You need to add the last point.
On the route 0 —> 1 there is usually no problem of going 1.002 when calculating the perturbation.
On the route 1 —> 0, some parameters may not make sense with negative values, so do a run at 0.002 and use the FDTI from that point.
You can see an example in lammps/examples/USER/fep/CH4hyd/fdti10
(examples have been updated since 2018; the code I don’t think so, at least in it’s functional aspects)
I’m finding out the free energy of CO2 in oil