In chapter 6.3.1 of the thesis Purely Functional Data Structures, says:
Then, whenever we create a new tree from a new element and a segment of trees of ranks 0... r-1, we simply compare the new element with the first root in the segment (i.e.,the root of the rank 0 tree). The smaller element becomes the new root and the larger element becomes the rank 0 child of the root.
- T0' is the new tree has rank 0
- T0..T(r-1) are the original trees rank 0 to r-1
- The smaller element becomes the new root and the larger element becomes rank 0 child of the root
The question is that step 3 result in two rank 1 trees, which is conflict with the binomial heaps.
Am I misunderstanding?