[vset VERSION 1.1]
[manpage_begin struct::disjointset n [vset VERSION]]
[keywords {disjoint set}]
[keywords {equivalence class}]
[keywords find]
[keywords {merge find}]
[keywords partition]
[keywords {partitioned set}]
[keywords union]
[moddesc   {Tcl Data Structures}]
[titledesc {Disjoint set data structure}]
[category  {Data structures}]
[require Tcl 8.6]
[require struct::disjointset [opt [vset VERSION]]]
[description]
[para]

This package provides [term {disjoint sets}]. An alternative name for
this kind of structure is [term {merge-find}].

[para]

Normally when dealing with sets and their elements the question is "Is
this element E contained in this set S?", with both E and S known.

[para]

Here the question is "Which of several sets contains the element
E?". I.e. while the element is known, the set is not, and we wish to
find it quickly. It is not quite the inverse of the original question,
but close.

Another operation which is often wanted is that of quickly merging two
sets into one, with the result still fast for finding elements. Hence
the alternative term [term merge-find] for this.

[para]

Why now is this named a [term disjoint-set] ?

Because another way of describing the whole situation is that we have

[list_begin itemized]
[item] a finite [term set] S, containing
[item] a number of [term elements] E, split into
[item] a set of [term partitions] P. The latter term
       applies, because the intersection of each pair P, P' of
       partitions is empty, with the union of all partitions
       covering the whole set.
[item] An alternative name for the [term partitions] would be
       [term {equvalence classes}], and all elements in the same
       class are considered as equal.
[list_end]

Here is a pictorial representation of the concepts listed above:
[example {
	+-----------------+ The outer lines are the boundaries of the set S.
	|           /     | The inner regions delineated by the skewed lines
	|  *       /   *  | are the partitions P. The *'s denote the elements
	|      *  / \     | E in the set, each in a single partition, their
	|*       /   \    | equivalence class.
	|       /  *  \   |
	|      / *   /    |
	| *   /\  * /     |
	|    /  \  /      |
	|   /    \/  *    |
	|  / *    \       |
	| /     *  \      |
	+-----------------+
}]

[para]

For more information see [uri http://en.wikipedia.org/wiki/Disjoint_set_data_structure].

[section API]

The package exports a single command, [cmd ::struct::disjointset]. All
functionality provided here can be reached through a subcommand of
this command.

[para]

[list_begin definitions]

[call [cmd ::struct::disjointset] [arg disjointsetName]]

Creates a new disjoint set object with an associated global Tcl
command whose name is [emph disjointsetName]. This command may be used
to invoke various operations on the disjointset. It has the following
general form:

[list_begin definitions]

[call [arg disjointsetName] [arg option] [opt [arg {arg arg ...}]]]

The [cmd option] and the [arg arg]s determine the exact behavior of
the command. The following commands are possible for disjointset
objects:

[list_end]

[call [arg disjointsetName] [method add-element] [arg item]]

Creates a new partition in the specified disjoint set, and fills it
with the single item [arg item].  The command maintains
the integrity of the disjoint set, i.e. it verifies that none of the
[arg elements] are already part of the disjoint set and throws an
error otherwise.

[para]

The result of this method is the empty string.

[para]

This method runs in constant time.

[call [arg disjointsetName] [method add-partition] [arg elements]]

Creates a new partition in specified disjoint set, and fills it with
the values found in the set of [arg elements]. The command maintains
the integrity of the disjoint set, i.e. it verifies that none of the
[arg elements] are already part of the disjoint set and throws an
error otherwise.

[para]

The result of the command is the empty string.

[para]

This method runs in time proportional to the size of [arg elements]].


[call [arg disjointsetName] [method partitions]]

Returns the set of partitions the named disjoint set currently
consists of. The form of the result is a list of lists; the inner
lists contain the elements of the partitions.

[para]

This method runs in time O(N*alpha(N)),
where N is the number of elements in the disjoint set and alpha
is the inverse Ackermann function.

[call [arg disjointsetName] [method num-partitions]]

Returns the number of partitions the named disjoint set currently
consists of.

[para]

This method runs in constant time.

[call [arg disjointsetName] [method equal] [arg a] [arg b]]

Determines if the two elements [arg a] and [arg b] of the disjoint set
belong to the same partition. The result of the method is a boolean
value, [const True] if the two elements are contained in the same
partition, and [const False] otherwise.

[para]

An error will be thrown if either [arg a] or [arg b] are not elements
of the disjoint set.

[para]

This method runs in amortized time O(alpha(N)), where N is the number of
elements in the larger partition and alpha is the inverse Ackermann function.

[call [arg disjointsetName] [method merge] [arg a] [arg b]]

Determines the partitions the elements [arg a] and [arg b] are
contained in and merges them into a single partition.  If the two
elements were already contained in the same partition nothing will
change.

[para]

The result of the method is the empty string.

[para]

This method runs in amortized time O(alpha(N)), where N is the number of
items in the larger of the partitions being merged. The worst case time
is O(N).

[call [arg disjointsetName] [method find] [arg e]]

Returns a list of the members of the partition of the disjoint set
which contains the element
[arg e].

[para]

This method runs in O(N*alpha(N)) time, where N is the total number of
items in the disjoint set and alpha is the inverse Ackermann function,
See [method find-exemplar] for a faster method, if all that is needed
is a unique identifier for the partition, rather than an enumeration
of all its elements.

[call [arg disjointsetName] [method exemplars]]

Returns a list containing an exemplar of each partition in the disjoint
set. The exemplar is a member of the partition, chosen arbitrarily.

[para]

This method runs in O(N*alpha(N)) time, where N is the total number of items
in the disjoint set and alpha is the inverse Ackermann function.

[call [arg disjointsetName] [method find-exemplar] [arg e]]

Returns the exemplar of the partition of the disjoint set containing
the element [arg e].  Throws an error if [arg e] is not found in the
disjoint set.  The exemplar is an arbitrarily chosen member of the partition.
The only operation that will change the exemplar of any partition is
[method merge].

[para]

This method runs in O(alpha(N)) time, where N is the number of items in
the partition containing E, and alpha is the inverse Ackermann function.

[call [arg disjointsetName] [method destroy]]

Destroys the disjoint set object and all associated memory.

[list_end]

[vset CATEGORY {struct :: disjointset}]
[include ../common-text/feedback.inc]
[manpage_end]