In discussion with my friend, we reached a conclusion that transformation formula of velosity v to another IFR moving V, i.e.
v'=\frac{v+V}{1+vV/c^2}
is valid even if v is hypothetical velocity,i,e,
v=\frac{x_2-x_1}{t_2-t_1}
v'=\frac{x'_2-x'_1}{t'_2-t'_1}
where interval of ##(t_1,x_1)\rightarrow...