Changelog

v0.9.2

Added nonuniform centralities and fixed the node_edge_centrality method #600 (@nwlandry).
Fixed Issue #615. Added a docstring to the watts_strogatz_hypergraph method #618 (@nwlandry).
Made the HOW_TO_CONTRIBUTE guide more friendly to new contributors and accurate to our current release process #614 (@nwlandry).
Fixed the local_clustering_coefficient method #608 (@cosimoagostinelli).
Created an initial sphinx-gallery #609 (@maximelucas).
Changed aggregate stats type from numpy ints and floats to Python ints and floats and added unique method. #603 (@nwlandry).

v0.9.1

Fixed Issue #361 for random_hypergraph #597 (@nwlandry).
Added the ability to generate Erdős-Rényi hypergraphs without multiedges #596 (@nwlandry).
Added support for Python 3.13 #599 (@nwlandry).
Added a module for simpliciality measures #587 (@nwlandry).
Tweaked the HIF read/write functions to match updates #598 (@nwlandry).
Added a conditional API note #595 (@nwlandry).
Fixed broken links in the README #594 (@nwlandry).

v0.9

Added methods to (1) cut networks by order (cut_to_order) and (2) generate the k-skeleton of a SC (k_skeleton) #578 (@thomasrobiglio).
Fixed an issue where the latest version displayed "dev" #592 (@nwlandry).
Fixed the issue in the XGI version switcher so it no longer displays "Choose version" #590 (@nwlandry).
Added the ability to switch between XGI versions on ReadTheDocs, added a favicon, added a theme-dependent logo, added the ability to auto-retrieve the license date range, and the ability to auto-retrieve the date of the latest release #586 (@nwlandry).
Fixed Issue #580 and added related unit tests #581 (@thomasrobiglio).
Added the ability to change the website color themes (light, dark, auto), removed unnecessary files from the PyData migration, and fixed Sphinx errors #582 (@nwlandry).
Fixed broken links on the PyPI landing page from relative paths #579 (@nwlandry).

v0.8.10

Migrating to the PyData theme for the online docs #576 (@maximelucas).

v0.8.9

Added a new recipe to address Issue #549 #574 (@nwlandry).
Added the edge_ec argument to the draw method to specify edge colors #575 (@maximelucas).
Switched from setup.py to pyproject.toml for installation #573 (@nwlandry).

v0.8.8

Added the ability to read and write files according the HIF functionality standard #572 (@nwlandry).
Implemented the add_node_to_edge and remove_node_from_edge methods for DiHypergraphs #571 (@nwlandry).
Allow empty edges #565 (@nwlandry).
Simplified the cleanup() methods #569 (@nwlandry).
Fix Issue #566 #567 (@maximelucas).
Added documentation about N vs. N-1 in the var() and std() methods in the stats module #562 (@nwlandry).
Fix Issue #552 #561 (@nwlandry).

v0.8.7

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).

v0.8.6

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).

v0.8.5

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 the exact=True keyword reproducible #526 (@nwlandry).

v0.8.4

Added the ability to supply user-defined functions to filterby and filterby_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 and argmin 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).

v0.8.3

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).

v0.8.2

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).

v0.8.1

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).

v0.8

Made IDViews respect edge insertion order in the from_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 the neighbors 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).

v0.7.4

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).

v0.7.3

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 to DiHypergraph and SimplicialComplex and added a connected argument to the Hypergraph 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 to to_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).

v0.7.2

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).

v0.7.1

Fix: Converting from a SimplicialComplex to a Hypergraph now adds all of its faces to the hypergraph, not just the maximal faces. Added unit tests for converting between Hypergraph and SimplicialComplex 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 to HOW_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).

v0.7

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() and memberships() methods so that views are read-only #371 (@nwlandry).
Added shortest path algorithms #368 (@acombretrenouard).

v0.6

Added new drawing layouts (circular_layout, spiral_layout, and barycenter_kamada_kawai_layout) #360 (@thomasrobiglio).

v0.5.8

Formatted the codebase with ruff #346 (@leotrs).
Improved the performance of flag_complex and random_flag_complex #355 (@maximelucas).
Moved the IDDict class to utilities and removed unnecessary arguments in the IDView 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 the maximal() method #332 (@nwlandry).
Feature: added trivial hypergraph #335 (@maximelucas).

v0.5.7

Changed the organization name to xgi-org and removed codecov from the test requirements file #334 (@nwlandry).
Fix: improved the degree_counts documentation #329 (@maximelucas).
Added the maximal() method to EdgeView, removed the maximal_simplices() method, and removed a bug from the duplicates() 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).

v0.5.6

Renamed convert_to_line_graph() to to_line_graph(), added an s 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).

v0.5.5

Refactored incidence_matrix() for ~4x speedup, made output consistent for empty matrices, added tests, and refactored multiorder_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. Added pca_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 the vector_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() and uniform_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).

v0.5.4

Fixed issue #270 #271 (@nwlandry).
Fixed the centralities so that they are positive and 1-normalized #274 (@nwlandry).

v0.5.3

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).

v0.5.2

Added the find_triangles() and flag_complex_d2() functions. The flag_complex_d2() function is much faster than flag_complex() for simplicial complexes of max order 2. Also refactored random_flag_complex_d2() to use flag_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).

v0.5.1

draw() now correctly plots simplicial complexes with the max_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() and random_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 match add_hyperedges_from() #220 (@maximelucas).

v0.5.0

Fixed #214, added a powerset() function, added a subfaces() 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() and incidence_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 to xgi.flag_complex() (#200).
Fixed Issue #198 (#199).
Refactored load_xgi_data() to call dict_to_hypergraph() and fixed a bug in dict_to_hypergraph() (#193).
Added num_edges_order() to get the number of edges of a given order and added an order parameter to the degree_counts() function (#192).
Fixed #182 and #186 by adding a max_order argument to draw() and load_xgi_data() (#173) .
Made draw() faster by refactoring _color_arg_to_dict() and _scalar_arg_to_dict() (#173).

Contributors: @leotrs, @maximelucas, and @nwlandry

v0.4.3

Hypergraph.has_edge is now IDView.lookup, Hypergraph.duplicate_edges is now IDView.duplicates, and utilities.convert_labels_to_integer is now function.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

v0.4.2

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

v0.4.1

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() and singletons() method from the Hypergraph class to the NodeView and EdgeView classes respectively (#146).
Fixed Hypergraph.copy() to not use the subhypergraph method (#145).
filterby() now accepts NodeStat and EdgeStat 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

v0.4

Added the stats package which implements NodeStat, EdgeStat and related functionality. This package now handles computation of edge size and degree (#120).
Removed the EdgeSizeView and DegreeView 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).
Moved the neighbors method in the Hypergraph class to the IDView class so that node and edge neighbors are now supported (PR #129).
Added a centrality module and added these methods to nodestats.py and edgestats.py (#130).
Moved the load_xgi_data method to the readwrite 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 and weighted_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

v0.3.1

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, and has_node)
Refactored the members() method as well as the NodeView and EdgeView 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

v0.3

Added the ability to convert to and from a NetworkX bipartite graph.
Removed the shape property from Hypergraph and renamed the number_of_nodes() and number_of_edges() methods to the num_nodes and num_edges properties, respectively.
Added random seed decorator as in NetworkX.
Added a SimplicialComplex class.
Added order and weighted arguments to the incidence_matrix and adjacency_matrix functions.
Added an intersection_profile function.
Added a laplacian function with argument order and a multiorder_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 added degree_counts based on #23.
Refactored the IDDegreeView class and changed the API. Added the ability to specify order and the datatype.
Added an abstract class IDDict to handle data validation.

Contributors: @iaciac, @leotrs, @lordgrilo, @maximelucas, @nwlandry, and @tlarock

v0.2

H[attr] now accesses hypergraph attributes
H.nodes[id] now accesses attributes, not bipartite neighbors
Removed the__call__() functionality from H.nodes and H.edges
H.nodes.memberships(id) and H.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(), and dcsbm_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.