Thank you very much.
Arun
this is easily confirmed by changing your input to:
pair_modify mix arithmetic compute no
kspace_style pppm 1.0e-4
delete_bonds all multi remove
compute 1 all pe pair
compute 2 all pe bond
compute 3 all pe angle
compute 4 all pe dihedral
compute 5 all pe kspace
thermo_style custom time pe c_1 c_2 c_3 c_4 c_5
BTW: this means that what you complained about, i.e. that the tail
correction was active is not correct either.
nevertheless, the patch i proposed should be applied, since compute no
is indeed incompatible with shift yes and tail yes.
axel.