An ordered linked list is a linear data structure consisting of a collection of nodes, each containing two main components: a data element, often referred to as the "value" or "key", and a "reference" or "link" to the next node in the sequence. These nodes are organized in a specific order based on the values of their data elements, typically in ascending or descending order.
In order to implement the ordered linked list, we must remember that the relative positions of the items are based on some underlying characteristic. The ordered linked list of integers given above (17, 26, 31, 54, 77, and 93) can be represented by a linked structure as shown in Figure 4.6.2. Again, the node and link structure is ideal for representing the relative positioning of the items.
To implement the OrderedList class, we will use the same technique as seen previously with unordered linked lists. Once again, an empty linked list will be denoted by a head reference to NULL (see Listing 4.6.3).
As we consider the operations for the ordered linked list, we should note that the isEmpty and size methods can be implemented the same as with unordered linked lists since they deal only with the number of nodes in the linked list without regard to the actual item values. Likewise, the remove method will work just fine since we still need to find the item and then link around the node to remove it. The two remaining methods, search and add, will require some modification.
The search of an unordered linked list required that we traverse the nodes one at a time until we either find the item we are looking for or run out of nodes (NULL). It turns out that the same approach would actually work with the ordered linked list and in fact in the case where we find the item it is exactly what we need. However, in the case where the item is not in the linked list, we can take advantage of the ordering to stop the search as soon as possible.
For example, Figure 4.6.4 shows the ordered linked list as a search is looking for the value 45. As we traverse, starting at the head of the linked list, we first compare against 17. Since 17 is not the item we are looking for, we move to the next node, in this case 26. Again, this is not what we want, so we move on to 31 and then on to 54. Now, at this point, something is different. Since 54 is not the item we are looking for, our former strategy would be to move forward. However, due to the fact that this is an ordered linked list, that will not be necessary. Once the value in the node becomes greater than the item we are searching for, the search can stop and return False. There is no way the item could exist further out in the linked list.
Listing 4.6.5 shows the complete search method. It is easy to incorporate the new condition discussed above by adding another boolean variable, stop, and initializing it to False (line 4). While stop is False (in other words, while the search is still ongoing) we can continue to look forward in the linked list (line 5). If any node is ever discovered that contains data greater than the item we are looking for, we will set stop to True (lines 9–10). The remaining lines are identical to the unordered linked list search.
The most significant method modification will take place in add. Recall that for unordered linked lists, the add method could simply place a new node at the head of the linked list. It was the easiest point of access. Unfortunately, this will no longer work with ordered linked lists. It is now necessary that we discover the specific place where a new item belongs in the existing ordered linked list.
Assume we have the ordered linked list consisting of 17, 26, 54, 77, and 93 and we want to add the value 31. The add method must decide that the new item belongs between 26 and 54. Figure 4.6.6 shows the setup that we need. As we explained earlier, we need to traverse the linked list looking for the place where the new node will be added. We know we have found that place when either we run out of nodes (current becomes NULL) or the value of the current node becomes greater than the item we wish to add. In our example, seeing the value 54 causes us to stop.
As we saw with unordered linked lists, it is necessary to have an additional reference, again called previous, since current will not provide access to the node that must be modified. Listing 4.6.7 shows the complete add method. Lines 2–3 set up the two external references and lines 9–10 again allow previous to follow one node behind current every time through the iteration. The condition (line 5) allows the iteration to continue as long as there are more nodes and the value in the current node is not larger than the item. In either case, when the iteration fails, we have found the location for the new node.
The remainder of the method completes the two-step process shown in Figure 4.6.6. Once a new node has been created for the item, the only remaining question is whether the new node will be added at the beginning of the linked list or some place in the middle. Again, previous == NULL (line 13) can be used to provide the answer.
The OrderedList class with methods discussed thus far can be found in ActiveCode 1. We leave the remaining methods as exercises. You should carefully consider whether the unordered implementations will work given that the linked list is now ordered.
Subsection4.6.1Analysis of Linked Lists
To analyze the complexity of the linked list operations, we need to consider whether they require traversal. Consider a linked list that has n nodes. The isEmpty method is \(O(1)\) since it requires one step to check the head reference for NULL. size, on the other hand, will always require n steps since there is no way to know how many nodes are in the linked list without traversing from head to end. Therefore, size is \(O(n)\text{.}\) Adding an item to an unordered linked list will always be O(1) since we simply place the new node at the head of the linked list. However, search and remove, as well as add for an ordered linked list, all require the traversal process. Although on average they may need to traverse only half of the nodes, these methods are all \(O(n)\) since in the worst case each will process every node in the linked list.
Checkpoint4.6.8.
Match the Big O() analysis to their corresponding opperation.
Try again!
O(1)
isEmpty, add (unordered linked list)
O(n)
length,add, search, and remove(ordered linked list)
Checkpoint4.6.9.
Q-3: After having read over unordered and ordered linked lists, what do you think a circular linked list would do differently from an ordered or unordered linked list? (Hint: think about the example from the beginning of the chapter.)
In a circular linked list, the head Node of the linked list contains a pointer to the last node in the list.
Wrong! the head Node of the list will only contain a pointer to the second Node.
In a circular linked list, the last Node of the linked list contains a pointer to the head node of the list rather than pointing to NULL.
Correct! the final Node of the linked list will contain a pointer to the first node so that it is possible to make "circles" around the list.
In a circular linked list, every node contains a pointer to the head of the list, making it possible to return back to the beginning of the list at any time.
Wrong! None of the nodes in the middle of the list will ever point to the head node in a circular linked list.
In a circular linked list, the head and final Node of the linked list point to each other, making it possible to traverse through the list in both directions.
Hint: This would be possible in a circular doubly linked list, but not a circular linked list.