EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

Home Lectures
Class Info Newsgroup
Links Assignments
STL Summary
This is a brief summary of the containers in the C++ Standard Library (once
informally known as the Standard Template Library or STL). It deliberately sacrifices
completeness for simplicity. Information is also available on the iterators and
algorithms.

In my opinion, the best overall reference to the Standard Library is Josuttis' The C++
Standard Library.

The containers described below are:

vectors
lists
deques
stacks
queues
priority queues
sets and multisets
maps and multimaps
pairs

For each operation, there is a guaranteed upper-bound on the complexity. See

Section 19.2.1 of Deitel for a brief description of "Big Oh" notation.

Vector
Header
#include <vector>

Constructors

vector<T> v; Make an empty vector. O(1)

vector<T> v(n); Make a vector with N elements. O(n)
vector<T> v(n, value); Make a vector with N elements, initialized to value. O(n)
vector<T> v(begin, Make a vector and copy the elements from begin to
end); O(n)
end.

Accessors

v[i] Return (or set) the I'th element. O(1)

v.at(i) Return (or set) the I'th element, with bounds checking. O(1)

1 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

v.size() Return current number of elements. O(1)

v.empty() Return true if vector is empty. O(1)
v.begin() Return random access iterator to start. O(1)
v.end() Return random access iterator to end. O(1)
v.front() Return the first element. O(1)
v.back() Return the last element. O(1)
v.capacity() Return maximum number of elements. O(1)

Modifiers

O(1)
v.push_back(value) Add value to end.
(amortized)
v.insert(iterator, Insert value at the position indexed by
value) O(n)
iterator.
v.pop_back() Remove value from end. O(1)
v.erase(iterator) Erase value indexed by iterator. O(n)
v.erase(begin, end) Erase the elements from begin to end. O(n)

Deque
Header
#include <deque>

Constructors

deque<T> d; Make an empty deque. O(1)

deque<T> d(n); Make a deque with N elements. O(n)
deque<T> d(n, value); Make a deque with N elements, initialized to value. O(n)
deque<T> d(begin, end); Make a deque and copy the values from begin to end. O(n)

Accessors

d[i] Return (or set) the I'th element. O(1)

d.at(i) Return (or set) the I'th element, with bounds checking. O(1)
d.size() Return current number of elements. O(1)
d.empty() Return true if deque is empty. O(1)
d.begin() Return random access iterator to start. O(1)
d.end() Return random access iterator to end. O(1)
d.front() Return the first element. O(1)
d.back() Return the last element. O(1)

Modifiers

2 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

O(1)
d.push_front(value) Add value to front.
(amortized)
O(1)
d.push_back(value) Add value to end.
(amortized)
d.insert(iterator, Insert value at the position indexed by
value) O(n)
iterator.
d.pop_front() Remove value from front. O(1)
d.pop_back() Remove value from end. O(1)
d.erase(iterator) Erase value indexed by iterator. O(n)
d.erase(begin, end) Erase the elements from begin to end. O(n)

List
Header
#include <list>

Constructors

list<T> l; Make an empty list. O(1)

list<T> l(begin, end); Make a list and copy the values from begin to end. O(n)

Accessors

l.size() Return current number of elements. O(1)

l.empty() Return true if list is empty. O(1)
l.begin() Return bidirectional iterator to start. O(1)
l.end() Return bidirectional iterator to end. O(1)
l.front() Return the first element. O(1)
l.back() Return the last element. O(1)

Modifiers

l.push_front(value) Add value to front. O(1)

l.push_back(value) Add value to end. O(1)
l.insert(iterator, value) Insert value after position indexed by iterator. O(1)
l.pop_front() Remove value from front. O(1)
l.pop_back() Remove value from end. O(1)
l.erase(iterator) Erase value indexed by iterator. O(1)
l.erase(begin, end) Erase the elements from begin to end. O(1)
l.remove(value) Remove all occurrences of value. O(n)
l.remove_if(test) Remove all element that satisfy test. O(n)
l.reverse() Reverse the list. O(n)

3 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

l.sort() Sort the list. O(n log n)

l.sort(comparison) Sort with comparison function. O(n logn)
l.merge(l2) Merge sorted lists. O(n)

Stack
In the C++ STL, a stack is a container adaptor. That means there is no primitive
stack data structure. Instead, you create a stack from another container, like a list,
and the stack's basic operations will be implemented using the underlying container's
operations.

Header
#include <stack>

Constructors

stack<T> s; Make an empty stack using a deque. O(1)

stack<T, container<T> > s; Make an empty stack using the given container. O(1)

Accessors

s.top() Return the top element. O(1)

s.size() Return current number of elements. O(1)
s.empty() Return true if stack is empty. O(1)

Modifiers

s.push(value) Push value on top. Same as push_back() for underlying container.

s.pop() Pop value from top. O(1)

Queue
In the C++ STL, a queue is a container adaptor. That means there is no primitive
queue data structure. Instead, you create a queue from another container, like a list,
and the queue's basic operations will be implemented using the underlying
container's operations.

Don't confuse a queue with a deque or a priority_queue.

Header
#include <queue>

Constructors

4 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

queue<T> q; Make an queue stack using a deque. O(1)

queue<T, container<T> > q; Make an empty queue using the given container. O(1)

Accessors

q.front() Return the front element. O(1)

q.back() Return the rear element. O(1)
q.size() Return current number of elements. O(1)
q.empty() Return true if queue is empty. O(1)

Modifiers

Same for push_back() for underlying

q.push(value) Add value to end.
container.
Remove value from
q.pop() O(1)
front.

Priority Queue
In the C++ STL, a priority queue is a container adaptor. That means there is no
primitive priorty queue data structure. Instead, you create a priority queue from
another container, like a deque, and the priority queue's basic operations will be
implemented using the underlying container's operations.

Priority queues are neither first-in-first-out nor last-in-first-out. You push objects
onto the priority queue. The top element is always the "biggest" of the elements
currently in the priority queue. Biggest is determined by the comparison predicate
you give the priority queue constructor.

If that predicate is a "less than" type predicate, then biggest means

largest.

If it is a "greater than" type predicate, then biggest means smallest.

Header

#include <queue> -- not a typo!

Constructors

priority_queue<T, Make an empty priority queue using the given O(1)

container<T>,
comparison<T> > container to hold values, and comparison to
q; compare values. container defaults to vector<T>
and comparison defaults to less<T>.

Accessors

5 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

q.top() Return the "biggest" element. O(1)

q.size() Return current number of elements. O(1)
q.empty() Return true if priority queue is empty. O(1)

Modifiers

q.push(value) Add value to priority queue. O(log n)

q.pop() Remove biggest value. O(log n)

Set and Multiset

Sets store objects and automatically keep them sorted and quick to find. In a set,
there is only one copy of each object. If you try to add another equal object, nothing
happens. multisets are declared and used the same as sets but allow duplicate
elements.

Sets are implemented with balanced binary search trees, typically red-black trees.
Thus, they provide logarithmic storage and retrieval times. Because they use search
trees, sets need a comparison predicate to sort the keys. operator<() will be used by
default if none is specified a construction time.

In a set, one object is considered equal to another if it is neither less than

nor greater than the other object. operator==() is not used.

Header
#include <set>

Constructors

Make an empty set. compare should be a binary predicate

set< type, compare
> s; for ordering the set. It's optional and will default to a O(1)
function that uses operator<.
set< type, compare O(n
> s(begin, end); Make a set and copy the values from begin to end.
log n)

Accessors

s.find(key)
Return an iterator pointing to an occurrence of key in s, or O(log
s.end() if key is not in s. n)
s.lower_bound(key)
Return an iterator pointing to the first occurrence of key in O(log
s, or s.end() if key is not in s. n)
Return an iterator pointing to the first occurrence of an
O(log
s.upper_bound(key) item greater than key in s, or s.end() if no such item is
n)
found.
O(log
s.equal_range(key) Returns a pair of lower_bound(key) and upper_bound(key).
n)

6 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

O(log
s.count(key) Returns the number of items equal to key in s.
n)
s.size() Return current number of elements. O(1)
s.empty() Return true if set is empty. O(1)
s.begin() Return an iterator pointing to the first element. O(1)
s.end() Return an iterator pointing one past the last element. O(1)

Modifiers

Inserts key into s. iterator is taken as a "hint" but key will

s.insert(iterator, O(log
key) go in the correct position no matter what. Returns an
n)
iterator pointing to where key went.
Inserts key into s and returns a pair p where p.first is an
s.insert(key)
iterator pointing to where key was stored, and p.second is O(log
true if key was actually inserted, i.e., was not already in n)
the set.

Map and Multimap

Maps can be thought of as generalized vectors. They allow map[key] = value for any
kind of key, not just integers. Maps are often called associative tables in other
languages, and are incredibly useful. They're even useful when the keys are integers,
if you have very sparse arrays, i.e., arrays where almost all elements are one value,
usually 0.

Maps are implemented as sets of pairs of keys and values. The pairs are sorted based
on the keys. Thus, they provide logarithmic storage and retrieval times, but require a
comparison predicate for the keys. operator<() will be used by default if none is
specified a construction time.

Map types are a bit complicated because of the pairs, so it's best to use typedef to
create more readable type names, like this:
typedef map<string, double> ValueMap;
typedef ValueMap::value_type VMPair;
typedef ValueMap::iterator VMIterator;

Definitions like the above will make find() and insert() a lot simpler:
ValueMap vm;
vm[ "abc" ] = 2.0;
vm[ "def" ] = 3.2;
vm.insert( VMPair( "ghi", 6.7 ) );

VMIterator iter = vm.find( "def" );

if ( iter != vm.end() ) {
cout << "Value of " << iter->first " << " is " << iter->second << endl;
}

You can just use map[key] to get the value directly without an iterator.

Warning: map[key] creates a dummy entry for key if one wasn't in the map

7 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

before. Sometimes, that's just what you want. When it isn't, use find().

multimaps are like map except that they allow duplicate keys. map[key] is not defined for
multimaps. Instead you must use insert() to add entry pairs, and find(), or
lower_bound() and upper_bound(), or equal_range() to retrieve entry pairs.

Header
#include <map>

Constructors

map< key_type, Make an empty map. key_compare should be a binary

value_type, key_compare predicate for ordering the keys. It's optional and will O(1)
> m;
default to a function that uses operator<.
map< key_type,
O(n
value_type, key_compare Make a map and copy the values from begin to end.
> m(begin, end); log n)

Accessors

m[key]
Return the value stored for key. This adds a default value if O(log
key not in map. n)
m.find(key)
Return an iterator pointing to a key-value pair, or m.end() O(log
if key is not in map. n)

m.lower_bound(key)
Return an iterator pointing to the first pair containing key, O(log
or m.end() if key is not in map. n)
m.upper_bound(key)
Return an iterator pointing one past the last pair O(log
containing key, or m.end() if key is not in map. n)
Return a pair containing the lower and upper bounds for
O(log
m.equal_range(key) key. This may be more efficient than calling those
n)
functions separately.
m.size() Return current number of elements. O(1)
m.empty() Return true if map is empty. O(1)
m.begin() Return an iterator pointing to the first pair. O(1)
m.end() Return an iterator pointing one past the last pair. O(1)

Modifiers

O(log
m[key] = value Store value under key in map.
n)
m.insert(pair)
Inserts the <key, value> pair into the map. Equivalent to the O(log
above operation. n)

Pair
A pair is a bit like a Lisp CONS cell. It holds just two values. They can be different

8 of 9 3/28/2010 8:02 AM
EECS 211: STL Summary http://www.cs.northwestern.edu/academics/courses/211/html/stl-summary...

types. For simplicity, pairs are simple generic struct's with two public data members:
first and second and a simple constructor that takes the two values to store.

Header
#include <utility>

Constructors

pair< first_type, second_type > p( first, Makes a pair. Both values must be
second ); O(1)
given.
pair< first_type, second_type > p( pair ); Makes a pair from another pair. O(1)

Accessors

p.first Returns the first value in the pair. O(1)

p.second Returns the second value in the pair. O(1)

Modifiers

There are no modifiers.

Comments? Send mail to [email protected].

9 of 9 3/28/2010 8:02 AM