See https://github.com/igraph/rigraph/blob/3299d31/src/vendor/cigraph/CHANGELOG.md for a complete changelog of the bundled C core, and https://github.com/igraph/rigraph/compare/f3fa58b..3299d31#diff-aeb78e0159780a9b26daabaf6f95f450b0cfec7161fc735f27ad69145a57dc84 for the changes since the igraph 2.0.1. (A permanent link to the most recent changelog of the C core used in the R package is https://github.com/igraph/rigraph/blob/main/src/vendor/cigraph/CHANGELOG.md.)
- GMP is no longer a dependency (#1256), libxml2 (#1282) and glpk are optional.
- Update vendored sources to igraph/igraph@857a125069c226f266562b3781b82833fe1d59d9.
- New
voronoi_cells()
to compute the Voronoi partitioning of a graph (#1173).
- Fix
rglplot(edge_label = )
(#1267).
- Run examples with sanitizer (#1288).
- Add scheduled builds.
- Make
x11()
usage in example happen only in interactive sessions (#1301). - Remove misleading comment about warning given as no warning is given (#1294).
- Improve
min_st_separators()
documentation (#1264). - Add link to discussion forum (#1279) and logo (#1280).
- Add code finding duplicate
@seealso
, and use it (#1270). - Remove duplicate
@seealso
from?sample_pa
(#1268). - Remove incorrect claim about handling of complete graphs by
is_separator()
andis_min_separator()
(#1263). - Fix error messages mentioning to mention
upgrade_graph()
instead of the nonexistingupgrade_version()
(#1252). - Split
is_bipartite()
manual page from other manual page (#1230). - Improve bug report template.
CITATION.cff
contains only the first three version components.- Install lock workflow to lock stale discussions after one year (#1304).
- Replace use of deprecated
barabasi.game()
withsample_pa()
(#1291). - Auto-generate
are_adjacent()
, avoid deprecatedigraph_are_connected()
C function (#1254).
See https://github.com/igraph/rigraph/blob/f3fa58b/src/vendor/cigraph/CHANGELOG.md for a complete changelog of the bundled C core, and https://github.com/igraph/rigraph/compare/1bd2bf79..f3fa58b#diff-aeb78e0159780a9b26daabaf6f95f450b0cfec7161fc735f27ad69145a57dc84 for the changes since the igraph 2.0.1. (A permanent link to the most recent changelog of the C core used in the R package is https://github.com/igraph/rigraph/blob/main/src/vendor/cigraph/CHANGELOG.md.)
g + vertices(1, 2, foo = 3)
works again, regression introduced in igraph 2.0.0 (#1247).sample_pa()
respects theout.seq
andout.dist
arguments again, regression introduced in igraph 2.0.0 (#1226).isomorphisms()
andsubgraph_isomorphisims(method = "vf2")
work again, regression introduced in 2.0.0 (#1219).biconnected_components()
now returns edge and vertex sequences again, regression introduced in 2.0.0 (#1213).- Remove zeros from the
order
andorder.out
components returned bydfs()
, regression introduced in 2.0.0 (#1179). - Memory leaks when converting data to C (#1196).
realize_bipartite_degseq()
creates a bipartite graph from two degree sequences (#1212).is_biconnected()
checks if a graph is biconnected (#1204).distances()
now supports the Floyd-Warshall algorithm (#1186).
- Use more culturally diverse names in intro vignettes (#1246).
- Formatting tweaks in introductory vignettes (#1243).
- Recommend {pak} instead of {remotes} (#1228).
- Fix typo in
mean_distance()
docs. - Update troubleshooting document, emphasize issues with Anaconda environments (#1209).
- Improved docs for shortest path functions (#1201).
- Document
"dsatur"
heuristic forgreedy_vertex_coloring()
(#1206). - Remove scg related docs (#1167).
- Fix typo in
?articulation_points
(#1191). - Improve installation and troubleshooting instructions (#1184).
- Improve docs of assortativity (#1151).
- Add tests for
isomorphisms()
andsubgraph_isomorphisms()
(#1225).
- Always use bundled mini-gmp (#1233).
config.h
definesHAVE___UINT128_T
(#1216).- Do not rely on
which
program during configuration (#1232). configure
manage libxml multiple include paths (#1197).- Remove empty string in
configure
(#1235). - Link Fortran runtime on Windows as needed by arpack. (#1215).
- Workaround for deprecated enum values not being supported with old GCC (#1205).
HAVE_GFORTRAN
flag forwin
anducrt
(#1171).make_empty_graph()
is now fully auto-generated (#1068).- Eliminate manual edits from autogenerated files (#1207).
- Add read-only comments for RStudio IDE (#1152).
- Remove unused patch files (#1234).
- Update stimulus to 0.21.4 (#1210).
- Avoid duplicate objects (#1223).
- Eliminate a compiler warning from simpleraytracer (#1185).
- Avoid
is.R()
which is deprecated in R 4.4.0.
- Use cpp11 for simpleraytracer glue to avoid
-Wlto
warnings (#1163).
This major release brings development in line with the C core at https://github.com/igraph/igraph. See https://github.com/igraph/rigraph/blob/1bd2bf79/src/vendor/cigraph/CHANGELOG.md for a complete changelog, in particular the section "Breaking changes". (A permanent link to the most recent changelog of the C core used in the R package is https://github.com/igraph/rigraph/blob/main/src/vendor/cigraph/CHANGELOG.md.)
Version 1.6.0 of the R package used version 0.9.9 of the C core. The changes in the 0.10 series of the C core are relevant for version 2.0.0 of the R package, but not all C core functions are exposed in the R interface. Changes to the implementation of some functions in the C core may be reflected in functions in the R package.
The change log below is a summary of the additional changes in the R interface.
-
get.edge.ids(multi = TRUE)
andas_adjacency_matrix(edges = )
are no longer supported (#1101, #1080). -
Remove
NA
padding fordfs(unreachable = FALSE)$order
and$order.out
and forbfs(unreachable = FALSE)$order
, also forigraph_options(return.vs.es = FALSE)
(#1062, #1124). -
laplacian_matrix(normalized = TRUE)
uses a different normalization method (#1102). -
fit_power_law()
no longer returns aKS.p
component. -
Remove default value for
loops
argument ofcentr_degree_tmax()
(#1114). -
as_adjacency_matrix()
no longer supports attributes of typecharacter
(#1072). -
Remove
graph.isomorphic.34()
(#1065, #1067). -
Use
lifecycle::deprecate_soft()
for most deprecated functions (#1024, #1025, #1104). -
The system libraries GLPK, GMP, and libxml2 are now mandatory for compiling from source.
-
Add scalar conversion checks in a few critical places (#1069).
-
Check that we receive a scalar when expecting a scalar in C code (#1051).
-
Check that matrix sizes are in the supported range before passing them to R (#1066).
-
transitivity()
now produces a named vector in the local (weighted) case (#1057). -
plot()
correctly computes intersections between edges and rectangle vertices (#1021). -
Fix compilation on Windows with spaces in the Rtools path (#1000).
-
Use 30 bits of R's RNG (#1079).
-
Breaking change: Remove
NA
padding fordfs(unreachable = FALSE)$order
and$order.out
and forbfs(unreachable = FALSE)$order
, also forigraph_options(return.vs.es = FALSE)
(#1062, #1124). -
New
k_shortest_paths()
to compute the first k shortest paths between two vertices (#1028).
-
Add GitHub bug report form.
-
New
vignette("deprecated-dot-case")
that lists deprecated functions and their alternatives (#1013). -
Move deprecation badge to top of doc pages.
-
Remove usage of
get.adjacency()
from intro vignettes (#1084). -
Fix math formatting in
fit_power_law()
docs. -
Fix incorrect usage in
subgraph.edges()
example. -
Clarify that Infomap considers edge directions.
-
Improve edge connectivity docs (#1119).
-
Add some missing diacritics.
-
Use
[]
assignment for converting matrices (#1035). -
Move towards autogenerating all bindings (#1043).
-
Use %||% from rlang (#1112).
-
Replace loop by
lapply()
when returning vertex and edge sets (#1033). -
Do not duplicate prototypes, use a common header for
rinterface.c
andrinterface_extra.c
(#1055). -
Clean up some auto-generation code (#1031), remove unused parts from code generation (#1032), eliminate the use of some deprecated C functions.
-
Use integers for mode enum in
cluster_label_prop()
. -
Standardize the handling of some stimulus enum types (#1064).
-
Remove unused vectorlist / matrixlist destructors (#1070).
-
Remove unused stimulus type VERTEXSETLIST_INT (#1049).
-
Remove unused types (#1060).
-
Fix Stimulus definitions (#997).
-
Avoid deprecated
R_igraph_automorphisms()
(#999). -
Use new ARPACK_DEFAULTS symbol from C core for default arpack options.
-
Ignore upstream CITATION file to avoid R CMD check NOTE (#1007).
-
Add Aviator configuration.
-
Add
as_biadjacency_matrix()
tests for named vectors (#1154, #1155). -
Test transferring colors in isomorphism functions (#1050).
-
New
arpack_defaults()
replacesigraph.arpack.default
andarpack_defaults
lists (#800), with consistent usage checks (#966). -
Define methods for
ape::as.phylo()
instead of ownas_phylo()
generic, remove deprecatedasPhylo()
(#763). -
bfs()
with unreachable nodes returns zero instead ofNaN
asrank
(#956), and negative instead ofNaN
fordist
(#926). -
random_walk()
adds one more step, the resulting list of vertices is of lengthsteps + 1
(#934). -
girth()
returnsInf
for acyclic graphs (#931). -
Remove some behaviour / parameters that were deprecated in 1.3 (#842).
-
Remove
scg_eps()
,scg_group()
andscg_semi_proj()
, broken and no longer maintained (#946).
-
Fix
printf()
warning (#975). -
Fix calls to
isSymmetric()
for matrix objects with a class (#965). -
Reduce frequency of interruption checks (#957).
-
Change "incidence matrix" naming to "biadjacency matrix" (#962).
-
Add aliases using the neighborhood terminology for all ego functions: (#950).
-
Rename arguments to
assortativity(values, values.in)
fromtypes1
andtypes2
(#952). -
Add ellipsis to
cluster_label_prop()
(#954). -
Placeholders for
is_acyclic()
andis_forest()
, to be implemented in igraph 2.0.0 (#945). -
is_tree(details = TRUE)
designates the first vertex as root for non-trees (#935). -
all_shortest_paths()
returns the vertex list in bothres
andvpaths
components (#930). -
Improve symmetry check for adjacency matrices on input (#911).
-
Warn if non-symmetric undirected adjacency matrix (#904).
-
Remove duplicate
\seealso{}
(#948). -
Update
knnk()
andsir()
docs. -
Fix formula in
reciprocity()
docs. -
strength()
does not give a warning when no weights are given. -
Update
sample_smallworld()
manual page (#895). -
Correct documentation for weighted cliques.
-
Update Erdos-Renyi model docs.
-
Clean up tests (#967).
-
Stabilize tests for
cluster_label_prop()
(#932),sample_pa()
(#927),cluster_louvain()
(#933), SIR (#936),sample_forestfire()
(#929). -
Hard-code graph used in test (#928).
-
Update GML test file (#925).
-
Improve test for sampling from Dirichlet distribution (#923).
-
Omit test that no longer holds with igraph 0.10 (#922).
-
Stabilize test in preparation for upgrade (#920).
-
Stabilize plot test (#919).
-
Fix checks.
- Breaking change: start deprecation of
estimate_betweenness()
,estimate_edge_betweenness()
,estimate_closeness()
(#852).
identical_graphs()
now correctly detects identical graphs without node or edge attributes (#757).
-
Change ownership rules of attribute objects (#870).
-
R_SEXP_to_igraph()
andR_SEXP_to_igraph_copy()
getigraph_t
object from external pointer (#865). -
Remove
ETIME()
call from Fortran code, already in CRAN version 1.5.0.1 (#858). -
Ensure that
_GNU_SOURCE
is always defined (#877). -
Fix
make clean
(#860). -
Fix generation of code for functions with
VERTEX_COLOR
out-arguments (#850). -
Use
-lquadmath
inMakevars.win
, for compatibility with alternative R distributions such as Microsoft R Open (#855). -
getRversion()
uses strings.
-
Add cffr file and a GHA workflow that updates it automatically (#873).
-
Undeprecate
neighborhood()
(#851). -
Remove redundant reference to AUTHORS from DESCRIPTION.
- Breaking change: start deprecation of
estimate_betweenness()
,estimate_edge_betweenness()
,estimate_closeness()
(#852).
The internal format of graph objects has changed in a mostly backward-compatible way, to prepare for upgrading the C core to 0.10. Details are described at https://github.com/igraph/rigraph/wiki/The-igraph-object-format. Accessing graph objects that have been created with an older igraph version give a clean error message with instructions (#832). The new format cannot be read by igraph 1.4.3 or older, the following error is raised when trying to do so:
This graph was created by an old(er) igraph version.
Call upgrade_graph() on it to use with the current igraph version
For now we convert it on the fly...
Error in is_directed(object) :
REAL() can only be applied to a 'numeric', not a 'NULL'
The only supported remedy is to upgrade the igraph package to version 1.5.0 or later.
graph_version()
now returns an integer scalar (#832, #847), 4
as of igraph 1.5.0 (#835).
-
Vertex and edge sequences are converted to numeric vectors when used in attributes (#808).
-
New
largest_component()
returns the largest connected component (#786, @ngmaclaren).
- Fix error message in
make_graph()
whensimplify = ...
is used with a non-formula (#834).
-
Add more tests for
graph_from_literal()
(#826). -
Reenable serialization test and tests for
dyad_census()
, stabilize tests (#809, #822, #823).
-
The documentation for the R package is now hosted at https://r.igraph.org/ (#780).
-
Update
vignette("installation-troubleshooting")
. -
Fix use of deprecated functions in examples, e.g., replace
gsize()
byecount()
(#827). -
Fix typos in
?eigen_centrality
docs (@JJ). -
Update CONTRIBUTING.md and ORCID information (#791, #774).
-
Add DOI to CITATION (#773).
-
Add data for old igraph versions as constructed objects, and tests (#838).
-
Ensure we're always using named indexes to access the internal data structure (#784).
-
Prepare migration to igraph/C 0.10 (#781).
-
Update generated interface (#765).
-
Fix tests for dev waldo (#779, @hadley).
-
Fix linking on Windows: gfortran needs quadmath. (#778).
- Mention limitation of Pajek reader (#776).
-
Remove
igraph.eigen.default()
andeigen_defaults
, introduce internaleigen_defaults()
as a function (#741). -
Remove broken
nexus*()
functions (#705), andsrand()
(#701).
-
Update C core.
-
ARPACK-based calculations are now interruptible.
-
shortest_paths()
andall_shortest_paths()
no longer crash when an invalidfrom
vertex is passed and weights are being used.
See diff for details.
-
Use true vertex names for printing vertex sets. If a vertex set captures a relationship between vertices (e.g., the
father
component ofbfs()
), the vertex set is printed as a named vector (#754). -
Suggest restarting R session after fatal error (#745).
-
as_long_data_frame()
now correctly processes vertex attributes and works with graphs without vertex attributes (#748). -
as.hclust(hrg.fit(g))
works again (#721).
-
The documentation is now available at https://igraph.org/ (#743).
-
Reorganize function reference (#662).
-
Replace
graph()
withmake_graph()
in examples (#738). -
Add docs for
as.hclust.igraphHRG()
(#733). -
Merged man page of
hub_score()
andauthority_score()
(#698). -
Refactor contributors listing (#647).
-
Improve "family" titles (#679).
-
Improve docs of ego/neighborhood functions.
-
Improve
transitivity()
docs.
-
Introduce cpp11 package to improve error handling in the long run (#720).
-
Avoid longjmp for error handling and interrupts (#751).
-
as.hclust.igraphHRG
uses.Call()
interface (#727).
-
console()
now works again and provides a Tcl/Tk based UI where igraph can post status messages and progress info (#664). -
Fix errors when printing long vertex names (#677, @ahmohamed).
-
Fix regression that broke builds on some systems (e.g., GCC version 5 or earlier), introduced in igraph 1.4.0 (#670, #671).
-
fit_hrg()
does not crash any more when called with a graph that has less than three vertices.
- Various improvements (#663, @maelle; #667).
-
Fix warning about
yyget_leng()
returning wrong type when using LTO (#676). -
Don't mention C++11 or C++17 for best compatibility with both newest R and older compilers, while still requesting a C++ compiler for linking.
-
Don't ignore
build/
when building the package because the vignette index is built there. -
Skip plot test entirely on R-devel.
-
Avoid submodules for building igraph (#674).
-
Makevars cleanup (#671).
-
Add Zenodo configuration file.
- Breaking change: Allow change of attribute type when setting attribute for all vertices or edges; only attributes of length 1 or the length of the target set allowed (#633).
tkplot()
gained apalette
argument and it is now using the same palette asplot()
by default, for sake of consistency.plot.igraph()
gained aloop.size
argument that can be used to scale the common radius of the loop edges.
- The default maximum number of iterations for ARPACK has been increased to 3000 to match that of the igraph C core.
- Rare convergence problems have been corrected in
cluster_leading_eigen()
. - All ARPACK-based functions now respect random seeds set in R when generating a random starting vector.
igraph_version()
returned an invalid value in 1.3.4, this is now corrected.- The value of
par(xpd=...)
is now restored after plotting a graph. - Fixed a bug in
as.dendrogram.communities()
for large dendrograms, thanks to @pkharchenko (see PR #292). - Fixed two bugs in
graph_from_incidence_matrix()
that prevented the creation of directed graphs withmode="all"
from dense or sparse matrices. dfs()
accidentally returned zero-based root vertex indices in the result object; this is now fixed and the indices are now 1-based.as_graphnel()
does not duplicate loop edges any more.as_graphnel()
now checks that the input graph has no multi-edges. Multi-edges are not supported by the graphNEL class.convex_hull()
now returns the vertices of the convex hull with 1-based indexing.- Some
rgl.*()
function calls in the codebase were replaced with equivalent*3d()
function calls in preparation for upcoming deprecations inrgl
(see PR #619) plot.igraph()
does not use theframe=...
partial argument any more when callingplot.default()
. The defaultNULL
value offrame.plot
is now also handled correctly.hub_score()
andauthority_score()
considered self-loops only once on the diagonal of the adjacency matrix of undirected graphs, thus the result was not identical to that obtained byeigen_centrality()
on loopy undirected graphs. This is now corrected.distances()
no longer ignores themode
parameter whenalgorithm='johnson'
.
automorphisms()
was renamed tocount_automorphisms()
; the old name is still available, but it is deprecated.
- Documentation improvements.
- The Github repository was now moved to a single-branch setup where the package can be built from the
main
branch directly. - Added igraph extended tutorial as an R vignette (#587).
- igraph now has a homepage based on
pkgdown
thanks to @maelle (see #645). This will eventually become the official homepage.
Added:
mark.groups=...
argument ofplot.igraph()
now acceptscommunities
objects
Fixed:
-
Negative degree exponents are not allowed any more in
sample_pa()
andsample_aging_pa()
. -
Package updated to be compatible with Matrix 1.5.
Other:
- Documentation improvements and fixes.
Added:
sample_asym_pref()
now returns the generated types of the vertices in the vertex attributes namedouttype
andintype
.
Fixed:
-
layout_nicely()
does not recurse infinitely any more if it is assigned to thelayout
attribute of a graph -
layout_nicely()
now ignores edge weights when there are non-positive edge weights. This is needed because igraph 1.3.3 started validating edge weights inlayout_with_fr()
andlayout_with_drl()
, resulting in errors whenlayout_nicely()
was used on weighted graphs with negative weights. Sincelayout_nicely()
is the default layout algorithm forplot()
, most users were not even aware that they were using the FR or DrL layouts behind the scenes. Now the policy is thatlayout_nicely()
attempts to get the job done without errors if possible, even if that means that edge weights must be ignored. A warning is printed if this is the case.
Added:
-
reverse_edges()
reverses specific or all edges in a graph. -
Single-bracket indexing of
V()
andE()
resolves attribute names in the indexing expressions by default (for instance,E(g)[weight > x]
matches edges with a weight larger than a threshold). This can be problematic if the attribute masks one of the variables in the local evaluation context. We now have a pronoun called.env
(similarly torlang::.env
) that allows you to force attribute name lookup to the calling environment. For sake of completeness, we also provide.data
(similarly torlang::.data
) to force attribute name lookup to the vertex / edge attributes only. These pronouns are automatically injected into the environment where the indexing expression is evaluated.
Deprecated:
- Names of functions that can be used inside a
V()
orE()
indexing start with a dot since igraph 1.1.1; however, the old dotless names did not print a deprecation warning so this may have gone unnoticed for years. We are introducting a deprecation warning fornei()
,innei()
,outnei()
,inc()
,from()
andto()
inside single-bracket indexing of vertex and edge sequences and will remove the old variants soon.
The C core is updated to 0.9.9, fixing a range of bugs.
Fixed:
- The length of size-zero
communities
objects is now reported correctly. layout_with_kk()
would fail to produce reasonable results with the default initial coordinates. This has been corrected, however, this function no longer produces precisely the same output for a given graph as before. To restore the previous behaviour, uselayout_with_kk(g, coord=layout_in_circle(g))
in 2D orlayout_with_kk(g, dim=3, coord=layout_on_sphere(g))
in 3D.- Indexing an
igraph.vs
object withv[x, na_ok=T]
now correctly handles thena_ok
argument in all cases; previous versions ignored it whenx
was a single number.
Other:
- Documentation improvements and fixes.
Fixed:
graph_from_adjacency_matrix()
now works with sparse matrices even if the cell values in the sparse matrix are unspecified.- Fixed crash in
cluster_walktrap()
whenmodularity=FALSE
andmembership=FALSE
. cluster_walktrap()
no longer accepts invalid weight vectors.cluster_walktrap()
no longer returns a modularity vector of invalid length for disconnected graphs. This also fixes some rare failures of this function on weighted disconnected graphs.edge_attr()
does not ignore itsindex=...
argument any more.automorphisms()
,automorphism_group()
andcanonical_permutation()
now allow all possible values supported by the C core in thesh
argument. Earlier versions supported only"fm"
.- The
vertex.frame.width
plotting parameter now allows zero and negative values; these will simply remove the outline of the corresponding vertex. - The documentation of the
sh
argument of the BLISS isomorphism algorithm inisomorphic()
was fixed; earlier versions incorrectly referred tosh1
andsh2
. dominator_tree()
now conforms to its documentation with respect to thedom
component of the result: it contains the indices of the dominator vertices for each vertex and-1
for the root of the dominator tree.- Mentions of the
"power"
algorithm ofpage_rank()
have been removed from the documentation, as this method is no longer available. - Several other documentation fixes to bring the docs up to date with new behaviours in igraph 1.3.
The C core is updated to 0.9.7, fixing a range of bugs and introducing a number of new functions.
Added:
has_eulerian_path()
andhas_eulerian_cycle()
decides whether there is an Eulerian path or cycle in the graph.eulerian_path()
andeulerian_cycle()
returns the edges and vertices in an Eulerian path or cycle in the graph.any_loop()
checks whether a graph contains at least one loop edge.is_tree()
checks whether a graph is a tree and also finds a possible rootto_prufer()
converts a tree graph into its Prufer sequencemake_from_prufer()
creates a tree graph from its Prufer sequencesample_tree()
to sample labelled trees uniformly at randomsample_spanning_tree()
to sample spanning trees of an undirected graph uniformly at randomautomorphisms()
andcanonical_permutation()
now supports vertex colorsrandom_edge_walk()
to record the edges traversed during a random walkharmonic_centrality()
calculates the harmonic centrality of vertices, optionally with a cutoff on path lengthsmean_distance()
now supports edge weights and it can also return the number of unconnected vertex pairs whendetails=TRUE
is passed as an argumentgreedy_vertex_coloring()
finds vertex colorings based on a simple greedy algorithm.bridges()
finds the bridges (cut-edges) of a graph- The frame width of circle, rectangle and square vertex shapes can now be
adjusted on plots with the
frame.width
vertex attribute or thevertex.frame.width
keyword argument, thanks to @simoncarrignon . See PR #500 for more details. automorphism_group()
returns a possible (not necessarily minimal) generating set of the automorphism group of a graph.global_efficiency()
calculates the global efficiency of the graph.local_efficiency()
calculates the local efficiency of each vertex in a graph.average_local_efficiency()
calculates the average local efficiency across the set of vertices in a graph.rewire(each_edge(...))
now supports rewiring only one endpoint of each edge.realize_degseq()
generates graphs from degree sequences in a deterministic manner. It is also available asmake_(degseq(..., deterministic=TRUE))
.clique_size_counts()
counts cliques of different sizes without storing them all.feedback_arc_set()
finds a minimum-weight feedback arc set in a graph, either with an exact integer programming algorithm or with a linear-time approximation.make_bipartite_graph()
now handles vertices with names.shortest_paths()
now supports graphs with negative edge weights.min_cut()
now supports s-t mincuts even ifvalue.only=FALSE
.as.matrix()
now supports converting an igraph graph to an adjacency or edge list matrix representation. Seeas.matrix.igraph()
for more details. This function was migrated fromintergraph
; thanks to Michal Bojanowski.
Fixed:
is_connected()
now returns FALSE for the null graph- Calling
length()
on a graph now returns the number of vertices to make it consistent with indexing the graph with[[
. diameter()
now corrently returns infinity for disconnected graphs whenunconnected=FALSE
. Previous versions returned the number of vertices plus one, which was clearly invalid for weighted graphs.mean_distance()
now correctly treats the path length between disconnected vertices as infinite whenunconnected=FALSE
. Previous versions used the number of vertices plus one, adding a bias towards this number, even if the graph was weighted and the number of vertices plus one was not a path length that could safely have been considered as being longer than any "valid" path.layout_with_sugiyama()
now handles the case of exactly one extra virtual node correctly; fixes #85bfs()
anddfs()
callback functions now correctly receive 1-based vertex indices and ranks; it used to be zero-based in earlier versions- Accidentally returning a non-logical value from a
bfs()
ordfs()
callback does not crash R any more - Calling
print()
on a graph with a smallmax.lines
value (smaller than the number of lines needed to print the attribute list and the header) does not raise an error any more; fixes #179 as_adjacency_matrix(edges=TRUE, sparse=TRUE)
now consistently returns the last edge ID for each cell in the matrix instead of summing them.- Using the
+
and-
operators with apath()
object consisting of two vertices is now handled correctly; fixes #355 topo_sort()
now throws an error if the input graph is not acyclic instead of returning an incorrect partial ordering.- Weighted transitivity calculations (i.e.
transitivity(mode="barrat")
now throw an error for multigraphs; the implementation does not work correctly for multigraphs and earlier versions did not warn about this.
Changed:
- The
neimode
argument ofbfs()
anddfs()
was renamed tomode
for sake of consistency with other functions. The old argument name is deprecated and will be removed in 1.4.0. bfs()
anddfs()
callback functions now correctly receive 1-based vertex indices and ranks; it used to be zero-based in earlier versions. (This is actually a bugfix so it's also mentioned in the "Fixed" section).closeness()
,betweenness()
andedge_betweenness()
now all take acutoff
argument on their own.estimate_closeness()
,estimate_betweenness()
andestimate_edge_betweenness()
became aliases, with identical signature. They are not deprecated but their implementation might change in future versions to provide proper estimation schemes instead of a simple cutoff-based approximation. If you explicitly need cutoffs and you want your results to be reproducible with future versions, usecloseness()
,betweenness()
andedge_betweenness()
in your code with acutoff
argument.closeness()
now only considers reachable vertices during the calculation; in other words, closeness centrality is now calculated on a per-component basis for disconnected graphs. Earlier versions considered all vertices.
Deprecated:
- Using
cutoff=0
forcloseness()
,betweenness()
andedge_betweenness()
is deprecated; if you want exact scores, use a negative cutoff.cutoff=0
will be interpreted literally from igraph 1.4.0. centr_degree_tmax()
now prints a warning when it is invoked without an explicitloops
argument.loops
will be mandatory from igraph 1.4.0.- The
nexus_list()
,nexus_info()
,nexus_get()
andnexus_search()
functions now return an error informing the user that the Nexus graph repository has been taken offline (actually, several years ago). These functions will be removed in 1.4.0. - The
edges
argument ofas_adjacency_matrix()
is deprecated; it will be removed in igraph 1.4.0.
Removed:
- The deprecated
page_rank_old()
function and the deprecatedpower
method ofpage_rank()
were removed.
Dec 27, 2021
No user visible changes.
Dec 14, 2021
Fixed:
- The macOS versions of
igraph
were accidentally built without GraphML support on CRAN; this should now be fixed.
Nov 22, 2021
No user visible changes.
Oct 26, 2021
No user visible changes.
Oct 15, 2021
The C core is updated to 0.8.5, fixing a range of bugs and introducing a number of new functions.
Added:
- cluster_leiden added (#399).
- cluster_fluid_communities added (#454)
Fixed:
make_lattice()
correctly roundslength
to the nearest integer while printing a warning (#115).make_empty_graph(NULL)
now prints an error instead of producing an invalid graph (#404).make_graph(c())
now produces an empty graph instead of printing a misleading error message (#431).- Printing a graph where some edges have NA as the names of both endpoints does not produce a misleading error message any more (#410).
- The
types
argument of functions related to bipartite graphs now prints a warning when the types are coerced to booleans (#476). - Betweenness normalisation no longer overflows (#442).
layout_with_sugiyama()
returns a layout of type matrix even if there is only one vertex in the graph (#408).- Plotting a null graph (i.e. a graph with no vertices) does not throw an error any more (#387).
Deprecated:
- The
membership
argument ofmodularity.matrix()
is now deprecated as the function never needed it anyway. modularity()
now prints a warning when it is applied on a directed graph because the implementation in igraph's C core does not support directed graphs as of version 0.8.5. The warning will be turned into an error in the next minor (1.3.0) version of the R interface; the error will be removed later when the C core is updated to a version that supports modularity for directed networks.transitivity()
now prints a warning when its local variant (type="local"
) is called on a directed graph or a graph with multiple edges beecause the implementation in the C core of igraph does not work reliably in these cases as of version 0.8.5. The warning will be turned into an error in the next minor (1.3.0) version of the R interface; the error will be removed later when the C core is updated to a version that supports transitivity for networks with multiple edges.
Misc:
- Documentation improvements.
Oct 5, 2020
No user visible changes.
Mar 27, 2020
No user visible changes.
Feb 13, 2019
No user visible changes.
Jan 27, 2019
No user visible changes.
Jul 27, 2018
No user visible changes.
- The GLPK library is optional, if it is not available, then the
cluster_optimal()
function does not work. Unfortunately we cannot bundle the GLPK library into igraph on CRAN any more, because CRAN maintainers forbid the pragmas in its source code. - Removed the NMF package dependency, and related functions.
- Fix compilation without libxml2
Jul 20, 2017
- Fix compilation on Solaris
Jul 13, 2017
- Graph id is printed in the header, and a
graph_id
function was added - Fix
edge_attr
for some index values - Fix a
bfs()
bug,restricted
argument was zero-based match_vertices
is exported now%>%
is re-exported in a better way, to avoid interference with other packagesego_
functions default toorder = 1
now- New function
igraph_with_opt
to run code with temporary igraph options settings - Fix broken
sample_asym_pref
function - Fix
curve_multiple
to avoid warnings for graphs with self-loops. - The
NMF
package is only suggested now, it is not a hard dependency - Fix gen_uid.c _SVID_SOURCE issues
- Avoid drawing straight lines as Bezier curves
- Use the
pkgconfig
package for options. This allows setting options on a per-package basis. E.g. a package using igraph can setreturn.vs.es
toFALSE
in its.onLoad()
function, and then igraph will return plain numeric vectors instead of vertex/edge sequences if called from this package. igraph_options()
returns the old values of the updated options, this is actually useful, returning the new values was not.with_igraph_opt()
function to temporarily change values of igraph options.get.edge()
is deprecated, useends()
instead. (This was already the case for igraph 1.0.0, but we forgot to add a NEWS point for it.)- Do not redefine
.Call()
, to make native calls faster. - Speed up special cases of indexing vertex sequences.
- Removed an
anyNA()
call, to be compatible with older R versions. - Fixed a fast-greedy community finding bug, igraph/igraph#836
- Fixed
head_of()
andtail_of()
, they were mixed up. - Plot: make
label.dist
independent of label lengths, fixes #63. - Plot: no error for unknown graphical parameters.
- Import functions from base packages, to eliminate
R CMD check
NOTE
s. - Re-add support for edge weights in Fruchterman-Reingold layout
- Check membership vector in
modularity()
. - Rename
str.igraph()
toprint_all()
. - Use the igraph version in exported graphs, instead of @VERSION@ #75.
- Functions that can be used inside a
V()
orE()
indexing now begin with a dot. Old names are deprecated. New names:.nei()
,.innei()
,.outnei()
,.inc()
,.from()
,.to()
. #22 - Fix packages that convert graphs to graph::graphNEL: they don't need to attach 'graph' manually any more.
- Fix a bugs in
layout_with_dh
,layout_with_gem
andlayout_with_sugiyama
. They crashed in some cases.
June 26, 2015
Some minor updates:
- Documentation fixes.
- Do not require a C++-11 compiler any more.
- Fedora, Solaris and Windows compilation fixes.
June 21, 2015
This is a new major version of igraph, and then why not call it 1.0.0. This does not mean that it is ready, it'll never be ready.
The biggest changes in the release are
- the new function names. Most functions were renamed to make them more consistent and readable. (Relax, old names can still be used, no need to update any code.)
- Better operations for vertex and edge sequences. Most functions return proper vertex/edge sequences instead of numeric ids.
- The versatile
make_()
andmake_graph()
functions to create graphs.
- Many functions were renamed. Old names are not documented, but can still be used.
- A generic
make_graph()
function to create graphs. - A generic
layout_()
(not the underscore!) function to create graph layouts, see alsoadd_layout_()
. - The igraph data type has changed. You need to call
upgrade_graph()
on graphs created with previous igraph versions. - Vertex and edge sequence operations: union, intersection, etc.
- Vertex and edge sequences can only be used with the graphs they belong to. This is now strictly checked.
- Most functions that return a (sub)set of vertices or edges return vertex or edge sequences instead.
- Vertex and edge sequences have a
[[
operator now, for easy viewing of vertex/edge metadata. - Vertex and edge sequences are implemented as weak references.
See also the
as_ids()
function to convert them to simple ids. - Vertex order can be specified for the circle layout now.
- Davidson-Harel layout algorithm
layout_with_dh()
. - GEM layout algorithm
layout_with_gem()
. - Neighborhood functions have a
mindist
parameter for the smallest distance to consider. all_simple_paths()
function to list all simple paths in a graph.triangles()
lists all triangles in a graph.- Fruchterman-Reingold and Kamada-Kawai layout algorithms rewritten from scratch. They are much faster and follow the original publications closely.
- Nicer printing of graphs, vertex and edge sequences.
local_scan()
function calculates scan statistics.- Embeddings:
embed_adjacency_matrix()
andembed_laplacian_matrix()
. - Product operator:
*
, the same graph multiple times. Can be also used asrep()
. - Better default colors, color palettes for vertices.
- Random walk on a graph:
random_walk()
adjacent_vertices()
andincident_edges()
functions, they are vectorized, as opposed toneighhors()
andincident()
.- Convert a graph to a long data frame with
as_long_data_frame()
.
Too many to list. Please try if your issue was fixed and (re-)report it if not. Thanks!
April 21, 2014
Some bug fixes, to make sure that the code included in 'Statistical Analysis of Network Data with R' works. See https://github.com/kolaczyk/sand
- Graph drawing: fix labels of curved edges, issue #181.
- Graph drawing: allow fixing edge labels at given positions, issue #181.
- Drop the 'type' vertex attribute after bipartite projection, the projections are not bipartite any more, issue #255.
- Print logical attributes in header properly (i.e. encoded by
l
, notx
, which is for complex attributes. Issue #578. - Add a constructor for
communities
objects, seecreate.communities()
. Issue #547. - Better error handling in the GraphML parser.
- GraphML reader is a bit more lenient now; makes it possible to read GraphML files saved from yWorks apps.
- Fixed a bug in
constaint()
, issue #580. - Bipartite projection now detects invalid edges instead of giving a cryptic error, issue #543.
- Fixed the
simplify
argument ofgraph.formula()
, which was broken, issue #586. - The function
crossing()
adds better names to the result, fixes issue #587. - The
sir()
function gives an error if the input graph is not simple, fixes issue #582. - Calling igraph functions from igraph callbacks is not allowed now, fixes issue #571.
February 4, 2014
There are a bunch of new features in the library itself, and other important changes in the life of the project. Thanks everyone for sending code and reporting bugs!
igraph's development has moved from Launchpad to github. This has actually happened several month ago, but never announced officially. The place for reporting bugs is at https://github.com/igraph/igraph/issues.
igraph's homepage is now hosted at http://igraph.org, and it is brand new. We wanted to make it easier to use and modern.
You can download nightly builds from igraph at http://igraph.org/nightly. Source and binary R packages (for windows and OSX), are all built.
- Added a demo for hierarchical random graphs, invoke it via
demo(hrg)
. - Make attribute prefixes optional when writing a GraphML file.
- Added function
mod.matrix()
. - Support edge weights in leading eigenvector community detection.
- Added the LAD library for checking (sub)graph isomorphism, version 1.
- Logical attributes.
- Added
layout.bipartite()
function, a simple two-column layout for bipartite graphs. - Support incidence matrices in bipartite Pajek files.
- Pajek files in matrix format are now directed by default, unless they are bipartite.
- Support weighted (and signed) networks in Pajek when file is in matrix format.
- Fixed a bug in
barabasi.game()
, algorithm psumtree-multiple just froze. - Function
layout.mds()
by default returns a layout matrix now. - Added support for Boolean attributes in the GraphML and GML readers and writer.
- Change MDS layout coordinates, first dim is according to first eigenvalue, etc.
plot.communities()
(plot.igraph()
, really) draws a border around the marked groups by default.- printing graphs now converts the
name
graph attribute to character - Convenience functions to query and set all attributes at once:
vertex.attriubutes()
,graph.attributes()
andedge.attributes()
. - Function
graph.disjoint.union()
handles attributes now. - Rewrite
graph.union()
to handle attributes properly. rewire()
: now supports the generation and destruction of loops.- Erdos-Renyi type bipartite random graphs:
bipartite.random.game()
. - Support the new options (predecessors and inbound_edges) of
get_shortest_paths()
, reorganized the output ofget.shortest.paths()
completely. - Added
graphlets()
and related functions. - Fix modularity values of multilevel community if there are no merges at all.
- Fixed bug when deleting edges with FALSE in the matrix notation.
- Fix
bonpow()
andalpha.centrality()
and make sure that the sparse solver is called. tkplot()
news: enable setting coordinates from the command line viatkplot.setcoords()
and access to the canvas viatkplot.canvas()
.- Fixed a potential crash in
igraph_edge_connectivity()
, because of an un-initialized variable in the C code. - Avoiding overflow in
closeness()
and related functions. - Check for NAs after converting 'type' to logical in
bipartite.projection()
. graphNEL
conversion functions only load the 'graph' package if it was not loaded before and they load it at the end of the search path, to minimize conflicts.- Fixed a bug when creating graphs from adjacency matrices, we now convert them to double, in case they are integers.
- Fixed an invalid memory read (and a potential crash) in the infomap community detection.
- Fixed a memory leak in the functions with attribute combinations.
- Removed some memory leaks from the SCG functions.
- Fixed some memory leaks in the ray tracer.
- Fixed memory leak in
graph.bfs()
andgraph.dfs()
. - Fix a bug in triad census that set the first element of the result to NaN.
- Fixed a crash in
is.chordal()
. - Fixed a bug in weighted modularity calculation, sum of the weights was truncated to an integer.
- Fixed a bug in weighted multilevel communtiies, the maximum weight was rounded to an integer.
- Fixed a bug in
centralization.closeness.tmax()
. - Reimplement push-relabel maximum flow with gap heuristics.
- Maximum flow functions now return some statistics about the push relabel algorithm steps.
- Function
arpack()
now gives error message if unknown options are given. - Fixed missing whitespace in Pajek writer when the ID attribute was numeric.
- Fixed a bug that caused the GML reader to crash when the ID attribute was non-numeric.
- Fixed issue #500, potential segfault if the two graphs in BLISS differ in the number of vertices or edges.
- Added
igraphtest()
function. - Fix dyad census instability, sometimes incorrect results were reported.
- Dyad census detects integer overflow now and gives a warning.
- Function
add.edges()
does not allow now zeros in the vertex set. - Added a function to count the number of adjacent triangles:
adjacent.triangles()
. - Added
graph.eigen()
function, eigenproblems on adjacency matrices. - Added some workarounds for functions that create a lot of
graphs,
decompose.graph()
andgraph.neighborhood()
use it. Fixes issue #508. - Added weights support for
optimal.community()
, closes #511. - Faster maximal clique finding.
- Added a function to count maximum cliques.
- Set operations: union, intersection, disjoint, union, difference, compose now work based on vertex names (if they are present) and keep attributes, closes #20.
- Removed functions
graph.intersection.by.name()
,graph.union.by.name()
,graph.difference.by.name()
. - The
+
operator on graphs now callsgraph.union()
if both argument graphs are named, and callsgraph.disjoint.union()
otherwise. - Added function
igraph.version()
. - Generate graphs from a stochastic block model:
sbm.game()
. - Do not suggest the stats, XML, jpeg and png packages any more.
- Fixed a
set.vertex/edge.attribute
bug that changed both graph objects, after copying (#533) - Fixed a bug in
barabasi.game
that caused crashes. - We use PRPACK to calculate PageRank scores see https://github.com/dgleich/prpack
- Added
'which
argument tobipartite.projection
(#307). - Add
normalized
argument to closeness functions, fixes issue #3. - R: better handling of complex attributes,
[[
on vertex/edge sets, fixes #231. - Implement the
start
argument inhrg.fit
(#225). - Set root vertex in Reingold-Tilford layout, solves #473.
- Fix betweenness normalization for directed graphs.
- Fixed a bug in
graph.density
that resulted in incorrect values for undirected graphs with loops - Fixed a bug when many graphs were created in one C call
(e.g. by
graph.decompose
), causing #550. - Fixed sparse
graph.adjacency
bugs for graphs with one edge, and graphs with zero edges. - Fixed a bug that made Bellman-Ford shortest paths calculations fail.
- Fixed a
graph.adjacency
bug for undirected, weighted graphs and sparse matrices. main
,sub
,xlab
andylab
are proper graphics parameters now (#555).graph.data.frame
coerces arguments to data frame (#557).- Fixed a minimum cut bug for weighted undirected graphs (#564).
- Functions for simulating epidemics (SIR model) on networks,
see the
sir
function. - Fixed argument ordering in
graph.mincut
and related functions. - Avoid copying attributes in query functions and print (#573), these functions are much faster now for graphs with many vertices/edges and attributes.
- Speed up writing GML and GraphML files, if some attributes are integer. It was really-really slow.
- Fix multiple root vertices in
graph.bfs
(#575).
Released Oct 28, 2013
Some bugs fixed:
- Fixed a potential crash in the infomap.community() function.
- Various fixed for the operators that work on vertex names (#136).
- Fixed an example in the arpack() manual page.
- arpack() now gives error message if unknown options are supplied (#492).
- Better arpack() error messages.
- Fixed missing whitespace in Pajek writer when ID attribute was numeric.
- Fixed dyad census instability, sometimes incorrect results were reported (#496).
- Fixed a bug that caused the GML reader to crash when the ID attribute was non-numeric
- Fixed a potential segfault if the two graphs in BLISS differ in the number of vertices or edges (#500).
- Added the igraphtest() function to run tests from R (#485).
- Dyad census detects integer overflow now and gives a warning (#497).
- R: add.edges() does not allow now zeros in the vertex set (#503).
- Add C++ namespace to the files that didn't have one. Fixes some incompatibility with other packages (e.g. rgl) and mysterious crashes (#523).
- Fixed a bug that caused a side effect in set.vertex.attributes(), set.edge.attributes() and set.graph.attributes() (#533).
- Fixed a bug in degree.distribution() and cluster.distribution() (#257).
Released May 16, 2013
Worked two CRAN check problems, and a gfortran bug (string bound checking does not work if code is called from C and without string length arguments at the "right" place).
Otherwise identical to 0.6.5-1.
Released February 27, 2013
Fixing an annoying bug, that broke two other packages on CRAN:
- Setting graph attributes failed sometimes, if the attributes were lists or other complex objects.
Released February 24, 2013
This is a minor release, to fix some very annoying bugs in 0.6.4:
- igraph should now work well with older R versions.
- Eliminate gap between vertex and edge when plotting an edge without an arrow. Fixes #1118448.
- Fixed an out-of-bounds array indexing error in the DrL layout, that potentially caused crashes.
- Fixed a crash in weighted betweenness calculation.
- Plotting: fixed a bug that caused misplaced arrows at rectangle vertex shapes.
Released February 2, 2013
The version number is not a mistake, we jump to 0.6.4 from 0.6, for technical reasons. This version was actually never really released, but some R packages of this version were uploaded to CRAN, so we include this version in this NEW file.
-
Added a vertex shape API for defining new vertex shapes, and also a couple of new vertex shapes.
-
Added the get.data.frame() function, opposite of graph.data.frame().
-
Added bipartite support to the Pajek reader and writer, closes bug #1042298.
-
degree.sequence.game() has a new method now: "simple_no_multiple".
-
Added the is.degree.sequence() and is.graphical.degree.sequence() functions.
-
rewire() has a new method: "loops", that can create loop edges.
-
Walktrap community detection now handles isolates.
-
layout.mds() returns a layout matrix now.
-
layout.mds() uses LAPACK instead of ARPACK.
-
Handle the '~' character in write.graph and read.graph. Bug #1066986.
-
Added k.regular.game().
-
Use vertex names to plot if no labels are specified in the function call or as vertex attributes. Fixes issue #1085431.
-
power.law.fit() can now use a C implementation.
-
Fixed a bug in barabasi.game() when out.seq was an empty vector.
-
Fixed a bug that made functions with a progress bar fail if called from another package.
-
Fixed a bug when creating graphs from a weighted integer adjacency matrix via graph.adjacency(). Bug #1019624.
-
Fixed overflow issues in centralization calculations.
-
Fixed a minimal.st.separators() bug, some vertex sets were incorrectly reported as separators. Bug #1033045.
-
Fixed a bug that mishandled vertex colors in VF2 isomorphism functions. Bug #1032819.
-
Pajek exporter now always quotes strings, thanks to Elena Tea Russo.
-
Fixed a bug with handling small edge weights in shortest paths calculation in shortest.paths() (Dijkstra's algorithm.) Thanks to Martin J Reed.
-
Weighted transitivity uses V(graph) as 'vids' if it is NULL.
-
Fixed a bug when 'pie' vertices were drawn together with other vertex shapes.
-
Speed up printing graphs.
-
Speed up attribute queries and other basic operations, by avoiding copying of the graph. Bug #1043616.
-
Fixed a bug in the NCV setting for ARPACK functions. It cannot be bigger than the matrix size.
-
layout.merge()'s DLA mode has better defaults now.
-
Fixed a bug in layout.mds() that resulted vertices on top of each other.
-
Fixed a bug in layout.spring(), it was not working properly.
-
Fixed layout.svd(), which was completely defunct.
-
Fixed a bug in layout.graphopt() that caused warnings and on some platforms crashes.
-
Fixed community.to.membership(). Bug #1022850.
-
Fixed a graph.incidence() crash if it was called with a non-matrix argument.
-
Fixed a get.shortest.paths bug, when output was set to "both".
-
Motif finding functions return NA for isomorphism classes that are not motifs (i.e. not connected). Fixes bug #1050859.
-
Fixed get.adjacency() when attr is given, and the attribute has some complex type. Bug #1025799.
-
Fixed attribute name in graph.adjacency() for dense matrices. Bug #1066952.
-
Fixed erratic behavior of alpha.centrality().
-
Fixed igraph indexing, when attr is given. Bug #1073705.
-
Fixed a bug when calculating the largest cliques of a directed graph. Bug #1073800.
-
Fixed a bug in the maximal clique search, closes #1074402.
-
Warn for negative weights when calculating PageRank.
-
Fixed dense, unweighted graph.adjacency when diag=FALSE. Closes issue #1077425.
-
Fixed a bug in eccentricity() and radius(), the results were often simply wrong.
-
Fixed a bug in get.all.shortest.paths() when some edges had zero weight.
-
graph.data.frame() is more careful when vertex names are numbers, to avoid their scientific notation. Fixes issue #1082221.
-
Better check for NAs in vertex names. Fixes issue #1087215
-
Fixed a potential crash in the DrL layout generator.
-
Fixed a bug in the Reingold-Tilford layout when the graph is directed and mode != ALL.
Released June 11, 2012
See also the release notes at http://igraph.sf.net/relnotes-0.6.html
- Vertices and edges are numbered from 1 instead of 0. Note that this makes most of the old R igraph code incompatible with igraph 0.6. If you want to use your old code, please use the igraph0 package. See more at http://igraph.sf.net/relnotes-0.6.html.
- The '[' and '[[' operators can now be used on igraph graphs, for '[' the graph behaves as an adjacency matrix, for '[[' is is treated as an adjacency list. It is also much simpler to manipulate the graph structure, i.e. add/remove edges and vertices, with some new operators. See more at ?graph.structure.
- In all functions that take a vector or list of vertices or edges, vertex/edge names can be given instead of the numeric ids.
- New package 'igraphdata', contains a number of data sets that can be used directly in igraph.
- Igraph now supports loading graphs from the Nexus online data repository, see nexus.get(), nexus.info(), nexus.list() and nexus.search().
- All the community structure finding algorithm return a 'communities' object now, which has a bunch of useful operations, see ?communities for details.
- Vertex and edge attributes are handled much better now. They are kept whenever possible, and can be combined via a flexible API. See ?attribute.combination.
- R now prints igraph graphs to the screen in a more structured and informative way. The output of summary() was also updated accordingly.
- It is possible to mark vertex groups on plots, via shading. Communities and cohesive blocks are plotted using this by default.
- Some igraph demos are now available, see a list via 'demo(package="igraph")'.
- igraph now tries to select the optimal layout algorithm, when plotting a graph.
- Added a simple console, using Tcl/Tk. It contains a text area for status messages and also a status bar. See igraph.console().
- Reimplemented igraph options support, see igraph.options() and getIgraphOpt().
- Igraph functions can now print status messages.
- The multi-level modularity optimization community structure detection algorithm by Blondel et al. was added, see multilevel.community().
- Distance between two community structures: compare.communities().
- Community structure via exact modularity optimization, optimal.community().
- Hierarchical random graphs and community finding, porting the code from Aaron Clauset. See hrg.game(), hrg.fit(), etc.
- Added the InfoMAP community finding method, thanks to Emmanuel Navarro for the code. See infomap.community().
- Eccentricity (eccentricity()), and radius (radius()) calculations.
- Shortest path calculations with get.shortest.paths() can now return the edges along the shortest paths.
- get.all.shortest.paths() now supports edge weights.
- Centralization scores for degree, closeness, betweenness and eigenvector centrality. See centralization.scores().
- Personalized Page-Rank scores, see page.rank().
- Subgraph centrality, subgraph.centrality().
- Authority (authority.score()) and hub (hub.score()) scores support edge weights now.
- Support edge weights in betweenness and closeness calculations.
- bonpow(), Bonacich's power centrality and alpha.centrality(), Alpha centrality calculations now use sparse matrices by default.
- Eigenvector centrality calculation, evcent() now works for directed graphs.
- Betweenness calculation can now use arbitrarily large integers, this is required for some lattice-like graphs to avoid overflow.
- Support the DL file format in graph.read(). See http://www.analytictech.com/networks/dataentry.htm.
- Support writing the LEDA file format in write.graph().
- Star layout: layout.star().
- Layout based on multidimensional scaling, layout.mds().
- New layouts layout.grid() and layout.grid.3d().
- Sugiyama layout algorithm for layered directed acyclic graphs, layout.sugiyama().
- New graph generators: static.fitness.game(), static.power.law.game().
- barabasi.game() was rewritten and it supports three algorithms now, the default algorithm does not generate multiple or loop edges. The graph generation process can now start from a supplied graph.
- The Watts-Strogatz graph generator, igraph_watts_strogatz() can now create graphs without loop edges.
- Added the Spectral Coarse Graining algorithm, see scg().
- The cohesive.blocks() function was rewritten in C, it is much faster now. It has a nicer API, too. See demo("cohesive").
- Added generic breadth-first and depth-first search implementations with many callbacks, graph.bfs() and graph_dfs().
- Support vertex and edge coloring in the VF2 (sub)graph isomorphism functions (graph.isomorphic.vf2(), graph.count.isomorphisms.vf2(), graph.get.isomorphisms.vf2(), graph.subisomorphic.vf2(), graph.count.subisomorphisms.vf2(), graph.get.subisomorphisms.vf2()).
- Assortativity coefficient, assortativity(), assortativity.nominal() and assortativity.degree().
- Vertex operators that work by vertex names: graph.intersection.by.name(), graph.union.by.name(), graph.difference.by.name(). Thanks to Magnus Torfason for contributing his code!
- Function to calculate a non-induced subgraph: subgraph.edges().
- More comprehensive maximum flow and minimum cut calculation, see functions graph.maxflow(), graph.mincut(), stCuts(), stMincuts().
- Check whether a directed graph is a DAG, is.dag().
- has.multiple() to decide whether a graph has multiple edges.
- Added a function to calculate a diversity score for the vertices, graph.diversity().
- Graph Laplacian calculation (graph.laplacian()) supports edge weights now.
- Biconnected component calculation, biconnected.components() now returns the components themselves.
- bipartite.projection() calculates multiplicity of edges.
- Maximum cardinality search: maximum.cardinality.search() and chordality test: is.chordal()
- Convex hull computation, convex.hull().
- Contract vertices, contract.vertices().
Released November 22, 2009
- Some small changes to make 'R CMD check' clean
- Fixed a bug in graph.incidence, the 'directed' and 'mode' arguments were not handled correctly
- Betweenness and edge betweenness functions work for graphs with many shortest paths now (up to the limit of long long int)
- When compiling the package, the configure script fails if there is no C compiler available
- igraph.from.graphNEL creates the right number of loop edges now
- Fixed a bug in bipartite.projection() that caused occasional crashes on some systems
Released April 10, 2009
See also the release notes at http://igraph.sf.net/relnotes-0.5.2.html
- Added progress bar support to beweenness() and betweenness.estimate(), layout.drl()
- Speeded up betweenness estimation
- Speeded up are.connected()
- Johnson's shortest paths algorithm added
- shortest.paths() has now an 'algorithm' argument to choose from the various implementations manually
- Always quote symbolic vertex names when printing graphs or edges
- Average nearest neighbor degree calculation, graph.knn()
- Weighted degree (also called strength) calculation, graph.strength()
- Some new functions to support bipartite graphs: graph.bipartite(), is.bipartite(), get.incidence(), graph.incidence(), bipartite.projection(), bipartite.projection.size()
- Support for plotting curved edges with plot.igraph() and tkplot()
- Added support for weighted graphs in alpha.centrality()
- Added the label propagation community detection algorithm by Raghavan et al., label.propagation.community()
- cohesive.blocks() now has a 'cutsetHeuristic' argument to choose between two cutset algorithms
- Added a function to "unfold" a tree, unfold.tree()
- New tkplot() arguments to change the drawing area
- Added a minimal GUI, invoke it with tkigraph()
- The DrL layout generator, layout.drl() has a three dimensional mode now.
- Fixed a bug in VF2 graph isomorphism functions
- Fixed a bug when a sparse adjacency matrix was requested in get.adjacency() and the graph was named
- VL graph generator in degree.sequence.game() checks now that the sum of the degrees is even
- Many fixes for supporting various compilers, e.g. GCC 4.4 and Sun's C compiler
- Fixed memory leaks in graph.automorphisms(), Bellman-Ford shortest.paths(), independent.vertex.sets()
- Fix a bug when a graph was imported from LGL and exported to NCOL format (#289596)
- cohesive.blocks() creates its temporary file in the session temporary directory
- write.graph() and read.graph() now give error messages when unknown arguments are given
- The GraphML reader checks the name of the attributes to avoid adding a duplicate 'id' attribute
- It is possible to change the 'ncv' ARPACK parameter for leading.eigenvector.community()
- Fixed a bug in path.length.hist(), 'unconnected' was wrong for unconnected and undirected graphs
- Better handling of attribute assingment via iterators, this is now also clarified in the manual
- Better error messages for unknown vertex shapes
- Make R package unload cleanly if unloadNamespace() is used
- Fixed a bug in plotting square shaped vertices (#325244)
- Fixed a bug in graph.adjacency() when the matrix is a sparse matrix of class "dgTMatrix"
Released July 14, 2008
See also the release notes at http://igraph.sf.net/relnotes-0.5.1.html
- A new layout generator called DrL.
- Uniform sampling of random connected undirected graphs with a given degree sequence.
- Edge labels are plotted at 1/3 of the edge, this is better if the graph has mutual edges.
- Initial and experimental vertex shape support in 'plot'.
- New function, 'graph.adjlist' creates igraph graphs from adjacency lists.
- Conversion to/from graphNEL graphs, from the 'graph' R package.
- Fastgreedy community detection can utilize edge weights now, this was missing from the R interface.
- The 'arrow.width' graphical parameter was added.
- graph.data.frame has a new argument 'vertices'.
- graph.adjacency and get.adjacency support sparse matrices, the 'Matrix' package is required to use this functionality.
- graph.adjacency adds column/row names as 'name' attribute.
- Weighted shortest paths using Dijkstra's or the Belmann-Ford algorithm.
- Shortest path functions return 'Inf' for unreachable vertices.
- New function 'is.mutual' to find mutual edges in a directed graph.
- Added inverse log-weighted similarity measure (a.k.a. Adamic/Adar similarity).
- preference.game and asymmetric.preference.game were rewritten, they are O(|V|+|E|) now, instead of O(|V|^2).
- Edge weight support in function 'get.shortest.paths', it uses Dijkstra's algorithm.
- A bug was corrected in write.pajek.bgraph.
- Several bugs were corrected in graph.adjacency.
- Pajek reader bug corrected, used to segfault if '*Vertices' was missing.
- Directedness is handled correctly when writing GML files. (But note that 'correct' conflicts the standard here.)
- Corrected a bug when calculating weighted, directed PageRank on an undirected graph. (Which does not make sense anyway.)
- Several bugs were fixed in the Reingold-Tilford layout to avoid edge crossings.
- A bug was fixed in the GraphML reader, when the value of a graph attribute was not specified.
- Fixed a bug in the graph isomorphism routine for small (3-4 vertices) graphs.
- Corrected the random sampling implementation (igraph_random_sample), now it always generates unique numbers. This affects the Gnm Erdos-Renyi generator, it always generates simple graphs now.
- The basic igraph constructor (igraph_empty_attrs, all functions are expected to call this internally) now checks whether the number of vertices is finite.
- The LGL, NCOL and Pajek graph readers handle errors properly now.
- The non-symmetric ARPACK solver returns results in a consistent form now.
- The fast greedy community detection routine now checks that the graph is simple.
- The LGL and NCOL parsers were corrected to work with all kinds of end-of-line encodings.
- Hub & authority score calculations initialize ARPACK parameters now.
- Fixed a bug in the Walktrap community detection routine, when applied to unconnected graphs.
- Several small memory leaks were removed, and a big one from the Spinglass community structure detection function
Released February 14, 2008
See also the release notes at http://igraph.sf.net/relnotes-0.5.html
-
The 'rescale', 'asp' and 'frame' graphical parameters were added
-
Create graphs from a formula notation (graph.formula)
-
Handle graph attributes properly
-
Calculate the actual minimum cut for undirected graphs
-
Adjacency lists, get.adjlist and get.adjedgelist added
-
Eigenvector centrality computation is much faster now
-
Proper R warnings, instead of writing the warning to the terminal
-
R checks graphical parameters now, the unknown ones are not just ignored, but an error message is given
-
plot.igraph has an 'add' argument now to compose plots with multiple graphs
-
plot.igraph supports the 'main' and 'sub' arguments
-
layout.norm is public now, it can normalize a layout
-
It is possible to supply startup positions to layout generators
-
Always free memory when CTRL+C/ESC is pressed, in all operating systems
-
plot.igraph can plot square vertices now, see the 'shape' parameter
-
graph.adjacency rewritten when creating weighted graphs
-
We use match.arg whenever possible. This means that character scalar options can be abbreviated and they are always case insensitive
-
VF2 graph isomorphism routines can check subgraph isomorphism now, and they are able to return matching(s)
-
The BLISS graph isomorphism algorithm is included in igraph now. See canonical.permutation, graph.isomorphic.bliss
-
We use ARPACK for eigenvalue/eigenvector calculation. This means that the following functions were rewritten: page.rank, leading.eigenvector.community.*, evcent. New functions based on ARPACK: hub.score, authority.score, arpack.
-
Edge weights for Fruchterman-Reingold layout (layout.fruchterman.reingold).
-
Line graph calculation (line.graph)
-
Kautz and de Bruijn graph generators (graph.kautz, graph.de.bruijn)
-
Support for writing graphs in DOT format
-
Jaccard and Dice similarity coefficients added (similarity.jaccard, similarity.dice)
-
Counting the multiplicity of edges (count.multiple)
-
The graphopt layout algorithm was added, layout.graphopt
-
Generation of "famous" graphs (graph.famous).
-
Create graphs from LCF notation (graph.cf).
-
Dyad census and triad cencus functions (dyad.census, triad.census)
-
Cheking for simple graphs (is.simple)
-
Create full citation networks (graph.full.citation)
-
Create a histogram of path lengths (path.length.hist)
-
Forest fire model added (forest.fire.game)
-
DIMACS reader can handle different file types now
-
Biconnected components and articulation points (biconnected.components, articulation.points)
-
Kleinberg's hub and authority scores (hub.score, authority.score)
-
as.undirected handles attributes now
-
Geometric random graph generator (grg.game) can return the coordinates of the vertices
-
Function added to convert leading eigenvector community structure result to a membership vector (community.le.to.membership)
-
Weighted fast greedy community detection
-
Weighted page rank calculation
-
Functions for estimating closeness, betweenness, edge betweenness by introducing a cutoff for path lengths (closeness.estimate, betweenness.estimate, edge.betweenness.estimate)
-
Weighted modularity calculation
-
Function for permuting vertices (permute.vertices)
-
Betweenness and closeness calculations are speeded up
-
read.graph can handle all possible line terminators now (\r, \n, \r\n, \n\r)
-
Error handling was rewritten for walktrap community detection, the calculation can be interrupted now
-
The maxflow/mincut functions allow to supply NULL pointer for edge capacities, implying unit capacities for all edges
- Fixed a bug in cohesive.blocks, cohesive blocks were sometimes not calculated correctly
Released January 1, 2008
New:
- Cohesive block finding in the R interface, thanks to Peter McMahan for contributing his code. See James Moody and Douglas R. White, 2003, in Structural Cohesion and Embeddedness: A Hierarchical Conception of Social Groups American Sociological Review 68(1):1-25
- Biconnected components and articulation points.
- R interface: better printing of attributes.
- R interface: graph attributes can be used via '$'.
Bug fixed:
- Erdos-Renyi random graph generators rewritten.
Released October 3, 2007
This release should work seamlessly with the new R 2.6.0 version. Some other bugs were also fixed:
- A bug was fixed in the Erdos-Renyi graph generator, which sometimes added an extra vertex.
Released August 13, 2007
The next one in the sequence of bugfix releases. Thanks to many people sending bug reports. Here are the changes:
- Some memory leaks removed when using attributes from R or Python.
- GraphML parser: entities and character data in multiple chunks are now handled correctly.
- A bug corrected in edge betweenness community structure detection, it failed if called many times from the same program/session.
- Edge betweeness community structure: handle unconnected graphs properly.
- Fixed bug related to fast greedy community detection in unconnected graphs.
- Use a different kind of parser (Push) for reading GraphML files. This is almost invisible for users but fixed a nondeterministic bug when reading in GraphML files.
- R interface: plot now handles properly if called with a vector as the edge.width argument for directed graphs.
- R interface: bug (typo) corrected for walktrap.community and weighted graphs.
Released June 7, 2007
This is another bugfix release, as there was a serious bug in the R package of the previous version: it could not read and write graphs to files in any format under MS Windows.
Some other bits added:
- circular Reingold-Tilford layout generator for trees
- corrected a bug, Pajek files are written properly under MS Windows now.
- arrow.size graphical edge parameter added in the R interface.
Released May 23, 2007
This is a minor release, it corrects a number of bugs, mostly in the R package.
Released May 21, 2007
The major new additions in this release is a bunch of community detection algorithms and support for the GML file format. Here is the complete list of changes:
-
as the internal representation changed, graphs stored with 'save' with an older igraph version cannot be read back with the new version reliably.
-
neighbors returns ordered lists
-
is.loop and is.multiple were added
-
topological sorting
-
VF2 isomorphism algorithm
-
support for reading graphs from the Graph Database for isomorphism
-
graph.mincut can calculate the actual minimum cut
-
girth calculation added, thanks to Keith Briggs
-
support for reading and writing GML files
-
Walktrap community detection algorithm added, thanks to Matthieu Latapy and Pascal Pons
-
edge betweenness based community detection algorithm added
-
fast greedy algorithm for community detection by Clauset et al. added thanks to Aaron Clauset for sharing his code
-
leading eigenvector community detection algorithm by Mark Newman added
-
functions for creating dendrograms from the output of the community detection algorithms added
-
community.membership supporting function added, creates a membership vector from a community structure merge tree
-
modularity calculation added
-
graphics parameter handling is completely rewritten, uniform handling of colors and fonts, make sure you read ?igraph.plotting
-
new plotting parameter for edges: arrow.mode
-
a bug corrected when playing a nonlinear barabasi.game
-
better looking plotting in 3d using rglplot: edges are 3d too
-
rglplot layout is allowed to be two dimensional now
-
rglplot suspends updates while drawing, this makes it faster
-
loop edges are correctly plotted by all three plotting functions
-
better printing of attributes when printing graphs
-
summary of a graph prints attribute names
-
is.igraph rewritten to make it possible to inherit from the 'igraph' class
-
somewhat better looking progress meter for functions which support it
- many functions benefit from the new internal representation and are faster now: transitivity, reciprocity, graph operator functions like intersection and union, etc.
- corrected a bug when reading Pajek files: directed graphs were read as undirected
Released Dec 19, 2006
This is a new major release, it contains many new things:
- bonpow function ported from SNA to calculate Bonacich power centrality
- get.adjacency supports attributes now, this means that it sets the colnames and rownames attributes and can return attribute values in the matrix instead of 0/1
- grg.game, geometric random graphs
- graph.density, graph density calculation
- edge and vertex attributes can be added easily now when added new edges with add.edges or new vertices with add.vertices
- graph.data.frame creates graph from data frames, this can be used to create graphs with edge attributes easily
- plot.igraph and tkplot can plot self-loop edges now
- graph.edgelist to create a graph from an edge list, can also handle edge lists with symbolic names
- get.edgelist has now a 'names' argument and can return symbolic vertex names instead of vertex ids, by default id uses the 'name' vertex attribute is returned
- printing graphs on screen also prints symbolic symbolic names (the 'name' attribute if present)
- maximum flow and minimum cut functions: graph.maxflow, graph.mincut
- vertex and edge connectivity: edge.connectivity, vertex.connectivity
- edge and vertex disjoint paths: edge.disjoint.paths, vertex.disjoint.paths
- White's cohesion and adhesion measure: graph.adhesion, graph.cohesion
- dimacs file format added
- as.directed handles attributes now
- constraint corrected, it handles weighted graphs as well now
- weighted attribute to graph.adjacency
- spinglass-based community structure detection, the Joerg Reichardt -- Stefan Bornholdt algorithm added: spinglass.community
- graph.extended.chordal.ring, extended chordal ring generation
- no.clusters calculates the number of clusters without calculating the clusters themselves
- minimum spanning tree functions updated to keep attributes
- transitivity can calculate local transitivity as well
- neighborhood related functions added: neighborhood, neighborhood.size, graph.neighborhood
- new graph generators based on vertex types: preference.game and asymmetric.preference.game
- attribute handling bug when deleting edges corrected
- GraphML escaping and NaN handling corrected
- bug corrected to make it possible compile the R package without the libxml2 library
- a bug in Erdos-Renyi graph generation corrected: it had problems with generating large directed graphs
- bug in constraint calculation corrected, it works well now
- fixed memory leaks in the GraphML reader
- error handling bug corrected in the GraphML reader
- bug corrected in R version of graph.laplacian when normalized Laplacian is requested
- memory leak corrected in get.all.shortest.paths in the R package
Released Aug 23, 2006
This is a bug-fix release. Bugs fixed:
- reciprocity corrected to avoid segfaults
- some docs updates
- various R package updates to make it conform to the CRAN rules
Released Aug 18, 2006
Release time at last! There are many new things in igraph 0.2, the most important ones:
- reading writing Pajek and GraphML formats with attributes (not all Pajek and GraphML files are supported, see documentation for details)
- the RANDEDU fast motif search algorithm is implemented
- many new graph generators, both games and regular graphs
- many new structural properties: transitivity, reciprocity, etc.
- graph operators: union, intersection, difference, structural holes, etc.
- conversion between directed and undirected graphs
- new layout algorithms for trees and large graphs, 3D layouts and many more.
New things specifically in the R package:
- support for CTRL+C
- new functions: Graph Laplacian, Burt's constraint, etc.
- vertex/edge sequences totally rewritten, smart indexing (see manual)
- new R manual and tutorial: `Network Analysis with igraph', still under development but useful
- very basic 3D plotting using OpenGL
Although this release was somewhat tested on Linux, MS Windows, Mac OSX, Solaris 8 and FreeBSD, no heavy testing was done, so it might contain bugs, and we kindly ask you to send bug reports to make igraph better.
Released Jan 30, 2006
After about a year of development this is the first "official" release of the igraph library. This release should be considered as beta software, but it should be useful in general. Please send your questions and comments.