Skip to content
This repository has been archived by the owner on Oct 8, 2021. It is now read-only.

Improve Kruskal and use in-built disjoint set data structure #896

Merged
merged 4 commits into from Jun 4, 2018

Conversation

SohamTamba
Copy link
Contributor

Improvements due to sorting:

For a complete graph with 1,000 vertices, the run-time improved from 193 ms to 28 ms.
For a path graph with 10,000 vertices, the run-time improved from 3.88 ms to 1.15 ms.

I removed the disjoint set data structure I had created and replaced it with the Inbuilt Disjoint Set. because the improvement in performance was little (28ms to to 26ms)

@codecov
Copy link

codecov bot commented Jun 1, 2018

Codecov Report

Merging #896 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #896      +/-   ##
==========================================
- Coverage   99.69%   99.69%   -0.01%     
==========================================
  Files          60       60              
  Lines        2279     2269      -10     
==========================================
- Hits         2272     2262      -10     
  Misses          7        7

@sbromberger
Copy link
Owner

@SohamTamba - thanks for this!
@somil55 - for your review / decision, please. If you say OK, then I'll be happy to merge.

@dsrivastavv
Copy link
Contributor

@sbromberger Will read through this by tonight.


Return a vector of edges representing the minimum spanning tree of a connected, undirected graph `g` with optional
distance matrix `distmx` using [Kruskal's algorithm](https://en.wikipedia.org/wiki/Kruskal%27s_algorithm).
"""
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add doc string

for e in edge_list[sortperm(weights)]
if !DataStructures.in_same_set(connected_vs, e.src, e.dst)
DataStructures.union!(connected_vs, e.src, e.dst)
push!(mst, e)
if length(mst) >= nv(g) - 1
break
end
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Put a shorthand notation: (length(mst)>= nv(g)-1 ) && break

Changes requested by @somil55
@dsrivastavv
Copy link
Contributor

@sbromberger Please merge this!

@sbromberger sbromberger merged commit c7406f1 into sbromberger:master Jun 4, 2018
matbesancon added a commit to matbesancon/LightGraphs.jl that referenced this pull request Jun 5, 2018
* removed flow algorithms (sbromberger#815)

* fixes sbromberger#820, plus tests (sbromberger#821)

* change show() for empty graphs (ref JuliaGraphs/MetaGraphs.jl#20 (comment)) (sbromberger#822)

* Pull request clique_percolation.jl (sbromberger#826)

clique percolation is a method of uncovering the overlapping community structure of complex networks in nature and society

* add src/community/clique_percolation.jl
* tests in file test/community/clique_percolation.jl
* cites the original clique percolation paper
* for undirected graphs only using traitfn

* in_ / out_ -> in / out (sbromberger#830)

* (in,out)_neighbors -> (in,out)neighbors

* all_neighbors -> allneighbors

* Pull request clique_percolation.jl (sbromberger#826)

clique percolation is a method of uncovering the overlapping community structure of complex networks in nature and society

* add src/community/clique_percolation.jl
* tests in file test/community/clique_percolation.jl
* cites the original clique percolation paper
* for undirected graphs only using traitfn

* revert allneighbors

* expected_degree_graph (Chung-Lu model) (sbromberger#832)

* Expected degree random graph generator implemented, including tests

* algorithm corrected

* Missing seed corrected in expected_degree_graph

* expected_degree_graph! implemented

* Added return in function, comment with references removed, references in docs added (expected_degree_graph)

* Update randgraphs.jl

minor doc update

* Update randgraphs.jl

* Fixing problems with MST Kruskal's on SimpleWeightedGraphs (sbromberger#835)

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update prim.jl

* reverting changes

* Revert "reverting changes"

This reverts commit ac1760b.

* Revert "Update prim.jl"

This reverts commit 677f6fa.

* Revert "Update kruskal.jl"

This reverts commit a0e9c47.

* Revert "Update prim.jl"

This reverts commit 793bac4.

* Revert "Update kruskal.jl"

This reverts commit 6114e16.

* Revert "Update prim.jl"

This reverts commit 551f1e6.

* Revert "Update kruskal.jl"

This reverts commit 941005e.

* Revert "Update kruskal.jl"

This reverts commit a404514.

* Revert "Update prim.jl"

This reverts commit 2d43a60.

* Revert "Update kruskal.jl"

This reverts commit 4577920.

* fix problems with SimpleWeightedGraphs

* fix problems with SimpleWeightedGraphs

* fix problems with SimpleWeightedGraphs

* bipartite_map on 2-order graphs fixed. Added proper tests. Fixed test connected to bipartite_map (sbromberger#836)

* Correct pre-allocation of memory in Prim's MST (sbromberger#839)

* Improve Kruskal's MST by optimizing Union-Find (sbromberger#843)

* add missing backtick (sbromberger#846)

* Add greedy_color for Graph Coloring (sbromberger#844)

* Add greedy_color for Graph Coloring

* Improve Kruskal's MST by optimizing Union-Find (sbromberger#843)

* Update README.md

* Update README.md

* first cut at 0.7 compatibility (sbromberger#848)

* using LightGraphs does not error

* Switch to LinearAlgebra and SparseArrays std libs

* Fix most of linalg tests

* Add SharedArrays for distance tests to compile

* Add Random and Markdown to stdlibs used

* Fix connectivity tests

* IntSet -> BitSet

* Add DelimitedFiles stdlib for readcsv

* Fix failing test

* first cut

* Use mauro/SimpleTraits.jl/m3/where-fns in tests

* Fix SimpleTraits checkout (sbromberger#851)

* Move up SimpleTraits checkout (sbromberger#852)

* Update runtests.jl

* Update REQUIRE

* Update REQUIRE

* femtocleaner with additional depwarn fixes (sbromberger#856)

fix deprecation warnings based on local femtocleaner run

* use equalto in degeneracy.jl (sbromberger#858)

* fix depwarns in linalg submodule (sbromberger#860)

* update linalg/spectral to fix deprecations

* fix depwarns in graphmatrices

* fixes doc deprecation warnings (sbromberger#861)

* fixes doc deprecation warnings

* adding Base64 to runtests

* Update README.md

* remove add/remove vertex/edge from core, minor bug fix (sbromberger#862)

* remove add/remove vertex/edge from core, minor bug fix

* fix tests

* export add/rem vertex

* remove long-term deprecation warnings (sbromberger#863)

* uninitialized -> undef, blkdiag -> blockdiag, and removed import of d… (sbromberger#865)

* uninitialized -> undef, blkdiag -> blockdiag, and removed import of deprecated functions from LG to LinAlg

* test coverage for digraph eltype

* removes equalto (sbromberger#867)

* optional sorting algorithm for gdistances (sbromberger#868)

add the ability to pass RadixSort to gdistances!

* update url and mention directed graphs explicilty (sbromberger#837)

* update url and mention directed graphs explicilty

* Update graphtypes.md

* Update graphtypes.md

fixed references.

* Speed improvements for function transitiveclosure! (sbromberger#870)

* Speed improvements for function transitiveclosure!

Instead of checking for all paths i -> k and k -> j for a given vertex k
we only iterate over the in- and outneighbours of k.

* Merged some conditionals into a single statement

* Cache efficient Floyd Warshall (sbromberger#873)

* Update floyd-warshall.jl

* Cache efficient Floyd Warshall

* fixed an error where smallgraph(:frucht) had 20 vertices instead of 12 (sbromberger#878)

* Delete .Rhistory

* Added function transitivereduction (sbromberger#877)

* added function transitivereduction

* Update transitivity.jl

docstring formatting

* Fixed some tests && added testdigraphs for all tests

* Johnson Shortest Path for Sparse Graphs (sbromberger#884)

* Johnson Shortest Path for Sparse Graphs

Johnson Shortest Path for Sparse Graphs

* Improved memory efficiency if distmx is mutable

* Improved memory efficiency for parallel implementation

* Update index.md

* Added constructors to create graphs from a vector or an iterator of edges (sbromberger#880)

* Added constructors to create SimpleGraphs and SimpleDiGraphs from a vector or an iterator of edges

* Added constructors to create SimpleGraphs and SimpleDiGraphs from a vector or an iterator of edges

* Slyles1001/892 (sbromberger#894)

* comments are your friend

* Most of LightGraphs warnings are fixed

* Delete HITS.jl

* Slyles1001/872 (sbromberger#891)

* DataStructures fixed

* missed heappop!, now it tests clean

* spaces

* Update LightGraphs.jl

* Update runtests.jl

* fixes most depwarns as of 20180529 (sbromberger#895)

* fixes most depwarns as of 20180529

* graphmatrices problems

* remove tabs

* tabs, again

* Update CONTRIBUTING.md (sbromberger#897)

* Improve Kruskal and use in-built disjoint set data structure (sbromberger#896)

* Improve Kruskal and use in-built disjoint set data structure

* Update kruskal.jl

Changes requested by @somil55
sbromberger pushed a commit that referenced this pull request Jun 5, 2018
* Improve Kruskal and use in-built disjoint set data structure

* Update kruskal.jl

Changes requested by @somil55
sbromberger pushed a commit that referenced this pull request Jun 14, 2018
* ignore coverage file

* merge master (#2)

* removed flow algorithms (#815)

* fixes #820, plus tests (#821)

* change show() for empty graphs (ref JuliaGraphs/MetaGraphs.jl#20 (comment)) (#822)

* Pull request clique_percolation.jl (#826)

clique percolation is a method of uncovering the overlapping community structure of complex networks in nature and society

* add src/community/clique_percolation.jl
* tests in file test/community/clique_percolation.jl
* cites the original clique percolation paper
* for undirected graphs only using traitfn

* in_ / out_ -> in / out (#830)

* (in,out)_neighbors -> (in,out)neighbors

* all_neighbors -> allneighbors

* Pull request clique_percolation.jl (#826)

clique percolation is a method of uncovering the overlapping community structure of complex networks in nature and society

* add src/community/clique_percolation.jl
* tests in file test/community/clique_percolation.jl
* cites the original clique percolation paper
* for undirected graphs only using traitfn

* revert allneighbors

* expected_degree_graph (Chung-Lu model) (#832)

* Expected degree random graph generator implemented, including tests

* algorithm corrected

* Missing seed corrected in expected_degree_graph

* expected_degree_graph! implemented

* Added return in function, comment with references removed, references in docs added (expected_degree_graph)

* Update randgraphs.jl

minor doc update

* Update randgraphs.jl

* Fixing problems with MST Kruskal's on SimpleWeightedGraphs (#835)

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update prim.jl

* Update kruskal.jl

* Update prim.jl

* reverting changes

* Revert "reverting changes"

This reverts commit ac1760b.

* Revert "Update prim.jl"

This reverts commit 677f6fa.

* Revert "Update kruskal.jl"

This reverts commit a0e9c47.

* Revert "Update prim.jl"

This reverts commit 793bac4.

* Revert "Update kruskal.jl"

This reverts commit 6114e16.

* Revert "Update prim.jl"

This reverts commit 551f1e6.

* Revert "Update kruskal.jl"

This reverts commit 941005e.

* Revert "Update kruskal.jl"

This reverts commit a404514.

* Revert "Update prim.jl"

This reverts commit 2d43a60.

* Revert "Update kruskal.jl"

This reverts commit 4577920.

* fix problems with SimpleWeightedGraphs

* fix problems with SimpleWeightedGraphs

* fix problems with SimpleWeightedGraphs

* bipartite_map on 2-order graphs fixed. Added proper tests. Fixed test connected to bipartite_map (#836)

* Correct pre-allocation of memory in Prim's MST (#839)

* Improve Kruskal's MST by optimizing Union-Find (#843)

* add missing backtick (#846)

* Add greedy_color for Graph Coloring (#844)

* Add greedy_color for Graph Coloring

* Improve Kruskal's MST by optimizing Union-Find (#843)

* Update README.md

* Update README.md

* first cut at 0.7 compatibility (#848)

* using LightGraphs does not error

* Switch to LinearAlgebra and SparseArrays std libs

* Fix most of linalg tests

* Add SharedArrays for distance tests to compile

* Add Random and Markdown to stdlibs used

* Fix connectivity tests

* IntSet -> BitSet

* Add DelimitedFiles stdlib for readcsv

* Fix failing test

* first cut

* Use mauro/SimpleTraits.jl/m3/where-fns in tests

* Fix SimpleTraits checkout (#851)

* Move up SimpleTraits checkout (#852)

* Update runtests.jl

* Update REQUIRE

* Update REQUIRE

* femtocleaner with additional depwarn fixes (#856)

fix deprecation warnings based on local femtocleaner run

* use equalto in degeneracy.jl (#858)

* fix depwarns in linalg submodule (#860)

* update linalg/spectral to fix deprecations

* fix depwarns in graphmatrices

* fixes doc deprecation warnings (#861)

* fixes doc deprecation warnings

* adding Base64 to runtests

* Update README.md

* remove add/remove vertex/edge from core, minor bug fix (#862)

* remove add/remove vertex/edge from core, minor bug fix

* fix tests

* export add/rem vertex

* remove long-term deprecation warnings (#863)

* uninitialized -> undef, blkdiag -> blockdiag, and removed import of d… (#865)

* uninitialized -> undef, blkdiag -> blockdiag, and removed import of deprecated functions from LG to LinAlg

* test coverage for digraph eltype

* removes equalto (#867)

* optional sorting algorithm for gdistances (#868)

add the ability to pass RadixSort to gdistances!

* update url and mention directed graphs explicilty (#837)

* update url and mention directed graphs explicilty

* Update graphtypes.md

* Update graphtypes.md

fixed references.

* Speed improvements for function transitiveclosure! (#870)

* Speed improvements for function transitiveclosure!

Instead of checking for all paths i -> k and k -> j for a given vertex k
we only iterate over the in- and outneighbours of k.

* Merged some conditionals into a single statement

* Cache efficient Floyd Warshall (#873)

* Update floyd-warshall.jl

* Cache efficient Floyd Warshall

* fixed an error where smallgraph(:frucht) had 20 vertices instead of 12 (#878)

* Delete .Rhistory

* Added function transitivereduction (#877)

* added function transitivereduction

* Update transitivity.jl

docstring formatting

* Fixed some tests && added testdigraphs for all tests

* Johnson Shortest Path for Sparse Graphs (#884)

* Johnson Shortest Path for Sparse Graphs

Johnson Shortest Path for Sparse Graphs

* Improved memory efficiency if distmx is mutable

* Improved memory efficiency for parallel implementation

* Update index.md

* Added constructors to create graphs from a vector or an iterator of edges (#880)

* Added constructors to create SimpleGraphs and SimpleDiGraphs from a vector or an iterator of edges

* Added constructors to create SimpleGraphs and SimpleDiGraphs from a vector or an iterator of edges

* Slyles1001/892 (#894)

* comments are your friend

* Most of LightGraphs warnings are fixed

* Delete HITS.jl

* Slyles1001/872 (#891)

* DataStructures fixed

* missed heappop!, now it tests clean

* spaces

* Update LightGraphs.jl

* Update runtests.jl

* fixes most depwarns as of 20180529 (#895)

* fixes most depwarns as of 20180529

* graphmatrices problems

* remove tabs

* tabs, again

* Update CONTRIBUTING.md (#897)

* Improve Kruskal and use in-built disjoint set data structure (#896)

* Improve Kruskal and use in-built disjoint set data structure

* Update kruskal.jl

Changes requested by @somil55

* updated syntax for iterator protocol

* fixed iterator, broken inference
SohamTamba added a commit to SohamTamba/LightGraphs.jl that referenced this pull request Jun 25, 2018
…rger#896)

* Improve Kruskal and use in-built disjoint set data structure

* Update kruskal.jl

Changes requested by @somil55
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants