzz pe temp f_pimd[1] f_pimd[2] f_pimd[3] 
dump 1 all custom 100 ${ibead}${dump_relax}${potential}.xyz id type x y z 

#############relax with pimd #########
timestep ${timesp}
velocity all create 0.1 666${ibead} rot  yes  dist gaussian #  rot yes dist gaussian
fix pimd all pimd method pimd temp 0.1 nhc 4
run ${runstep}
clear 
next i
next potential
jump in.pimd
print "all done"
quit

This is far too old a LAMMPS version. It has many known bugs and that includes fix pimd (which is now called fix pimd/nvt).

That will need an expert in PIMD simulations to provide a competent answer. There are many possible sources of errors in all kinds of simulation. It is not easy to spot them all from just a vague description like yours and without specific references.