LAMMPS DumpReader returns incorrect coordinate transform

[wjq1228]

Expected behavior

The coordinate transform of "scaled" LAMMPS trajectory should be imported correctly.

Actual behavior

The MDAnalysis calculated incorrectly coordinates from scaled to unscaled. Calculation method of “scaled” is xs = (x-xlo)/(xhi-xlo). So, when coordinate transform, the coordinate can be calculated by x = (xhi-xlo)*xs + xlo. However, the LAMMPS DumpReader returns incorrect coordinate transform. For example, the scaled coordinate in my dump is

ITEM: TIMESTEP
0
ITEM: NUMBER OF ATOMS
6024
ITEM: BOX BOUNDS pp pp pp
3.3501000000000003e-02 6.2616999999999997e+01
-5.2700000000000002e-01 2.8887000000000000e+01
-1.5000000000000000e+01 4.0000000000000000e+01
ITEM: ATOMS id type xs ys zs
2608 2 0.0407033 0.457803 0.384662

The real coordinate of atom 2608 is [2.58086, 12.9388, 6.15641]. However, the coordinate calculated by DumpReader is [2.513854, 13.992818, 36.15641]. Calculation method of DumpReader seem to be x = (xhi-xlo)*xs - xlo. Other atoms are tested and the same error happened.

Code to reproduce the behavior

import numpy as np
import MDAnalysis as mda
from MDAnalysis.analysis.hydrogenbonds import HydrogenBondAnalysis
from MDAnalysis.analysis import hydrogenbonds

# load system
u = mda.Universe("carbonice3nm.data", "ice.lammpstrj", format="LAMMPSDUMP", lammps_coordinate_convention="scaled", dt=10)

O = u.select_atoms('id 2608')
print(O.positions)

## Current version of MDAnalysis ##

- Which version are you using? (2.8.0)
- Which version of Python? (3.10.0)
- Which operating system? (Windows)
- LAMMPS version 2 Aug 2023

[samkenzo]

If the coordinate transformation is indeed incorrect, you can manually calculate the coordinates using the correct formula as a workaround until an update is available.
mdsimulation.txt

i have attached the file here see if it works

[wjq1228]

Very thanks! Looking forward to updates soon!

[samkenzo]

appreciate it
!

[orbeckst]

@wjq1228 do you have a reference (eg from the LAMMPS docs) that defines the scaled coordinate transformation?

[hmacdope]

We will look into this, concerning if correct.