When it comes to collections, the Java standard library provides plenty of options to choose from. Among those options are two famous List implementations known as ArrayList and LinkedList, each with their own properties and use-cases.
In this tutorial, we're going to see how these two are actually implemented. Then, we'll evaluate different applications for each one.
Internally, ArrayList is using an array to implement the List interface. As arrays are fixed size in Java, ArrayList creates an array with some initial capacity. Along the way, if we need to store more items than that default capacity, it will replace that array with a new and more spacious one.
When we're creating an empty ArrayList, it initializes its backing array with a default capacity (currently 10):
Adding a new item while that array it not yet full is as simple as assigning that item to a specific array index. This array index is determined by the current array size since we're practically appending to the list:
backingArray[size] = newItem; size++;
So, in best and average cases, the time complexity for the add operation is O(1), which is pretty fast. As the backing array becomes full, however, the add implementation becomes less efficient:
To add a new item, we should first initialize a brand new array with more capacity and copy all existing items to the new array. Only after copying current elements can we add the new item. Hence, the time complexity is O(n) in the worst case since we have to copy n elements first.
Theoretically speaking, adding a new element runs in amortized constant time. That is, adding n elements requires O(n) time. However, some single additions may perform poorly because of the copy overhead.
2.2. Access by Index
Accessing items by their indices is where the ArrayList really shines. To retrieve an item at index i, we just have to return the item residing at the ith index from the backing array. Consequently, the time complexity for access by index operation is always O(1).
2.3. Remove by Index
Suppose we're going to remove the index 6 from our ArrayList, which corresponds to the element 15 in our backing array:
After marking the desired element as deleted, we should move all elements after it back by one index. Obviously, the nearer the element to the start of the array, the more elements we should move. So the time complexity is O(1) at the best-case and O(n) on average and worst-cases.
2.4. Applications and Limitations
Usually, ArrayList is the default choice for many developers when they need a List implementation. As a matter of fact, it's actually a sensible choice when the number of reads is far more than the number of writes.
Sometimes we need equally frequent reads and writes. If we do have an estimate of the maximum number of possible items, then it still makes sense to use ArrayList. If that's the case, we can initialize the ArrayList with an initial capacity:
int possibleUpperBound = 10_000; List<String> items = new ArrayList<>(possibleUpperBound);
This estimation may prevent lots of unnecessary copying and array allocations.
Moreover, arrays are indexed by int values in Java. So, it's not possible to store more than 232 elements in a Java array and, consequently, in ArrayList.
LinkedList, as its name suggests, uses a collection of linked nodes to store and retrieve elements. For instance, here's how the Java implementation looks after adding four elements:
Each node maintains two pointers: one pointing to the next element and another referring to the previous one. Expanding on this, the doubly linked list has two pointers pointing to the first and last items.
Again, let's evaluate this implementation with respect to the same fundamental operations.
In order to add a new node, first, we should link the current last node to the new node:
And then update the last pointer:
As both of these operations are trivial, the time complexity for the add operation is always O(1).
3.2. Access by Index
LinkedList, as opposed to ArrayList, does not support fast random access. So, in order to find an element by index, we should traverse some portion of the list manually.
In the best case, when the requested item is near the start or end of the list, the time complexity would be as fast as O(1). However, in the average and worst-case scenarios, we may end up with an O(n) access time since we have to examine many nodes one after another.
3.3. Remove by Index
In order to remove an item, we should first find the requested item and then un-link it from the list. Consequently, the access time determines the time complexity — that is, O(1) at best-case and O(n) on average and in worst-case scenarios.
LinkedLists are more suitable when the addition rate is much higher than the read rate.
Also, it can be used in read-heavy scenarios when most of the time we want the first or last element. It's worth mentioning that LinkedList also implements the Deque interface – supporting efficient access to both ends of the collection.
Generally, if we know their implementation differences, then we could easily choose one for a particular use-case.
For instance, let's say that we're going store a lot of time-series events in a list-like data structure. We know that we would receive bursts of events each second.
Also, we need to examine all events one after another periodically and provide some stats. For this use-case, LinkedList is a better choice because the addition rate is much higher than the read rate.
Also, we would read all the items, so we can't beat the O(n) upper bound.
In this tutorial, first, we took a dive into how ArrayList and LinkLists are implemented in Java.
We also evaluated different use-cases for each one of these.