Some question about calculating the invlambda

[YongyuG]

Thnaks for your work,

I dont understand why the invlambda_l is defineds as:
invlambda(1,delta+K:min(delta+K+sT-1, T))

In my view, i think it should be
invlambda(1,delta+K+1:min(delta+K+sT-1, T)) because delta is delay frame and K are the pass frames, so we are going to estimate the frame started at "delta + k +1"

can you tell me the difference or any my mistaken understanding.