Kruskal's Algorithm: Difference between revisions

From Rice Wiki
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
{{Infobox Algorithm|runtime=Sort + E cycle? + (V - 1) adding-edge|class=[[Minimum Spanning Tree]], [[Greedy Algorithm]]}}
{{Infobox Algorithm|runtime=O(E log E)|class=[[Minimum Spanning Tree]], [[Greedy Algorithm]]}}
 
= Approach: Greedy =
= Approach: Greedy =
The approach is to try to add the smallest edges as long as they do not create a cycle. Unlike [[Prims Algorithm|Prim's algorithm]], which prevents cycles by only choosing edges that crosses a cut of nodes already in the tree and nodes that aren't, Kruskal prevents cycles using a data structure known as disjoint set (aka. union-find).
The approach is to try to add the smallest edges as long as they do not create a cycle. Unlike [[Prims Algorithm|Prim's algorithm]], which prevents cycles by only choosing edges that crosses a cut of nodes already in the tree and nodes that aren't, Kruskal prevents cycles using a data structure known as disjoint set (aka. union-find).

Latest revision as of 18:06, 20 March 2024

Approach: Greedy

The approach is to try to add the smallest edges as long as they do not create a cycle. Unlike Prim's algorithm, which prevents cycles by only choosing edges that crosses a cut of nodes already in the tree and nodes that aren't, Kruskal prevents cycles using a data structure known as disjoint set (aka. union-find).

Given the MST of , the MST of should be that of plus the edge that connects to that is the shortest.

Implementation

Kruskal(G):
    let A be an empty graph
    for each v in V:
        makeSet(v)
    sort E in nondecreasing order by weight

    for each (u,v) in E:
        if findSet(u) != findSet(v):
            A = A U {(u,v)}
            Union(u,v)
    return A

Analysis

Sort edges + E (cycle?) + (V - 1) adding edge

Disjoint set allows O(log V) cycle checking and O(log V) edge adding.

Sorting takes E log E

For weighted disjoint set, checking cycle takes log V, and adding edge takes log V

For fast-find, where all members have the same ID, fast-set-id needs O(1) and union needs O(n)