Array versus linked-list

问题:

Why would someone want to use a linked-list over an array?

Coding a linked-list is, no doubt, a bit more work than using an array and one may wonder what would justify the additional effort.

I think insertion of new elements is trivial in a linked-list but it's a major chore in an array. Are there other advantages to using a linked list to store a set of data versus storing it in an array?

This question is not a duplicate of this question because the other question is asking specifically about a particular Java class while this question is concerned with the general data structures.

回答1:

• It's easier to store data of different sizes in a linked list. An array assumes every element is exactly the same size.
• As you mentioned, it's easier for a linked list to grow organically. An array's size needs to be known ahead of time, or re-created when it needs to grow.
• Shuffling a linked list is just a matter of changing what points to what. Shuffling an array is more complicated and/or takes more memory.
• As long as your iterations all happen in a "foreach" context, you don't lose any performance in iteration.

回答2:

Another good reason is that linked lists lend themselves nicely to efficient multi-threaded implementations. The reason for this is that changes tend to be local - affecting only a pointer or two for insert and remove at a localized part of the data structure. So, you can have many threads working on the same linked list. Even more, it's possible to create lock-free versions using CAS-type operations and avoid heavy-weight locks altogether.

With a linked list, iterators can also traverse the list while modifications are occurring. In the optimistic case where modifications don't collide, iterators can continue without contention.

With an array, any change that modifies the size of the array is likely to require locking a large portion of the array and in fact, it's rare that this is done without a global lock across the whole array so modifications become stop the world affairs.

回答3:

Wikipedia has very good section about the differences.

Linked lists have several advantages over arrays. Elements can be inserted into linked lists indefinitely, while an array will eventually either fill up or need to be resized, an expensive operation that may not even be possible if memory is fragmented. Similarly, an array from which many elements are removed may become wastefully empty or need to be made smaller.

On the other hand, arrays allow random access, while linked lists allow only sequential access to elements. Singly-linked lists, in fact, can only be traversed in one direction. This makes linked lists unsuitable for applications where it's useful to look up an element by its index quickly, such as heapsort. Sequential access on arrays is also faster than on linked lists on many machines due to locality of reference and data caches. Linked lists receive almost no benefit from the cache.

Another disadvantage of linked lists is the extra storage needed for references, which often makes them impractical for lists of small data items such as characters or boolean values. It can also be slow, and with a naïve allocator, wasteful, to allocate memory separately for each new element, a problem generally solved using memory pools.

http://en.wikipedia.org/wiki/Linked_list

回答4:

I'll add another - lists can act as purely functional data structures.

For instance, you can have completely different lists sharing the same end section

a = (1 2 3 4, ....)
b = (4 3 2 1 1 2 3 4 ...)
c = (3 4 ...)

i.e.:

b = 4 -> 3 -> 2 -> 1 -> a
c = a.next.next  

without having to copy the data being pointed to by a into b and c.

This is why they are so popular in functional languages, which use immutable variables - prepend and tail operations can occur freely without having to copy the original data - very important features when you're treating data as immutable.

回答5:

Merging two linked lists (especially two doubly linked lists) is much faster than merging two arrays (assuming the merge is destructive). The former takes O(1), the latter takes O(n).

EDIT: To clarify, I meant "merging" here in the unordered sense, not as in merge sort. Perhaps "concatenating" would have been a better word.

回答6:

Besides inserting into the middle of the list being easier - it's also much easier to delete from the middle of a linked list than an array.

But frankly, I've never used a linked list. Whenever I needed fast insertion and deletion, I also needed fast lookup, so I went to a HashSet or a Dictionary.

回答7:

A widely unappreciated argument for ArrayList and against LinkedList is that LinkedLists are uncomfortable while debugging. The time spent by maintenance developers to understand the program, e.g. to find bugs, increases and IMHO does sometimes not justify the nanoseconds in performance improvements or bytes in memory consumption in enterprise applicatons. Sometimes (well, of course it depends on the type of applications), it's better to waste a few bytes but have an application which is more maintainable or easier to understand.

For example, in a Java environment and using the Eclipse debugger, debugging an ArrayList will reveal a very easy to understand structure:

arrayList   ArrayList<String>
  elementData   Object[]
    [0] Object  "Foo"
    [1] Object  "Foo"
    [2] Object  "Foo"
    [3] Object  "Foo"
    [4] Object  "Foo"
    ...

On the other hand, watching the contents of a LinkedList and finding specific objects becomes a Expand-The-Tree clicking nightmare, not to mention the cognitive overhead needed to filter out the LinkedList internals:

linkedList  LinkedList<String>
    header  LinkedList$Entry<E>
        element E
        next    LinkedList$Entry<E>
            element E   "Foo"
            next    LinkedList$Entry<E>
                element E   "Foo"
                next    LinkedList$Entry<E>
                    element E   "Foo"
                    next    LinkedList$Entry<E>
                    previous    LinkedList$Entry<E>
                    ...
                previous    LinkedList$Entry<E>
            previous    LinkedList$Entry<E>
        previous    LinkedList$Entry<E>

回答8:

First of all, in C++ linked-lists shouldn't be much more trouble to work with than an array. You can use the std::list or the boost pointer list for linked lists. The key issues with linked lists vs arrays are extra space required for pointers and terrible random access. You should use a linked list if you

• you don't need random access to the data
• you will be adding/deleting elements, especially in the middle of the list

回答9:

For me it is like this,

1. Access

   • Linked Lists allow only sequential access to elements. Thus the algorithmic complexities is order of O(n)
   • Arrays allow random access to its elements and thus the complexity is order of O(1)

2. Storage

   • Linked lists require an extra storage for references. This makes them impractical for lists of small data items such as characters or boolean values.
   • Arrays do not need an extra storage to point to next data item. Each element can be accessed via indexes.

3. Size

   • The size of Linked lists are dynamic by nature.
   • The size of array is restricted to declaration.

4. Insertion/Deletion

   • Elements can be inserted and deleted in linked lists indefinitely.
   • Insertion/Deletion of values in arrays are very expensive. It requires memory reallocation.

回答10:

Two things:

Coding a linked list is, no doubt, a bit more work than using an array and he wondered what would justify the additional effort.

Never code a linked list when using C++. Just use the STL. How hard it is to implement should never be a reason to choose one data structure over another because most are already implemented out there.

As for the actual differences between an array and a linked list, the big thing for me is how you plan on using the structure. I'll use the term vector since that's the term for a resizable array in C++.

Indexing into a linked list is slow because you have to traverse the list to get to the given index, while a vector is contiguous in memory and you can get there using pointer math.

Appending onto the end or the beginning of a linked list is easy, since you only have to update one link, where in a vector you may have to resize and copy the contents over.

Removing an item from a list is easy, since you just have to break a pair of links and then attach them back together. Removing an item from a vector can be either faster or slower, depending if you care about order. Swapping in the last item over top the item you want to remove is faster, while shifting everything after it down is slower but retains ordering.

回答11:

Eric Lippert recently had a post on one of the reasons arrays should be used conservatively.

回答12:

Fast insertion and removal are indeed the best arguments for linked lists. If your structure grows dynamically and constant-time access to any element isn't required (such as dynamic stacks and queues), linked lists are a good choice.

回答13:

Here's a quick one: Removal of items is quicker.

回答14:

Linked-list are especially useful when the collection is constantly growing & shrinking. For example, it's hard to imagine trying to implement a Queue (add to the end, remove from the front) using an array -- you'd be spending all your time shifting things down. On the other hand, it's trivial with a linked-list.

回答15:

Other than adding and remove from the middle of the list, I like linked lists more because they can grow and shrink dynamically.

回答16:

No one ever codes their own linked list anymore. That'd be silly. The premise that using a linked list takes more code is just wrong.

These days, building a linked list is just an exercise for students so they can understand the concept. Instead, everyone uses a pre-built list. In C++, based the on the description in our question, that'd probably mean an stl vector (#include <vector> ).

Therefore, choosing a linked list vs an array is entirely about weighing the different characteristics of each structure relative to the needs of your app. Overcoming the additional programming burden should have zero impact on the decision.

回答17:

It's really a matter of efficiency, the overhead to insert, remove or move (where you are not simply swapping) elements inside a linked list is minimal, i.e. the operation itself is O(1), verses O(n) for an array. This can make a significant difference if you are operating heavily on a list of data. You chose your data-types based on how you will be operating on them and choose the most efficient for the algorithm you are using.

回答18:

Arrays make sense where the exact number of items will be known, and where searching by index makes sense. For example, if I wanted to store the exact state of my video output at a given moment without compression I would probably use an array of size [1024][768]. This will provide me with exactly what I need, and a list would be much, much slower to get the value of a given pixel. In places where an array does not make sense there are generally better data types than a list to deal with data effectively.

回答19:

Arrays Vs Linked List:

1. Array memory allocation will fail sometimes because of fragmented memory.
2. Caching is better in Arrays as all elements are allocated contiguous memory space.
3. Coding is more complex than Arrays.
4. No size constraint on Linked List, unlike Arrays
5. Insertion/Deletion is faster in Linked List and access is faster in Arrays.
6. Linked List better from multi-threading point of view.

回答20:

as arrays are static in nature, therefore all operations like memory allocation occur at the time of compilation only. So processor has to put less effort at its runtime .

回答21:

Suppose you have an ordered set, which you also want to modify by adding and removing elements. Further, you need ability to retain a reference to an element in such a way that later you can get a previous or next element. For example, a to-do list or set of paragraphs in a book.

First we should note that if you want to retain references to objects outside of the set itself, you will likely end up storing pointers in the array, rather than storing objects themselves. Otherwise you will not be able to insert into array - if objects are embedded into the array they will move during insertions and any pointers to them will become invalid. Same is true for array indexes.

Your first problem, as you have noted yourself, is insertion - linked list allows inserting in O(1), but an array would generally require O(n). This problem can be partially overcome - it is possible to create a data structure that gives array-like by-ordinal access interface where both reading and writing are, at worst, logarithmic.

Your second, and more severe problem is that given an element finding next element is O(n). If the set was not modified you could retain the index of the element as the reference instead of the pointer thus making find-next an O(1) operation, but as it is all you have is a pointer to the object itself and no way to determine its current index in the array other than by scanning the entire "array". This is an insurmountable problem for arrays - even if you can optimized insertions, there is nothing you can do to optimize find-next type operation.

回答22:

In an array you have the privilege of accessing any element in O(1) time. So its suitable for operations like Binary search Quick sort, etc. Linked list on the other hand is suitable for insertion deletion as its in O(1) time. Both has advantages as well as disadvantages and to prefer one over the other boils down to what you want to implement.

-- Bigger question is can we have a hybrid of both. Something like what python and perl implement as lists.

回答23:

Linked List

Its more preferable when it comes about insertion! Basically what is does is that it deals with the pointer

1 -> 3 -> 4

Insert (2)

1........3......4
.....2

Finally

1 -> 2 -> 3 -> 4

One arrow from the 2 points at 3 and the arrow of 1 points at 2

Simple!

But from Array

| 1 | 3 | 4 |

Insert (2) | 1 | 3 | | 4 | | 1 | | 3 | 4 | | 1 | 2 | 3 | 4 |

Well anyone can visualize the difference! Just for 4 index we are performing 3 steps

What if the array length is one million then? Is array efficient? The answer is NO! :)

The same thing goes for deletion! In Linked List we can simply use the pointer and nullify the element and next in the object class! But for array, we need to perform shiftLeft()

Hope that helps! :)

回答24:

Linked List are more of an overhead to maintain than array, it also requires additional memory storage all these points are agreed. But there are a few things which array cant do. In many cases suppose you want an array of length 10^9 you can't get it because getting one continous memory location has to be there. Linked list could be a saviour here.

Suppose you want to store multiple things with data then they can be easily extended in the linked list.

STL containers usually have linked list implementation behind the scene.

回答25:

Only reason to use linked list is that insert the element is easy (removing also).

Disadvatige could be that pointers take a lot of space.

And about that coding is harder: Usually you don't need code linked list (or only once) they are included in STL and it is not so complicated if you still have to do it.

回答26:

i also think that link list is more better than arrays. because we do traversing in link list but not in arrays

回答27:

Depending on your language, some of these disadvantages and advantages could be considered:

C Programming Language: When using a linked list (through struct pointers typically), special consideration must be made sure that you are not leaking memory. As was mentioned earlier, linked lists are easy to shuffle, because all were doing is changing pointers, but are we going to remember to free everything?

Java: Java has an automatic garbage collect, so leaking memory won't be an issue, but hidden from the high level programmer is the implementation details of what a linked list is. Methods such as removing a node from the middle of the list is more complicated of a procedure than some users of the language would expect it to be.

回答28:

Why a linked list over an array ? Well as some have already said, greater speed of insertions and deletions.

But maybe we don't have to live with the limits of either, and get the best of both, at the same time... eh ?

For array deletions, you can use a 'Deleted' byte, to represent the fact that a row has been deleted, thus reorging the array is no longer necessary. To ease the burden of insertions, or rapidly changing data, use a linked list for that. Then when referring to them, have your logic first search one, then the other. Thus, using them in combination gives you the best of both.

If you have a really large array, you could combine it with another, much smaller array or linked list where the smaller one hold thes 20, 50, 100 most recently used items. If the one needed is not in the shorter linked list or array, you go to the large array. If found there, you can then add it to the smaller linked list/array on the presumption that 'things most recently used are most likey to be re-used' ( and yes, possibly bumping the least recently used item from the list ). Which is true in many cases and solved a problem I had to tackle in an .ASP security permissions checking module, with ease, elegance, and impressive speed.

回答29:

While many of you have touched upon major adv./dis of linked list vs array, most of the comparisons are how one is better/ worse than the other.Eg. you can do random access in array but not possible in linked list and others. However, this is assuming link lists and array are going to be applied in a similar application. However a correct answer should be how link list would be preferred over array and vice-versa in a particular application deployment. Suppose you want to implement a dictionary application, what would you use ? Array : mmm it would allow easy retrieval through binary search and other search algo .. but lets think how link list can be better..Say you want to search "Blob" in dictionary. Would it make sense to have a link list of A->B->C->D---->Z and then each list element also pointing to an array or another list of all words starting with that letter ..

A -> B -> C -> ...Z
|    |    |
|    |    [Cat, Cave]
|    [Banana, Blob]
[Adam, Apple]

Now is the above approach better or a flat array of [Adam,Apple,Banana,Blob,Cat,Cave] ? Would it even be possible with array ? So a major advantage of link list is you can have an element not just pointing to the next element but also to some other link list/array/ heap/ or any other memory location. Array is a one flat contigous memory sliced into blocks size of the element it is going to store.. Link list on the other hand is a chunks of non-contigous memory units (can be any size and can store anything) and pointing to each other the way you want. Similarly lets say you are making a USB drive. Now would you like files to be saved as any array or as a link list ? I think you get the idea what I am pointing to :)

回答30:

Why would someone want to use a linked-list over an array?

This is only one reason - if you need a linked list data structure and a programming language you are using is not supports a pointers.