Question abou "Out of range atoms - cannot compute PPPM (src/KSPACE/pppm.cpp:1839)" when run lammps with dplr model

[datieniuyyds]

Deepmd-kit version used to run dplr model is v2.1.1
Deepmd-kit version used to run lammps with dplr model is v2.2.3
My system is liquid water and different concentration CaCl2 solutions. And in the wannierization process, i just done projections on O and Cl atoms to get the dw model (and confirmed that projections with or without ca atoms had little effect on wannier center). I can successfully used the dplr model to run lammps in the liquid water system for cell expansion, but when i ran lammps for other CaCl2 systems, I ran into difficulties with out of range atoms. And i did some tests as follows:

First, I used a frame of structure from a dp run lammps balanced as the initial structure file for a dplr run md. It does not work.

Second, I used the un-expanded cell structure (about 200 to 250 atoms) as the initial structure for a dplr run md. Lower concentrations that include only one CaCl2 can be run successfully, but other concentrations that are a bit higher still won't work.

Third, I used an unexpanded cell structure with a concentration a little higher than the second attempted concentration as the initial structure, and lowered the temperature in the input file to 300k (330k for the previous), and this concentration could be run successfully, but the other higher concentrations still could not be run successfully when the temperature was lowered.

In the output file of the failed run, the temperature rises dramatically. Attached are the input and output files for one of my failed columns.

I would like to ask you how this situation can be solved, is it a problem with the dplr model, the dp version, or something else?

CaCL2-test.zip
out.zip