respa thermo output

Hi All,

I have an issue with the run_style respa command. If I use
run_style respa 2 2 bond 1 angle 1 dihedral 1 pair 2 kspace 2

I find that with the thermodynamic output (vdwl/coul energies and pressure) is incorrect. This appears to be related to whether or not the pairwise interactions are calculated on the innermost loop (run_style respa 2 2 bond 1 angle 1 dihedral 1 pair 1 kspace 2 seems OK) .

I am attaching an input script (modified from the rhodo example) that produces the problem. It needs to be combined with the data.rhodo that is distributed with lammps in examples/gpu.

I find this problem with the 9-May version of lammps. Output below.

Regards,
Mike Salerno

NO RESPA:

log.lammps-norespa:Step Temp E_vdwl E_coul E_long E_bond E_angle E_dihed TotEng Press
log.lammps-norespa- 0 301.19057 1263.5796 239891.9 -269452.07 3520.624 14653.009 7646.8561 19591.976 9532.1941
log.lammps-norespa- 10 372.60694 1025.6181 236995.33 -269593.04 3032.4027 13461.978 7539.5529 19566.986 10449.13
log.lammps-norespa- 20 469.19976 450.35878 231076.4 -269844.13 3038.8345 13377.331 7439.0051 19487.761 11032.139
log.lammps-norespa-Loop time of 0.835453 on 12 procs for 20 steps with 32000 atoms

in.cpu.rhodo (660 Bytes)