Chapter 2, Building Abstractions with Data

Section - 2.3 Symbolic Data

Exercise 2.63


Yes, they will produce same results for every tree.

Looking closely reveals that both algorithms are similar in the following way:

  • Convert left tree to list, lets call the result left-list.
  • Convert right tree to list, lets call the result right-list.
  • Create a combined list as (left-list, current-entry, right-list)

They differ only in the way left and right lists are merged. The first procedure uses append while other procedure does it employing the recursive process.


Clearly both procedures are invoked number of times, where is number of leaves in the tree.

While in the first procedure, there is an additional invocation of append. In the second procedure there is no other function call(assuming null?, cons etc will take constant time) that changes the time complexity. Thus complexity of second procedure is .

Lets compute the complexity of first procedure.

Assuming tree is balanced, the additional call append takes in each step, where is the number of nodes present in the tree in that step. Since in each step, number of nodes are halved, the total time taken by compute is:


Thus total time will be .

Thus first procedure is roughly 3 times more costly than second procedure but note that both have same order of growth .