We define the complexity of the algorithm as a function of the geometric
complexity of the input, in this case the number of part vertices, . We
neglect the numerical complexity of representing vertices and angles as
rational numbers.
Step 1 (computing the squeeze function) can be performed in time [6]. Step 3 runs in time
. To see this, note that
we only need to consider positioning the box flush with each step and there
are
steps. We only need
iterations of Step 3, since
intervals can be constructed from continuous subsequences of the
steps in the squeeze function. Thus the algorithm runs in time
and finds plans of length
.