- Renamed the
_hypergraph
internal variable to_net_attr
#560 (@nwlandry). - Get rid of KeyErrors in
to_line_graph
#558 (@pgberlureau). - Fix
asfptype()
scipy error #559 (@nwlandry). - Added the ability for XGI to load data collections #540 (@nwlandry).
- Made the
DiHypergraph
class more consistent with other class internals #541 (@nwlandry). - Fix Numpy 2.0 breaking changes #547 (@nwlandry).
- Moved tutorials to the top level #550 (@nwlandry).
- Added numpy<2.0 to the requirements to avoid breaking changes with Numpy version 2.0 #545 (@nwlandry).
- Update license and contributing #539 (@nwlandry).
- Added error handling for JSON duplicate IDs #538 (@nwlandry).
- Added a new drawing recipe for drawing multiple hypergraphs with the same node positions #535 (@nwlandry).
- Added the ability to draw hypergraphs and dihypergraphs as bipartite graphs, added a bipartite spring layout, and added the ability to place edge markers at the barycenters of the positions of their member nodes with the
edge_positions_from _barycenters
function #492 (@nwlandry). - Updated the documentation to include (1) new relevant software packages organized by language, (2) documentation for the encapsulation DAG, (3) new gallery examples, (4) new projects using XGI, and (5) links to the XGI JOSS paper #529 (@nwlandry).
- More intelligible warning for unknown type of degree assortativity #533 (@doabell).
- Formatted the numbers in the XGI-DATA table by locale #532 (@doabell).
- Added a hypergraph random edge shuffle method #531 (@doabell).
- Updated the return type of Katz centrality to be a dictionary and added it as a NodeStat #530 (@nwlandry).
- Made
degree_assortativity
with theexact=True
keyword reproducible #526 (@nwlandry).
- Added the ability to supply user-defined functions to
filterby
andfilterby_attr
#524 (@nwlandry). - Reorganized the stats tests #525 (@nwlandry).
- Added a recipe for finding the maximal indices based on statistics #522 (@nwlandry).
- Added the ability to argsort stats #521 (@nwlandry).
- Added recipes for flagged triangular lattices and avg. shortest path length #513 (@thomasrobiglio).
- Add
argmax
andargmin
to the stats interface #518 (@nwlandry). - Updated the list of projects using XGI #519 (@nwlandry).
- Fixed the multilayer warning (Issue #515) that occurs when specifying colors for each node #517 (@nwlandry).
- Fixed the drawing warning (Issue #499) that occurs when specifying colors for each node #512 (@nwlandry).
- Formatted the tutorial notebooks with isort #502 (@nwlandry).
- Fixed a missing link #503 (@maximelucas).
- Added a workflow to check all the urls in the documentation work #498 (@nwlandry).
- Added an XGI-DATA page to ReadTheDocs with network statistics #496 (@nwlandry).
- Added syntax highlighting in the tutorials #497 (@thomasrobiglio).
- Changed the xgi-data index location to Github to support the move to Zenodo-hosted datasets #494 (@nwlandry)
- Update the list of publications using XGI #493 (@nwlandry).
- Added the ability to draw a convex hull as an option in
draw_hyperedges
#491 (@maximelucas). - Fixed sphinx documentation errors #487 (@nwlandry).
- Rewrote the
draw_multilayer
function #486 (@maximelucas). - Updated the "Projects Using XGI" page #489 (@nwlandry).
- Added support for Python 3.12 #488 (@nwlandry).
- Made
IDViews
respect edge insertion order in thefrom_view
method #482 (@nwlandry). - Fixed issues in drawing functions #476 (@maximelucas).
- Fixed Issue #480 which raised an error when users attempted to set edge properties with an id:value dictionary with an attribute name #481 (@nwlandry).
- Made minor changes and corrections to the tutorials in the documentation #479 (@thomasrobiglio).
- Rewrote the
draw_hyperedges()
function to utilize native matplotlib functionality and be more consistent #456 (@maximelucas). - Integrated the tutorials into the documentation #457 (@thomasrobiglio).
- Added an
s
parameter to theneighbors
method for edge overlap #450 (@nwlandry). - Fixed Issue #468 so that the
ashist
method doesn't try to create multiple bins for stats with a single unique value #473 (@nwlandry).
- Response to PyOpenSci review, fixing Issue #453 among other things #470 (@nwlandry).
- Fixed issues created from merging PR #380 #471 (@nwlandry).
- Added the capability to generate the complement of a hypergraph #380 (@acombretrenouard).
- Fixed the directionality of BiGG reactions (See Issue #458) #459 (@pietrotraversa, @nwlandry).
- Fixed Issue #461 #462 (@maximelucas).
- Added the
ashist()
method to the stats module#452 (@nwlandry). - Feature: return the node collections to allow a colorbar corresponding to node colors #441 (@maximelucas).
- Added tutorials: XGI in X minutes #415 (@thomasrobiglio).
- Fixed an empty edge error when loading BiGG data #447 (@nwlandry).
- Changed rho to an optional arg in the HPPM method #446 (@nwlandry).
- Added a
cleanup()
method toDiHypergraph
andSimplicialComplex
and added aconnected
argument to theHypergraph
method #440 (@nwlandry). - Added encapsulation DAG functionality #444 (@tlarock).
- Refactored the multilayer drawing function to be more standard and added headers to convert module #434 (@nwlandry).
- Added the
weights
option toto_line_graph()
function. #427 (@tlarock). - Added the ability for users to access the optional arguments of NetworkX layout functions. #439 (@nwlandry).
- Fixed Issue #331 #438 (@maximelucas).
- Refactored the draw module #435 (@maximelucas).
- Feature: added the
aspect
keyword for drawing, addressing Issue #430 #432 (@maximelucas).
- Listed the available statistics in the stats module #405 (@thomasrobiglio).
- Implemented functions from this article #400 (@maximelucas).
- Reorganized the
convert
module #423 (@nwlandry). - Refactored the core data structures and removed the
function.py
file, moving the functions to other locations #412 (@nwlandry). - Small documentation fixes #422 (@nwlandry).
- Added a webpage listing projects and papers that use XGI #416 (@nwlandry).
- Added an optinion so that only the text in the Jupyter Notebooks counts towards the line counts #417 (@nwlandry).
- Fixed the
_color_arg_to_dict
and_scalar_arg_to_dict
functions so they are more consistent #402 (@nwlandry). - Fixed the IDView.ids type #406 (@leotrs).
- Fix: Converting from a
SimplicialComplex
to aHypergraph
now adds all of its faces to the hypergraph, not just the maximal faces. Added unit tests for converting betweenHypergraph
andSimplicialComplex
classes #399 (@thomasrobiglio). - Moved the list of contributors from the readthedocs, license, setup, etc. into a dedicated file so that when contributors join or leave, we only update a single file. Renamed
CONTRIBUTING.md
toHOW_TO_CONTRIBUTE.md
. #401 (@nwlandry). - Added the ability to convert from a simplex dict to a
SimplicialComplex
#397 (@thomasrobiglio). - Updated index.rst to add the
DiHypergraph
class to the quick references #392 (@thomasrobiglio). - Addressed Issue #393 by adding the ability for
draw_hypergraph()
to handle nodes with non-integer IDs and isolated nodes. #394 (@thomasrobiglio). - Fixed the short description and landing page on PyPI so that it is more readable #391 (@nwlandry).
- Added the functionality to load datasets from the BiGG database #384 (@nwlandry).
- Added the
draw_dihypergraph()
method to draw directed hypergraphs #387 (@thomasrobiglio). - Changed "an hypergraph" to "a hypergraph" in the docstrings #386 (@acombretrenouard).
- Fixed mathematical expressions that were not represented properly in the docstrings part 2 #382 (@nwlandry).
- Fixed mathematical expressions that were not represented properly in the docstrings #381 (@nwlandry).
- Added the DiHypergraph class, directed view classes, and directed stats classes #372 (@nwlandry).
- Added Katz-centrality algorithm #370 (@acombretrenouard).
- Added the
draw_multilayer()
drawing function #369 (@thomasrobiglio). - Fixed
members()
andmemberships()
methods so that views are read-only #371 (@nwlandry). - Added shortest path algorithms #368 (@acombretrenouard).
- Added new drawing layouts (
circular_layout
,spiral_layout
, andbarycenter_kamada_kawai_layout
) #360 (@thomasrobiglio).
- Formatted the codebase with ruff #346 (@leotrs).
- Improved the performance of
flag_complex
andrandom_flag_complex
#355 (@maximelucas). - Moved the
IDDict
class to utilities and removed unnecessary arguments in theIDView
constructor #353 (@nwlandry). - Up-versioned sphinx to v6.xxx and up-versioned sphinx-rtd-theme to >= 1.2 to be compatible with Sphinx v6.xxx. Updated the contribution guide and the GH, PyPI, and RTD landing pages #350 (@nwlandry).
- Fixed a bug in the double edge swap method #349 (@nwlandry)
- Updated the convert module methods to be able to return a hypergraph, addressing Issue #327. Now the conversion from a simplicial complex to a hypergraph only includes maximal faces #345 (@thomasrobiglio)
- Place a ceiling on IPython so that it is Python 3.8 compatible #344 (@nwlandry).
- Started fixing the cross references in the "See Also" section in the function/class docstrings #343 (@nwlandry).
- Feature: added complete_hypergraph #337 (@maximelucas).
- Updated the quickstart notebook #338 (@nwlandry).
- Added
strict
keyword to themaximal()
method #332 (@nwlandry). - Feature: added trivial hypergraph #335 (@maximelucas).
- Changed the organization name to
xgi-org
and removedcodecov
from the test requirements file #334 (@nwlandry). - Fix: improved the
degree_counts
documentation #329 (@maximelucas). - Added the
maximal()
method to EdgeView, removed themaximal_simplices()
method, and removed a bug from theduplicates()
method #324 (@nwlandry). - Fix: documented max_order in
add_simplices_from()
#328 (@maximelucas). - Restructured folder for the generators and linalg modules #321 (@maximelucas).
- Added 3 clustering coefficient definitions to the
algorithms
module as well as to NodeStats #316 (@nwlandry). - Added the ability to choose whether to output the index-to-ID mappings from
to_bipartite_graph()
method. Fixes #322 #323 (@leotrs). - Added the ability to draw hypergraphs with hyperedges as convex hulls #320 (@thomasrobiglio).
- Renamed
convert_to_line_graph()
toto_line_graph()
, added ans
parameter to the function, and added corresponding unit tests #318 (@nwlandry). - Made the sparse warning in the
adjacency_matrix()
function more intelligible #315 (@nwlandry). - Added a function for the normalized hypergraph laplacian #314 (@nwlandry).
- Added tests for draw functions #312 (@maximelucas).
- Updated the centrality functions so they more gracefully handle empty and disconnected hypergraphs #313 (@nwlandry).
- Added
keep_isolates
argument to the subhypergraph function #308 (@maximelucas). - Fix: raise error for assortativity of empty hypergraph #307 (@maximelucas).
- Minor: renamed to from_max_simplices #306 (@maximelucas).
- Refactored
incidence_matrix()
for ~4x speedup, made output consistent for empty matrices, added tests, and refactoredmultiorder_laplacian()
so all internal variables are sparse if sparse=True. Fixes #301 #303 (@maximelucas). - Renamed plotting functions,
xgi_pylab
module, and node/hyperedge/simplex plotting functions. All drawing functions now return axes. Addedpca_transform()
to rotate the node positions relative to the principal axes #300 (@nwlandry). - Changed the Github actions to test all notebooks in the tutorial folder #299 (@nwlandry).
- Added the
convert_to_line_graph()
function and thevector_centrality()
function, which uses it #290 (@goznalo-git). - Fixed the quickstart notebook by updating the synchronization #294 (@nwlandry).
- Added more tests for the layout functions #296 (@maximelucas).
- Added basic tests for layout functions #293 (@maximelucas).
- Added tests for generators #291 (@maximelucas).
- Added the ability to specify sparsity in the matrix functions in the linalg module #284 (@nwlandry).
- Added the
uniform_HSBM()
anduniform_HPPM
generative models #286 (@nwlandry). - Up-versioned requirements to fix #287 and make compatible with NetworkX #288 (@nwlandry).
- Added code coverage with
codecov
and displayed coverage on main page #285 (@nwlandry). - Fixed a bug in the
add_edge()
method #289 (@nwlandry). - Added examples of sorting matrices by node/edge IDs to the documentation #282 (@nwlandry).
- Added the ability in
draw()
to plot any node positions by rescaling the plot area #279 (@maximelucas).
- Fixed issue #270 #271 (@nwlandry).
- Fixed the centralities so that they are positive and 1-normalized #274 (@nwlandry).
- Added support for NetworkX 3.0, removed support for Python 3.7, and changed all scipy sparse matrices to scipy sparse arrays #268 (@nwlandry).
- Added the ability to display the list of available datasets in xgi-data with
load_xgi_data()
#266 (@nwlandry).
- Added the
find_triangles()
andflag_complex_d2()
functions. Theflag_complex_d2()
function is much faster thanflag_complex()
for simplicial complexes of max order 2. Also refactoredrandom_flag_complex_d2()
to useflag_complex_d2()
#263 (@maximelucas). - add the
items()
method so NodeStats and EdgeStats are even more dict-like #233 (@leotrs). - Added the ability to cache the output of
load_xgi_data()
and added more interpretable errors when the http request fails #261 (@nwlandry). - Deleted the data folder #260 (@nwlandry).
- Split the simulation of the Kuramoto model and its order parameter into two functions #257 (@maximelucas).
- Added the ability to write/read xgi-data datasets to/from a file, and pointed the
load_xgi_data()
function to the new xgi-data collection in Gitlab #254 (@acuschwarze). - remove singletons from random generators #256 (@maximelucas).
- Remove references to the disGene dataset in the data folder #253 (@nwlandry).
- Updated the new release process #249 (@nwlandry).
draw()
now correctly plots simplicial complexes with themax_order
keyword #248 (@maximelucas).- Changed the
add_simplex
method to be non recursive #247 (@maximelucas). - Added tests for the SimplicialComplex class #245 (@maximelucas).
- Made all draw functions available from xgi #246 (@maximelucas).
- Added an indent to make hypergraph json files more readable #242 (@maximelucas).
- Improved the efficiency of the uid update function #239 (@nwlandry).
- Added the ability to display the node and hyperedge labels in
draw()
#234 (@mcontisc). - Fixed the uid counter initialisation #225 (@maximelucas).
- Added the ability to pickle hypergraphs #229 (@nwlandry).
- Made
random_hypergraph()
andrandom_simplicialcomplex()
faster #213 (@maximelucas). - Fixed a bug in
dynamical_assortativity()
#230 (@nwlandry). - Removed all random decorators #227 (@nwlandry).
- Modified
unique_edge_sizes()
so that the list of sizes is now sorted #226 (@nwlandry). - Added the
merge_duplicate_edges()
function to merge multi-edges #210 (@nwlandry). - Partial speed-up of
draw
function #211 (@iaciac). - Added a simplicial synchronization function #212 (@Marconurisso).
- Sped up the
add_simplices_from()
method #223 (@maximelucas). - Updated the
add_simplices_from()
method to matchadd_hyperedges_from()
#220 (@maximelucas).
- Fixed #214, added a
powerset()
function, added asubfaces()
function, and added examples of these functions (#209). - Refactored the NodeStats and EdgeStats classes to be more efficient (#209).
- Implemented set operations for NodeView and EdgeView (#208).
- Addressed #180 with
density()
andincidence_density()
functions (#204 and #207). - Added the
<<
operator to "add" two hypergraphs together (#203). - Improved the documentation (#202).
- Added Python 3.11 to the test suite (#201).
- Added an option to fill only some cliques with probabilities
ps
toxgi.flag_complex()
(#200). - Fixed Issue #198 (#199).
- Refactored
load_xgi_data()
to calldict_to_hypergraph()
and fixed a bug indict_to_hypergraph()
(#193). - Added
num_edges_order()
to get the number of edges of a given order and added anorder
parameter to thedegree_counts()
function (#192). - Fixed #182 and #186 by adding a
max_order
argument todraw()
andload_xgi_data()
(#173) . - Made
draw()
faster by refactoring_color_arg_to_dict()
and_scalar_arg_to_dict()
(#173).
Contributors: @leotrs, @maximelucas, and @nwlandry
-
Hypergraph.has_edge
is nowIDView.lookup
,Hypergraph.duplicate_edges
is nowIDView.duplicates
, andutilities.convert_labels_to_integer
is nowfunction.convert_labels_to_integer
(#150). - Added some unit tests for the convert module, the function module, and the classic generators module. Fixed for minor bugs encountered while writing tests and added documentation to Read The Docs for the drawing module. (#153)
- Fixed a bug in
remove_node_from_edge()
(#154). - Implemented computation of moments for NodeStat and EdgeStat (#155).
- Implemented weak and strong node removal as per issue #167 (#156).
- Added a dynamics module and created a Kuramoto model synchronization function (#159).
- Added a cleanup method that removes artifacts specified by the user: multi-edges, singletons, isolates. It also can convert all labels to consecutive integers (#161).
- Modified the
duplicates()
method to not include the first instance of the node/edge in the list of duplicates (#161). - Converted all instances of edges to sets from lists in response to issue #158 (#162).
- Added lambda function default arguments for
$f$ ,$g$ ,$\varphi$ ,$\psi$ as defined by Tudisco and Higham. Default behavior is identical as before. Fixes #132 (#165). - Added
sum()
as a stats method (#168). - Added a benchmarking suite for the core hypergraph data structure using airspeed velocity (#170).
- Fixed issue #171 (#172)
Contributors: @nwlandry, @leotrs, and @saad1282
- Keyword arguments are now consistent in the
draw()
function (#148). - Notebooks are now formatted with black and the requirements have been updated to reflect this (#148).
Contributors: @nwlandry
- Added the ability to color nodes and edges in
xgi.draw()
by value, iterable, or NodeStat/EdgeStat (#139, #142, and #143). - Fixed the distortion of the node aspect ratio with different figure sizes in Issue #137.
- Moved the
isolates()
andsingletons()
method from theHypergraph
class to theNodeView
andEdgeView
classes respectively (#146). - Fixed
Hypergraph.copy()
to not use thesubhypergraph
method (#145). filterby()
now acceptsNodeStat
andEdgeStat
objects instead of just strings (#144).- Removed edit-mode install to run the Github Actions test suite (#136).
- Added unit tests (#147).
Contributors: @nwlandry, @leotrs, and @maximelucas
- Added the
stats
package which implementsNodeStat
,EdgeStat
and related functionality. This package now handles computation of edge size and degree (#120). - Removed the
EdgeSizeView
andDegreeView
classes (#120). - Changed all imports to be relative in the
xgi
package (#121). - Added an assortativity module (#122).
- Improved the performance of accessing edge members (#124).
- Added more operations for node and edge attributes besides "eq" (#125).
- Added a function to convert all node and edge labels to integers and store the old labels as properties (#127).
-
- Renamed the
egonet
method to `edge_neighborhood (#129).
- Renamed the
- Moved the
neighbors
method in theHypergraph
class to theIDView
class so that node and edge neighbors are now supported (PR #129). - Added a centrality module and added these methods to
nodestats.py
andedgestats.py
(#130). - Moved the
load_xgi_data
method to thereadwrite
module (#130). - Added a generator for sunflower hypergraphs (#130).
- Added a Jupyter notebook as a quickstart guide (#131 and #134).
- Fixed a bug in the
barycenter_spring_layout
andweighted_barycenter_spring_layout
methods to handle non-integer node IDs (#133). - Added an isort configuration file so it no longer sorts the
__init__.py
files (#134).
Contributors: @leotrs, @nwlandry, and @iaciac
- Refactored the subhypergraph methods
- Moved functions not related to the core Hypergraph data structure to functions.py
- Removed unnecessary duplicated functions (
n_bunch_iter
,get_edge_data
, andhas_node
) - Refactored the
members()
method as well as theNodeView
andEdgeView
classes for significant speedup. - Github Actions now tests the docstrings and tutorial notebooks
- The
add_edges_from
method now supports different input formats. - Fixed various bugs in the generative models module.
- A method for double edge swaps is now implemented.
Contributors: @leotrs and @nwlandry
- Added the ability to convert to and from a NetworkX bipartite graph.
- Removed the
shape
property fromHypergraph
and renamed thenumber_of_nodes()
andnumber_of_edges()
methods to thenum_nodes
andnum_edges
properties, respectively. - Added random seed decorator as in NetworkX.
- Added a
SimplicialComplex
class. - Added order and weighted arguments to the
incidence_matrix
andadjacency_matrix functions
. - Added an
intersection_profile
function. - Added a
laplacian
function with argumentorder
and amultiorder_laplacian
function. - Fix: Return an empty array rather than a 1x1 zero array when appropriate.
- Fix: Ensured that the incidence matrix is always has dimensions num_nodes x num_edges.
- Added 2 generators of random (maximal) simplicial complexes, and toy star-clique generator
- Extensively rewrote the documentation, updating the content and format on Read The Docs.
- Added an egonet function to get the edges neighboring a specified node.
- Added functions to visualize hypergraphs and simplicial complexes.
- Added the ability to get nodes and edges of given degrees/sizes respectively.
- Extended the
members()
function to be able to get different data types and to either get a single edge or all edges. - Added the
load_xgi_data
function to load datasets from the xgi-data repository. - Added two additional tutorials: a tutorial on visualizing higher-order networks and a case study replicating a recent paper.
- Changed the API of
degree_histogram
and addeddegree_counts
based on #23. - Refactored the
IDDegreeView
class and changed the API. Added the ability to specifyorder
and the datatype. - Added an abstract class
IDDict
to handle data validation.
Contributors: @iaciac, @leotrs, @lordgrilo, @maximelucas, @nwlandry, and @tlarock
H[attr]
now accesses hypergraph attributesH.nodes[id]
now accesses attributes, not bipartite neighbors- Removed the
__call__()
functionality fromH.nodes
andH.edges
H.nodes.memberships(id)
andH.edges.members(id)
now access the bipartite neighbors- Created base classes for the Node/Edge Views and Degree/Edge Size Views to inherit from.
- Removed the NodeDataView and EdgeDataView.
- Updated the list of developers
__getitem__()
in the NodeView and EdgeView are now in a try-except block for greater efficiency.- Removed the name attribute and fixed methods and tests to match.
- Fixed the
erdos_renyi_hypergraph()
,chung_lu_hypergraph()
, anddcsbm_hypergraph()
methods such that the number of nodes will always match the number the user specifies, even if the resulting hypergraph is disconnected. - Changed the construction method from reading in a Pandas DataFrame to using the
add_node_to_edge()
method for a 2x speedup. - Added some basic unit tests for the generative models.
- Added the dual keyword for the read_bipartite_edgelist() method.
- Added small functions to the Hypergraph class
- Added generator of random hypergraph
- Added functions for finding and removing isolates
- Refactored the
has_edge()
method in the Hypergraph class.
Contributors: @leotrs, @maximelucas, and @nwlandry