All Versions
22
Latest Version
Avg Release Cycle
137 days
Latest Release
199 days ago

Changelog History
Page 1

  • v3.6.1 Changes

    December 08, 2025

    networkx 3.6.1

    🚀 We're happy to announce the release of networkx 3.6.1!

    API Changes

    • ➕ Add spectral bipartition community finding and greedy bipartition using node swaps (#8347).

    ✨ Enhancements

    • Nodelists for from_biadjacency_matrix (#7993).
    • ➕ Add spectral bipartition community finding and greedy bipartition using node swaps (#8347).
    • Fix draw_networkx_nodes with list node_shape and add regression test (#8363).

    🐛 Bug Fixes

    • 🛠 Fix: allow graph subclasses to have additional arguments (#8369).

    📚 Documentation

    • DOC: Improve benchmarking readme (#8358).
    • 🚀 DOC: More details re: RC releases in the release process devdocs (#8346).
    • DOC: clarify difference between G.nodes/G.nodes() and G.edges/G.edges() in tutorial (#8300).
    • ✅ DOC: Add blurb to contributor guide about drawing tests (#8370).
    • 📄 DOC: Fix underline lens in docstrings (#8371).
    • ✅ Rolling back shortest paths links (#8373).

    🚧 Maintenance

    • MAINT: Replace string literal with comment (#8359).
    • ⬆️ Bump actions/checkout from 5 to 6 in the actions group (#8360).
    • 📌 pin python 3.14 to be version 3.14.0 until dataclasses are fixed (#8365).
    • Blocklist Python 3.14.1 (#8372).

    Other

    • ✅ TST: add tests for unsupported graph types in MST algorithms (#8353).
    • ✅ TST: clean up isomorphism tests (#8364).

    Contributors

    🚀 10 authors added to this release (alphabetically):

    🚀 9 reviewers added to this release (alphabetically):

    These lists are automatically generated, and may not be complete or may contain
    duplicates.

  • v3.6 Changes

    November 24, 2025

    networkx 3.6

    🚀 We're happy to announce the release of networkx 3.6!

    🆕 New Features

    • ENH: add normalize option for subgraph_centrality and its _exp version (#8340).

    API Changes

    • Replace random_lobster with random_lobster_graph (#8067).
    • Replace maybe_regular_expander with maybe_regular_expander_graph (#8050).
    • Rm networkx.algorithms.threshold.swap_d (#8213).
    • Expire deprecation of compute_v_structures (#8281).
    • Rm unused dissuade_hubs kwarg from forceatlas2 (#8293).
    • 🗄 Expire deprecation of link kwarg in node_link fns (#8282).
    • 🗄 DEP: Deprecate metric_closure (#8304).

    ✨ Enhancements

    • 👌 Improve error message for removed random_tree function (#8105).
    • ⚡️ Update approx current_flow betweenness to use k directly (#8007).
    • ⚡️ Proposal: update semantics for nonisomorphic trees with order 0 or 1 (#8083).
    • SCC benchmarks and use of G._adj in Tarjan algorithm (#8064).
    • 🐎 Performance improvement and tests for edges_equal (#8077).
    • optimise is_reachable() (#8112).
    • Optimise harmonic centrality (#8158).
    • feat(drawing): add missing connection styles in draw_networkx_edge_labels and display (#8108).
    • ⚡️ Optimizing Dijkstra's paths to target (~50x faster for graphs with multiple-hops shortest path) (#8023).
    • ➕ Add all_triangles generator yielding all unique triangles in a graph (#8135).
    • 🔨 refactor: simplify k_factor (#8139).
    • feat: add directed star graph (#8151).
    • Faster intersection_array computation for checking distance-regularity (#7181).
    • enh: short-circuit in is_regular for directed graphs (#8138).
    • Avoid re-exploring nodes in Kosaraju's SCC algorithm (#8056).
    • ✅ trust rank implementation and testing (#8165).
    • Add hyper_wiener_index function (#8184).
    • Bidirectional dijkstra optimization: from 1.1x to 25x faster (#8206).
    • Implement the algorithm to find the centroid(s) of a tree (#8089).
    • Set length threshold in FR and use np.clip (#8145).
    • ➕ Add panther++ (#4400).
    • maint: use nx.circulant_graph to generate Harary graphs (#8189).
    • ENH: add directed kwarg to edges_equal (#8192).
    • ⚡️ Optimizing Dijkstra's path construction for all targets case (#8218).
    • 🔨 ENH: adds ISMAGS support for directed and multigraph with tests and refactor (#8274).
    • Dispatch classes such as nx.Graph(backend=...) (#7760).
    • ENH: Add is_perfect_graph using SPGT (follow-up to #8111) (#8318).
    • ➕ Add benchmark suite for shortest path algorithms on weighted graphs (#8059).
    • ➕ Adding Generalized Petersen Graph (#8147).
    • ENH: add normalize option for subgraph_centrality and its _exp version (#8340).
    • FEAT: Add a describe method for some basic info about graphs (#8338).

    🐛 Bug Fixes

    • ➕ Add Python 3.14 to testing matrix (#8096).
    • 🛠 Fix round-trip to and from pygraphviz.AGraph setting spurious graph attributes (#8121).
    • ➕ Add input validation to non_randomness() and clarify its behavior (#8057).
    • Ensure that backend names are valid Python identifiers (#8160).
    • 🛠 fix: resolve failure to pickle.loads(pickle.dumps(PlanarEmbedding())) (#8186).
    • ➕ Add GEXF 1.3 to the recognized GEXF versions (#8196).
    • BUG: Raise on directed graphs in nx.find_cliques_recursive (#8211).
    • fix optimize_edit_paths handling of self-loops (#8207).
    • BUG: add check for isolated nodes in degree_sequence_tree (#8235).
    • Mehlorn Steiner Tree (#8052).
    • 🐛 BUG/MAINT: fix edge betweenness centrality scaling when k<N and merge all b.c. rescale helper functions (#8256).
    • 🛠 Fix node attributes on lattice graphs (#8311).
    • 🐛 BUG: allow graphs with nonstandard node labels in FISTA (#8332).
    • 👉 Make dominance functions consistent with definitions (#8061).
    • 📌 Pin last value to 1.0 in cumulative_distribution to address floating-point errors (#8342).

    📚 Documentation

    • Fix min_weight_matching (#8062).
    • 🚀 Update deploy-docs yml to use Python 3.12 when deploying the docs (#8102).
    • 📄 DOC: Add missing params to bfs_layout docstring (#8086).
    • ➕ Add input validation to non_randomness() and clarify its behavior (#8057).
    • 📄 doc: improve docstring for hypercube_graph (#8012).
    • 👌 Improved documentation for boundary_expansion function (#7905).
    • 📄 DOC: Add docstring example count number of unique triangles (#8144).
    • Add function bfs_labeled_edges to docs (#8149).
    • 🛠 Fix issues with urls in HITS reference docs (#8156).
    • 📄 Correct the docs for display() keyword node_pos (#8153).
    • ➕ Adding Notes on Multi-Target Shortest Path Queries (#8169).
    • 3d facebook plot example (#6893).
    • ✅ trust rank implementation and testing (#8165).
    • 👌 Improve docs for all_neighbors() (#8166).
    • ➕ Adding shortest-paths documentation (#8187).
    • ➕ Add Linux Foundation health score badge to README (#8219).
    • DOC: Add docstring for number_of_cliques (#8216).
    • DOC: add docstring for degree_sequence_tree (#8236).
    • DOC: Add examples to contracted_nodes (#7856).
    • 📄 DOC: fix wrong reference in leiden docs (#8277).
    • 🛠 Fix over-indentation of list in chordless_cycles docstring (#8288).
    • ➕ Add iplotx to network drawing documentation (#8289).
    • 🛠 Fix sphinx build errors (#8303).
    • 🚚 DOC: Move deprecation procedure from contributing->dev guide (#8308).
    • DOC: add gallery example for metric_closure (#8306).
    • 👀 Cross-link Platonic graphs in See Also section (#8307).
    • ➕ Add seealso crosslinks between lattice graphs (#8310).
    • CI,DOC: Only run one parallel betweenness example (#8305).
    • 📄 DOC: rework betweenness centrality docstrings (#8264).
    • Rm 3D layout and animation from greedy_color example (#8315).
    • DOC: Clarify node and edge removal behavior in tutorial (#8321).
    • 📄 Improving connected module docs (#8267).
    • 📄 Docs: add nx-neptune backend documentation ([#8258](https://git...

  • v3.6.rc0 Changes

    November 04, 2025

    networkx 3.6rc0

    🚀 We're happy to announce the release of networkx 3.6rc0!

    API Changes

    • Replace random_lobster with random_lobster_graph (#8067).
    • Replace maybe_regular_expander with maybe_regular_expander_graph (#8050).
    • Rm networkx.algorithms.threshold.swap_d (#8213).
    • Expire deprecation of compute_v_structures (#8281).
    • Rm unused dissuade_hubs kwarg from forceatlas2 (#8293).
    • 🗄 Expire deprecation of link kwarg in node_link fns (#8282).
    • 🗄 DEP: Deprecate metric_closure (#8304).

    ✨ Enhancements

    • 👌 Improve error message for removed random_tree function (#8105).
    • ⚡️ Update approx current_flow betweenness to use k directly (#8007).
    • ⚡️ Proposal: update semantics for nonisomorphic trees with order 0 or 1 (#8083).
    • SCC benchmarks and use of G._adj in Tarjan algorithm (#8064).
    • 🐎 Performance improvement and tests for edges_equal (#8077).
    • optimise is_reachable() (#8112).
    • Optimise harmonic centrality (#8158).
    • feat(drawing): add missing connection styles in draw_networkx_edge_labels and display (#8108).
    • ⚡️ Optimizing Dijkstra's paths to target (~50x faster for graphs with multiple-hops shortest path) (#8023).
    • ➕ Add all_triangles generator yielding all unique triangles in a graph (#8135).
    • 🔨 refactor: simplify k_factor (#8139).
    • feat: add directed star graph (#8151).
    • Faster intersection_array computation for checking distance-regularity (#7181).
    • enh: short-circuit in is_regular for directed graphs (#8138).
    • Avoid re-exploring nodes in Kosaraju's SCC algorithm (#8056).
    • ✅ trust rank implementation and testing (#8165).
    • Add hyper_wiener_index function (#8184).
    • Bidirectional dijkstra optimization: from 1.1x to 25x faster (#8206).
    • Implement the algorithm to find the centroid(s) of a tree (#8089).
    • Set length threshold in FR and use np.clip (#8145).
    • ➕ Add panther++ (#4400).
    • maint: use nx.circulant_graph to generate Harary graphs (#8189).
    • ENH: add directed kwarg to edges_equal (#8192).
    • ⚡️ Optimizing Dijkstra's path construction for all targets case (#8218).
    • 🔨 ENH: adds ISMAGS support for directed and multigraph with tests and refactor (#8274).
    • Dispatch classes such as nx.Graph(backend=...) (#7760).
    • ENH: Add is_perfect_graph using SPGT (follow-up to #8111) (#8318).
    • ➕ Add benchmark suite for shortest path algorithms on weighted graphs (#8059).

    🐛 Bug Fixes

    • ➕ Add Python 3.14 to testing matrix (#8096).
    • 🛠 Fix round-trip to and from pygraphviz.AGraph setting spurious graph attributes (#8121).
    • ➕ Add input validation to non_randomness() and clarify its behavior (#8057).
    • Ensure that backend names are valid Python identifiers (#8160).
    • 🛠 fix: resolve failure to pickle.loads(pickle.dumps(PlanarEmbedding())) (#8186).
    • ➕ Add GEXF 1.3 to the recognized GEXF versions (#8196).
    • BUG: Raise on directed graphs in nx.find_cliques_recursive (#8211).
    • fix optimize_edit_paths handling of self-loops (#8207).
    • BUG: add check for isolated nodes in degree_sequence_tree (#8235).
    • Mehlorn Steiner Tree (#8052).
    • 🐛 BUG/MAINT: fix edge betweenness centrality scaling when k<N and merge all b.c. rescale helper functions (#8256).
    • 🛠 Fix node attributes on lattice graphs (#8311).
    • 🐛 BUG: allow graphs with nonstandard node labels in FISTA (#8332).
    • 👉 Make dominance functions consistent with definitions (#8061).

    📚 Documentation

    • Fix min_weight_matching (#8062).
    • 🚀 Update deploy-docs yml to use Python 3.12 when deploying the docs (#8102).
    • 📄 DOC: Add missing params to bfs_layout docstring (#8086).
    • ➕ Add input validation to non_randomness() and clarify its behavior (#8057).
    • 📄 doc: improve docstring for hypercube_graph (#8012).
    • 👌 Improved documentation for boundary_expansion function (#7905).
    • 📄 DOC: Add docstring example count number of unique triangles (#8144).
    • Add function bfs_labeled_edges to docs (#8149).
    • 🛠 Fix issues with urls in HITS reference docs (#8156).
    • 📄 Correct the docs for display() keyword node_pos (#8153).
    • ➕ Adding Notes on Multi-Target Shortest Path Queries (#8169).
    • 3d facebook plot example (#6893).
    • ✅ trust rank implementation and testing (#8165).
    • 👌 Improve docs for all_neighbors() (#8166).
    • ➕ Adding shortest-paths documentation (#8187).
    • ➕ Add Linux Foundation health score badge to README (#8219).
    • DOC: Add docstring for number_of_cliques (#8216).
    • DOC: add docstring for degree_sequence_tree (#8236).
    • DOC: Add examples to contracted_nodes (#7856).
    • 📄 DOC: fix wrong reference in leiden docs (#8277).
    • 🛠 Fix over-indentation of list in chordless_cycles docstring (#8288).
    • ➕ Add iplotx to network drawing documentation (#8289).
    • 🛠 Fix sphinx build errors (#8303).
    • 🚚 DOC: Move deprecation procedure from contributing->dev guide (#8308).
    • DOC: add gallery example for metric_closure (#8306).
    • 👀 Cross-link Platonic graphs in See Also section (#8307).
    • ➕ Add seealso crosslinks between lattice graphs (#8310).
    • CI,DOC: Only run one parallel betweenness example (#8305).
    • 📄 DOC: rework betweenness centrality docstrings (#8264).
    • Rm 3D layout and animation from greedy_color example (#8315).
    • DOC: Clarify node and edge removal behavior in tutorial (#8321).
    • 📄 Improving connected module docs (#8267).
    • 📄 Docs: add nx-neptune backend documentation (#8258).
    • ✅ Improving shortest paths docs when there is no path between source and target (#8327).
    • ➕ Adding floating point considerations to tutorial (#8324).
    • ➕ Adding Dijkstra's algo specific doc (#8286).
    • 📚 Minor documentation build improvements (#8329).
    • Clarify the meaning of the cutoff parameter in some path-finding functions (#7487).
    • Switch to the NumFOCUS Co...

  • v3.5 Changes

    May 29, 2025

    networkx 3.5

    🚀 We're happy to announce the release of networkx 3.5!

    API Changes

    • 💾 Save Layouts on Graphs (#7571).
    • Expire d_separated and minimum_d_separator functions (#7830).
    • 🗄 Expire all_triplets deprecation (#7828).
    • 🗄 Expire random_triad deprecation (#7829).
    • 👻 DEP: Raise an exception for k_core functions with multigraphs (#7831).
    • Deprecate graph_could_be_isomorphic (#7826).
    • Expire total_spanning_tree_weight deprecation (#7843).
    • 🗄 Expire deprecation of create kwarg in nonisomorphic_trees (#7847).
    • 🆕 New draw API (#7589).

    ✨ Enhancements

    • perf: optimise random_k_out_graph (#7702).
    • Clausets local community detection algorithm (#7691).
    • find_asteroidal_triple improvement (#7736).
    • ➕ Add weight to harmonic_diameter (#7636).
    • Densest Subgraph Problem: Greedy Peeling and Greedy++ Implementations (#7731).
    • single_source_all_shortest_paths: don't loop over all nodes (#7762).
    • Error message improvement for nbunch_iter ( NetworkXError raised with specific message on TypeError with "iter" in msg ) (#7790).
    • Faster computation of energy in Laplacian centrality (#7793).
    • 👉 Make forceatlas2_layout dispatchable (#7794).
    • ⚡️ Update dispatchable for forceatlas2_layout (#7798).
    • Enable backend-only functions where NetworkX is just an API (#7690).
    • Steinertree kou enhancement in response to issue 5889 type:Enhancements (#7767).
    • ➕ Add Leiden as a backend-only algorithm (#7743).
    • Bipartite layout nodes optional (#7756).
    • ✅ Densest Subgraph Problem: FISTA based algorithm + Large scale tests (#7770).
    • Dispatch get_node_attributes and a few more from nx.classes.function (#7824).
    • Faster could_be_isomorphic and number_of_cliques (#7855).
    • ➕ Add square_clustering to algorithm benchmarks (#7857).
    • Faster Implementation of Structural Holes (#7249).
    • Improve runtime of number_of_nonisomorphic_trees() (#7917).
    • 🛠 Fix write_gexf timeformat for dynamic Graphs (#7914).
    • Consolidate could_be_isomorphic (#7852).
    • Improving rooted_tree_isomorphism for deep trees (#7945).
    • 🛠 Fixing nx.diameter inconsistent results with usebounds=True (#7954).
    • Faster square_clustering (#7810).
    • Avoid repeated cache conversion failures for backends (#7768).
    • 📜 Improve _sparse_fruchterman_reingold with L-BFGS (#7889).
    • 👌 Improve Performance of Tree Isomorphism and Center Calculation (#7946).
    • ➕ Add option for biadjacency_matrix to be returned as a dense NumPy array (#7973).
    • ➕ Add Functions for Finding Connected Dominating Sets (#7774).
    • ➕ Add feature to make storing node contraction data optional (#7902).
    • Added "initial_node" param to generate_random_paths() to allow a starting node to be specified for generated walks (#8002).
    • 🛠 Fix behavior for iterable sources argument in bfs_layers (#8013).
    • Speed up connected_components and weakly_connected_components (#7971).
    • BiRank Algorithm Implementation (#7978).
    • Enforce correct graph types for graph matchers (#8043).

    🐛 Bug Fixes

    • Update _raise_on_directed to work with create_using pos arg (#7695).
    • trophic_levels now checks for paths from each node to a basal node (#7453).
    • 🛠 Fix TSP weight parameter issues (#7721).
    • 🛠 Fix for filtered MultiGraph views from edge_subgraph (#7724) (#7729).
    • 🐛 BUG: fixed the if condition in asadpour_atsp (#7753).
    • 💅 Implement Bar ConnectionStyle for labels (#7739).
    • 🛠 Fixed a divide by zero error in forceatlas2 (#7791).
    • 🛠 Fix for issue #7645: Do not preserve 'cw' and 'ccw' attributes in PlanarEmbedding.to_undirected() (#7750).
    • 🛠 fix typo in ramanujan branch (#7804).
    • 🛠 Fix with nx.config(backend_priority=backends): (#7814).
    • Fix handling of faux_infinite values in network_simplex (#7796).
    • 🛠 Fixed the return type from an empty dict to an empty set (#7910).
    • Add edge_attrs="weight" to forceatlas2_layout dispatch decorator (#7918).
    • 🛠 Fix graph_hash iteration counts and DiGraph handling (#7834).
    • 🔨 Refactored the working of chordless_cycles to handle self loops (#7901).
    • 🛠 Fix bc scale with k endpoints (#7908).
    • 🛠 Fix BC scaling for source nodes with k and endpoints=False (#7949).
    • 🐛 BUG: graph6 format invariant to trailing newline (#7941).
    • Fix random_degree_sequence_graph when input is an iterator (#7979).
    • 👌 Improve special cases in dispatch testing (paying off tech debt) (#7982).
    • 🛠 Fix bug when assigning list to nx.config.backend_priority (#8034).
    • A minimal fix for is_aperiodic (#8029).
    • 📜 fix bug of _sparse_fruchterman_reingold and remove try/except idiom (#8041).
    • 🛠 Fix edge case in ISMAGS symmetry detection (#8055).

    📚 Documentation

    • set nx-arangodb link to github (#7694).
    • 👍 Re-submission of gh-7087 with better file provenance (#7681).
    • 🛠 Fix code formatting of some examples (#7730).
    • Add examples for custom graph in the doc of soft_random_geometric_graph and thresholded_random_geometric_graph (#7749).
    • Gallery example: bipartite a/b-core motif (#7757).
    • ➕ Add blurb about pytest-mpl dependency to contributing guide (#7741).
    • Minor updates to single_source_shortest_path_length docstring (#7637).
    • ➕ Added a note to the contributor guideline to avoid numpy scalars as a… (#7773).
    • Correcting the example given under subgraph_is_monomorphic.py (#7779).
    • [easy] Add to Contributor List (#7801).
    • ⚡️ doc: mention the second major update (#7782).
    • DOC: Add details about more grants (#7823).
    • 🔨 Refactor: Moving backend docs from backends.py to backends.rst (#7776).
    • ⚡️ Update readwrite docstrings for the path parameter (#7835).
    • Fix docstring example of nx.generate_random_paths(index_map=...) (#7832).
    • ➕ Adds NVIDIA Corporation to list of supporters (#7846).
    • 🛠 Fix use of triple backticks in docstrings ([#7845](https://github.com/networkx/net...

  • v3.5.rc0 Changes

    May 09, 2025

    networkx 3.5rc0

    🚀 We're happy to announce the release of networkx 3.5rc0!

    API Changes

    • 💾 Save Layouts on Graphs (#7571).
    • Expire d_separated and minimum_d_separator functions (#7830).
    • 🗄 Expire all_triplets deprecation (#7828).
    • 🗄 Expire random_triad deprecation (#7829).
    • 👻 DEP: Raise an exception for k_core functions with multigraphs (#7831).
    • Deprecate graph_could_be_isomorphic (#7826).
    • Expire total_spanning_tree_weight deprecation (#7843).
    • 🗄 Expire deprecation of create kwarg in nonisomorphic_trees (#7847).
    • 🆕 New draw API (#7589).

    ✨ Enhancements

    • perf: optimise random_k_out_graph (#7702).
    • Clausets local community detection algorithm (#7691).
    • find_asteroidal_triple improvement (#7736).
    • ➕ Add weight to harmonic_diameter (#7636).
    • Densest Subgraph Problem: Greedy Peeling and Greedy++ Implementations (#7731).
    • single_source_all_shortest_paths: don't loop over all nodes (#7762).
    • Error message improvement for nbunch_iter ( NetworkXError raised with specific message on TypeError with "iter" in msg ) (#7790).
    • Faster computation of energy in Laplacian centrality (#7793).
    • 👉 Make forceatlas2_layout dispatchable (#7794).
    • ⚡️ Update dispatchable for forceatlas2_layout (#7798).
    • Enable backend-only functions where NetworkX is just an API (#7690).
    • Steinertree kou enhancement in response to issue 5889 type:Enhancements (#7767).
    • ➕ Add Leiden as a backend-only algorithm (#7743).
    • Bipartite layout nodes optional (#7756).
    • ✅ Densest Subgraph Problem: FISTA based algorithm + Large scale tests (#7770).
    • Dispatch get_node_attributes and a few more from nx.classes.function (#7824).
    • Faster could_be_isomorphic and number_of_cliques (#7855).
    • ➕ Add square_clustering to algorithm benchmarks (#7857).
    • Faster Implementation of Structural Holes (#7249).
    • Improve runtime of number_of_nonisomorphic_trees() (#7917).
    • 🛠 Fix write_gexf timeformat for dynamic Graphs (#7914).
    • Consolidate could_be_isomorphic (#7852).
    • Improving rooted_tree_isomorphism for deep trees (#7945).
    • 🛠 Fixing nx.diameter inconsistent results with usebounds=True (#7954).
    • Faster square_clustering (#7810).
    • Avoid repeated cache conversion failures for backends (#7768).
    • 📜 Improve _sparse_fruchterman_reingold with L-BFGS (#7889).
    • 👌 Improve Performance of Tree Isomorphism and Center Calculation (#7946).
    • ➕ Add option for biadjacency_matrix to be returned as a dense NumPy array (#7973).
    • ➕ Add Functions for Finding Connected Dominating Sets (#7774).
    • ➕ Add feature to make storing node contraction data optional (#7902).
    • Added "initial_node" param to generate_random_paths() to allow a starting node to be specified for generated walks (#8002).
    • 🛠 Fix behavior for iterable sources argument in bfs_layers (#8013).
    • Speed up connected_components and weakly_connected_components (#7971).

    🐛 Bug Fixes

    • Update _raise_on_directed to work with create_using pos arg (#7695).
    • trophic_levels now checks for paths from each node to a basal node (#7453).
    • 🛠 Fix TSP weight parameter issues (#7721).
    • 🛠 Fix for filtered MultiGraph views from edge_subgraph (#7724) (#7729).
    • 🐛 BUG: fixed the if condition in asadpour_atsp (#7753).
    • 💅 Implement Bar ConnectionStyle for labels (#7739).
    • 🛠 Fixed a divide by zero error in forceatlas2 (#7791).
    • 🛠 Fix for issue #7645: Do not preserve 'cw' and 'ccw' attributes in PlanarEmbedding.to_undirected() (#7750).
    • 🛠 fix typo in ramanujan branch (#7804).
    • 🛠 Fix with nx.config(backend_priority=backends): (#7814).
    • Fix handling of faux_infinite values in network_simplex (#7796).
    • 🛠 Fixed the return type from an empty dict to an empty set (#7910).
    • Add edge_attrs="weight" to forceatlas2_layout dispatch decorator (#7918).
    • 🛠 Fix graph_hash iteration counts and DiGraph handling (#7834).
    • 🔨 Refactored the working of chordless_cycles to handle self loops (#7901).
    • 🛠 Fix bc scale with k endpoints (#7908).
    • 🛠 Fix BC scaling for source nodes with k and endpoints=False (#7949).
    • 🐛 BUG: graph6 format invariant to trailing newline (#7941).
    • Fix random_degree_sequence_graph when input is an iterator (#7979).
    • 👌 Improve special cases in dispatch testing (paying off tech debt) (#7982).

    📚 Documentation

    • set nx-arangodb link to github (#7694).
    • 👍 Re-submission of gh-7087 with better file provenance (#7681).
    • 🛠 Fix code formatting of some examples (#7730).
    • Add examples for custom graph in the doc of soft_random_geometric_graph and thresholded_random_geometric_graph (#7749).
    • Gallery example: bipartite a/b-core motif (#7757).
    • ➕ Add blurb about pytest-mpl dependency to contributing guide (#7741).
    • Minor updates to single_source_shortest_path_length docstring (#7637).
    • ➕ Added a note to the contributor guideline to avoid numpy scalars as a… (#7773).
    • Correcting the example given under subgraph_is_monomorphic.py (#7779).
    • [easy] Add to Contributor List (#7801).
    • ⚡️ doc: mention the second major update (#7782).
    • DOC: Add details about more grants (#7823).
    • 🔨 Refactor: Moving backend docs from backends.py to backends.rst (#7776).
    • ⚡️ Update readwrite docstrings for the path parameter (#7835).
    • Fix docstring example of nx.generate_random_paths(index_map=...) (#7832).
    • ➕ Adds NVIDIA Corporation to list of supporters (#7846).
    • 🛠 Fix use of triple backticks in docstrings (#7845).
    • ➕ Add paragraph about university classes to mentored projects (#7838).
    • 🛠 Fix pygraphviz_layout example (#7849).
    • ➕ Add test-extras to optional dependencies (#7854).
    • doc: hash size are in bytes (#7866).
    • 🚚 DOC: Clean up mentored projects page: move visualization project to completed section (#7881).
    • ➕ added 2 projects for GSoC 2025 ([#7880](https://github.com/networkx/netw...

  • v3.4.2 Changes

    October 21, 2024

    NetworkX 3.4.2

    🚀 We're happy to announce the release of networkx 3.4.2!

    🐛 Bug Fixes

    • 🛠 Fix docstrings of dispatchable functions (#7679).
    • Fix draw_networkx_nodes return type (#7685).

    📚 Documentation

    • ➕ Add disclaimer about LLM driven PRs (#7683).

    🚧 Maintenance

    • 🛠 Fix doc warnings from recently added docs (#7682).

    Contributors

    🚀 6 authors added to this release (alphabetically):

    🚀 4 reviewers added to this release (alphabetically):

    These lists are automatically generated, and may not be complete or may contain duplicates.

  • v3.4.1 Changes

    October 11, 2024

    networkx 3.4.1

    🚀 We're happy to announce the release of networkx 3.4.1!

    🚧 Maintenance

    • ✂ Remove old deprecation decorator (#7669).
    • MAINT: delay loading of backend_info to after imports (#7672).

    Contributors

    🚀 2 authors added to this release (alphabetically):

    🚀 3 reviewers added to this release (alphabetically):

    These lists are automatically generated, and may not be complete or may contain duplicates.

  • v3.4 Changes

    October 10, 2024

    networkx 3.4

    🚀 We're happy to announce the release of networkx 3.4!

    API Changes

    • 🗄 Expires the forest_str deprecation (#7414).
    • 🗄 [ENH, BUG]: added colliders and v_structures and deprecated compute_v_structures in dag.py (#7398).
    • 🗄 Expires the random_tree deprecation (#7415).
    • Expire deprecation for strongly_connected_components_recursive (#7420).
    • Expire deprecated sort_neighbors param in generic_bfs_edges (#7417).
    • 🗄 Rm deprecated normalized param from s_metric (#7418).
    • 🗄 Expire deprecated nx.join in favor of join_trees (#7419).
    • ✂ Remove depercated Edmonds class for 3.4 (#7447).
    • ✂ Remove deprecated MultiDiGraph_EdgeKey for 3.4 (#7448).
    • ➕ Add edges keyword/deprecate link keyword arguments in JSON input-output (#7565).
    • 🔗 Revert breaking change to node_link_* link defaults (#7652).

    ✨ Enhancements

    • Add a nodelist feature to from_numpy_array (#7412).
    • Prioritize edgelist representations in to_networkx_graph (#7424).
    • ➕ Adds initial debug logging calls to _dispatchable (#7300).
    • ➕ add: nodes attribute is modifiable (#7532).
    • Enable config to be used as context manager (#7363).
    • ➕ Added code to handle multi-graph in mst (#7454).
    • 0️⃣ Enable caching by default (#7498).
    • #7546 More detail error message for pydot (#7558).
    • Fix weakly_connected_components() performance on graph view (#7586).
    • 👮 Forceatlas2 (#7543).
    • avoid iteration and use boolean indexing (#7591).
    • Hide edges with a weight of None in simple_paths (#7583).
    • 👌 Improved running time for harmonic centrality (#7595).
    • ➕ Add remove attribute functions (#7569).
    • 🌲 Log "can/should run" and caching in dispatch machinery (#7568).
    • Individualize drawing attributes (#7570).
    • ➕ added nx-parallel gsoc project (#7620).
    • Harmonic diameter (#5251).
    • 👍 Allow dispatch machinery to fall back to networkx (#7585).
    • ➕ Add create_using parameter for random graphs (#5672).
    • ➕ Add config option to disable warning when using cached value (#7497).

    🐛 Bug Fixes

    • Fix graph name attribute for complete_bipartite_graph (#7399).
    • Remove import warnings during to_networkx_graph conversion (#7426).
    • Fix nx.from_pandas_edgelist so edge keys are not added as edge attributes and edge keys (#7445).
    • Fix from_pandas_edgelist for MultiGraph given edge_key (#7466).
    • 🛠 Fix dispatch tests when using numpy 2 (#7506).
    • 🗄 [ENH, BUG]: added colliders and v_structures and deprecated compute_v_structures in dag.py (#7398).
    • 🛠 Fix reading edgelist when delimiter is whitespace, e.g. tab (#7465).
    • Ensure we always raise for unknown backend in backend= (#7494).
    • Prevent to_agraph from modifying graph argument (#7610).
    • Implementing iterative removal of non_terminal_leaves in Steiner Tree approximation (#7422).
    • Only allow connected graphs in eigenvector_centrality_numpy (#7549).
    • CI: Fix typo in nightly run pip install (#7625).

    📚 Documentation

    • ✅ Document missing shortest_path functions (#7394).
    • Optimal Edit Paths Return Section Improved (#7375).
    • ⚡️ Minor updates to simple_cycles docstring (#7421).
    • 👻 DOC: Clarifying NetworkXPointlessConcept exception (#7434).
    • ⚡️ DOC: updated pairs.py (#7416).
    • ➕ Add docstring example for directed tree (#7449).
    • Change docs of shortest_path_length so return is number instead of int (#7477).
    • 👉 Use intersphinx_registry to manage intersphinx mapping (#7481).
    • 📄 Ma: fix some spelling errors in docs (#7480).
    • ⚡️ Update NetworkX reference links in doc index (#7500).
    • ⚡️ strong product docs update (#7511).
    • 🔨 Refactoring and enhancing user-facing Backend and Configs docs (#7404).
    • 🛠 Fixed the citation in dominance.py [Issue #7522] (#7524).
    • Clarify generation number in dorogovtsev_goltsev_mendes_graph() (#7473).
    • ➕ Add Introspection section to backends docs (#7556).
    • DOC: Added default_config in get_info's description (#7567).
    • Prettify README.rst (#7514).
    • DOC: Fix typo in the code snippet provided in the docstring of nx_pydot.pydot_layout() (#7572).
    • 🛠 Fix installation instructions for default extras in README (#7574).
    • ➕ Add missing metadata to v3.3 release notes (#7592).
    • Correct the members of steering council (#7604).
    • 🛠 Fix dispatch docs formatting (#7619).
    • ➕ Add to Contributor List (#7621).
    • Example fix for issue 7633 (#7634).
    • 🛠 Fix: Correct community color assignment in Girvan-Newman community detection (#7644).
    • Updated docstring for generators/karate_club_graph() (#7626).
    • 📚 Updates documentation to include details about using NetworkX with backends (#7611).
    • Add examples section to to_scipy_sparse_array (#7627).
    • ➕ Add examples to docstrings of subgraph_(iso/monomorphism) methods (#7622).

    🚧 Maintenance

    • Simplify flow func augmentation logic in connectivity module (#7367).
    • ✅ A few more doctest skips for mpl/np dependencies (#7403).
    • ✂ Remove repetitive words (#7406).
    • FilterAdjacency: len is recalculated unnecessarily #7377 (#7378).
    • ➕ Add check for empty graphs in flow_hierarchy (#7393).
    • Use nodelist feature of from_numpy_array (#7425).
    • 📦 Cleanup remaining usages of deprecated random_tree in package (#7411).
    • ➕ Add check for empty graphs in non_randomness (#7395).
    • ⚡️ Update tests for macOS Sonoma v14 (#7437).
    • ⚡️ Update doc requirements (#7435).
    • ⚡️ Update pygraphviz (#7441).
    • 👍 Always cache graph attrs for better cache behavior (#7455).
    • retain adjacency order in nx-loopback copy of networkx graph (#7432).
    • DEV: Add files generated by benchmarking to .gitignore (#7461).
    • ✂ Remove redundant graph copy in algorithms.bridges.bridges() (#7471).
    • CI: Add G...

  • v3.4.rc0 Changes

    October 06, 2024

    networkx 3.4rc0

    🚀 We're happy to announce the release of networkx 3.4rc0!

    API Changes

    • 🗄 Expires the forest_str deprecation (#7414).
    • 🗄 [ENH, BUG]: added colliders and v_structures and deprecated compute_v_structures in dag.py (#7398).
    • 🗄 Expires the random_tree deprecation (#7415).
    • Expire deprecation for strongly_connected_components_recursive (#7420).
    • Expire deprecated sort_neighbors param in generic_bfs_edges (#7417).
    • 🗄 Rm deprecated normalized param from s_metric (#7418).
    • 🗄 Expire deprecated nx.join in favor of join_trees (#7419).
    • ✂ Remove depercated Edmonds class for 3.4 (#7447).
    • ✂ Remove deprecated MultiDiGraph_EdgeKey for 3.4 (#7448).
    • ➕ Add edges keyword/deprecate link keyword arguments in JSON input-output (#7565).
    • 🔗 Revert breaking change to node_link_* link defaults (#7652).

    ✨ Enhancements

    • Add a nodelist feature to from_numpy_array (#7412).
    • Prioritize edgelist representations in to_networkx_graph (#7424).
    • ➕ Adds initial debug logging calls to _dispatchable (#7300).
    • ➕ add: nodes attribute is modifiable (#7532).
    • Enable config to be used as context manager (#7363).
    • ➕ Added code to handle multi-graph in mst (#7454).
    • 0️⃣ Enable caching by default (#7498).
    • #7546 More detail error message for pydot (#7558).
    • Fix weakly_connected_components() performance on graph view (#7586).
    • 👮 Forceatlas2 (#7543).
    • avoid iteration and use boolean indexing (#7591).
    • Hide edges with a weight of None in simple_paths (#7583).
    • 👌 Improved running time for harmonic centrality (#7595).
    • ➕ Add remove attribute functions (#7569).
    • 🌲 Log "can/should run" and caching in dispatch machinery (#7568).
    • Individualize drawing attributes (#7570).
    • ➕ added nx-parallel gsoc project (#7620).
    • Harmonic diameter (#5251).
    • 👍 Allow dispatch machinery to fall back to networkx (#7585).

    🐛 Bug Fixes

    • Fix graph name attribute for complete_bipartite_graph (#7399).
    • Remove import warnings during to_networkx_graph conversion (#7426).
    • Fix nx.from_pandas_edgelist so edge keys are not added as edge attributes and edge keys (#7445).
    • Fix from_pandas_edgelist for MultiGraph given edge_key (#7466).
    • 🛠 Fix dispatch tests when using numpy 2 (#7506).
    • 🗄 [ENH, BUG]: added colliders and v_structures and deprecated compute_v_structures in dag.py (#7398).
    • 🛠 Fix reading edgelist when delimiter is whitespace, e.g. tab (#7465).
    • Ensure we always raise for unknown backend in backend= (#7494).
    • Prevent to_agraph from modifying graph argument (#7610).
    • Implementing iterative removal of non_terminal_leaves in Steiner Tree approximation (#7422).
    • Only allow connected graphs in eigenvector_centrality_numpy (#7549).
    • CI: Fix typo in nightly run pip install (#7625).

    📚 Documentation

    • ✅ Document missing shortest_path functions (#7394).
    • Optimal Edit Paths Return Section Improved (#7375).
    • ⚡️ Minor updates to simple_cycles docstring (#7421).
    • 👻 DOC: Clarifying NetworkXPointlessConcept exception (#7434).
    • ⚡️ DOC: updated pairs.py (#7416).
    • ➕ Add docstring example for directed tree (#7449).
    • Change docs of shortest_path_length so return is number instead of int (#7477).
    • 👉 Use intersphinx_registry to manage intersphinx mapping (#7481).
    • 📄 Ma: fix some spelling errors in docs (#7480).
    • ⚡️ Update NetworkX reference links in doc index (#7500).
    • ⚡️ strong product docs update (#7511).
    • 🔨 Refactoring and enhancing user-facing Backend and Configs docs (#7404).
    • 🛠 Fixed the citation in dominance.py [Issue #7522] (#7524).
    • Clarify generation number in dorogovtsev_goltsev_mendes_graph() (#7473).
    • ➕ Add Introspection section to backends docs (#7556).
    • DOC: Added default_config in get_info's description (#7567).
    • Prettify README.rst (#7514).
    • DOC: Fix typo in the code snippet provided in the docstring of nx_pydot.pydot_layout() (#7572).
    • 🛠 Fix installation instructions for default extras in README (#7574).
    • ➕ Add missing metadata to v3.3 release notes (#7592).
    • Correct the members of steering council (#7604).
    • 🛠 Fix dispatch docs formatting (#7619).
    • ➕ Add to Contributor List (#7621).
    • Example fix for issue 7633 (#7634).
    • 🛠 Fix: Correct community color assignment in Girvan-Newman community detection (#7644).
    • Updated docstring for generators/karate_club_graph() (#7626).

    🚧 Maintenance

    • Simplify flow func augmentation logic in connectivity module (#7367).
    • ✅ A few more doctest skips for mpl/np dependencies (#7403).
    • ✂ Remove repetitive words (#7406).
    • FilterAdjacency: len is recalculated unnecessarily #7377 (#7378).
    • ➕ Add check for empty graphs in flow_hierarchy (#7393).
    • Use nodelist feature of from_numpy_array (#7425).
    • 📦 Cleanup remaining usages of deprecated random_tree in package (#7411).
    • ➕ Add check for empty graphs in non_randomness (#7395).
    • ⚡️ Update tests for macOS Sonoma v14 (#7437).
    • ⚡️ Update doc requirements (#7435).
    • ⚡️ Update pygraphviz (#7441).
    • 👍 Always cache graph attrs for better cache behavior (#7455).
    • retain adjacency order in nx-loopback copy of networkx graph (#7432).
    • DEV: Add files generated by benchmarking to .gitignore (#7461).
    • ✂ Remove redundant graph copy in algorithms.bridges.bridges() (#7471).
    • 📦 CI: Add GitHub artifact attestations to package distribution (#7459).
    • ➕ Add polynomials.py to needs_numpy (#7493).
    • MAINT: Rename LoopbackDispatcher to LoopbackBackendInterface and dispatcher to backend_interface (#7492).
    • ⚡️ CI: update action that got moved org (#7503).
    • ⚡️ Update momepy (#7507).
    • 🛠 Fix pygraphviz install on Windows ([#7512](htt...

  • v3.3 Changes

    April 06, 2024

    networkx 3.3

    🚀 We're happy to announce the release of networkx 3.3!

    API Changes

    • Disallow negative number of nodes in complete_multipartite_graph (#7057).
    • 🗄 DEP: Deprecate the all_triplets one-liner (#7060).
    • [A-star] Added expansion pruning via cutoff if cutoff is provided (#7073).
    • 👉 Make HITS raise exceptions consistent with power iterations (#7084).
    • 🗄 DEP: Deprecate random_triad (#7061).
    • ➕ Added feature modular graph product (#7227).
    • ENH: Speed up common/non_neighbors by using _adj dict operations (#7244).
    • 🗄 Deprecate the create argument of nonisomorphic_trees (#7316).
    • Improve total_spanning_tree_weight (#7100).
    • Update init.py (#7320).
    • add **kwargs to traveling_salesman_problem (#7371).

    ✨ Enhancements

    • ➕ Add Tadpole graph (#6999).
    • [A-star] Added expansion pruning via cutoff if cutoff is provided (#7073).
    • 💅 Implementation of $S1$ model (#6858).
    • [Feat] Random expanders utilities (#6761).
    • ✅ Compare graphs for generator functions when running tests with backend (#7066).
    • ➕ Add Kirchhoff index / Effective graph resistance (#6926).
    • 🔄 Changed return types of shortest path methods to improve consistency (#6584).
    • 🆕 New PR for Fixes minimal d-separator function failing to handle cases where no d-separators exist (#7019).
    • ENH : Provide non-normalized and normalized directed laplacian matrix calculation (#7199).
    • feat: drop the use of node attribute "first_nbr" in PlanarEmbedding (#7202).
    • ➕ Add functions to compute Schultz and Gutman Index (#3709).
    • Divisive community algorithms (#5830).
    • ➕ Added feature modular graph product (#7227).
    • ENH : added sort_neighbors to all functions in depth_first_search.py (#7196).
    • 🆕 New graph generator for the Kneser graph (#7146).
    • Draw MultiDiGraph edges and labels qa7008 (#7010).
    • 👉 Use github actions to run a comparison benchmark (#7268).
    • BFS layout implementation (#5179).
    • Add max_level= argument to louvain_communities to limit macro-iterations (#6909).
    • ⚡️ Review and update @nx._dispatchable usage since 3.2.1 (#7302).
    • Transmogrify _dispatchable objects into functions (#7298).
    • fix: make PlanarEmbedding.copy() use add_edges_from() from parent (closes #7223) (#7224).
    • 👍 Allow seed of np.random instance to exactly produce arbitrarily large integers (#6869).
    • Improve total_spanning_tree_weight (#7100).
    • add seed to nx.generate_random_paths (#7332).
    • 👍 Allow backends to implement should_run (#7257).
    • ➕ Adding tree broadcasting algorithm in a new module (#6928).
    • Option to include initial labels in weisfeiler_lehman_subgraph_hashes (#6601).
    • ➕ Add better error message when trying to get edge that is not present (#7245).
    • Make is_negatively_weighted dispatchable (#7352).
    • ➕ Add option to hide or show tick labels (#6018).
    • ENH: Cache graphs objects when converting to a backend (#7345).

    🐛 Bug Fixes

    • 🛠 Fix listing of release notes on Releases page (#7030).
    • 🛠 Fix syntax warning from bad escape sequence (#7034).
    • 🛠 Fix triangles to avoid using is to compare nodes (#7041).
    • 🛠 Fix error message for nx.mycielski_graph(0) (#7056).
    • Disallow negative number of nodes in complete_multipartite_graph (#7057).
    • Handle edge cases for greedy_modularity_communities (#6973).
    • 🛠 FIX: Match the doc description while copying over data (#7092).
    • fix: Include singleton/trivial paths in all_simple_paths & other functions (#6694).
    • Dinitz correction (#6968).
    • ✅ Modify GML test to fix invalid octal character warning (#7159).
    • Fix random_spanning_tree() for single node and empty graphs (#7211).
    • ⚡️ PlanarEmbedding.remove_edge() now updates removed edge's neighbors (#6798).
    • ➕ add seed to graph creation (#7241).
    • add seed to tests of fast_label_propatation_communities (#7242).
    • Fix rich_club_coefficient() for single node and empty graphs (#7212).
    • Fix minimum_spanning_arborescence regression (#7280).
    • 💅 Move arrowstyle input munging after intput validation (#7293).
    • 🛠 Fix empty GraphML attribute is not parsed (#7319).
    • Add new test result to test_asadpour_tsp and change linprog method (#7335).
    • 🛠 Fix custom weight attribute for Mehlhorn (#6681).

    📚 Documentation

    • 🚀 Update release process (#7029).
    • ⚡️ Update convert_matrix.py (#7018).
    • 🛠 fix extendability function name in bipartite.rst (#7042).
    • 🏗 Minor doc cleanups to remove doc build warnings (#7048).
    • DOC: Add example to generic_bfs_edges to demonstrate the neighbors param (#7072).
    • Hierarchical clustering layout gallery example (#7058).
    • 🛠 Fixed an error in the documentation of the katz centrality (#6294).
    • Create 3d_rotation_anime.py (#7025).
    • 📄 DOC: Add docstrings to filter view functions (#7086).
    • 📄 DOC: Add docstrings to Filter mapping views (#7075).
    • 📄 DOCS: Fix internal links to other functions in isomorphvf2 (#6706).
    • ➕ added note for the triangle inequality case in TSP (#6995).
    • ➕ Add note about importance of testing to contributor guide (#7103).
    • Proposal to add centrality overview to mentored projects (#7104).
    • 👌 Improve documentation of Component Algorithms (#5473).
    • ➕ Add dot io to readwrite (#5061).
    • ➕ Add Python versions to release notes (#7113).
    • 📄 DOC: Turn on inline plots in graph generators docstrings (#6401).
    • 🛠 Fix duplicate numbering in contributor guide (#7116).
    • DOC: remove unnecessary 'or' in planted_partition_graph (#7115).
    • DOC: Link methods in functions to base Graph methods/properties (#7125).
    • Connect docs to doc_string for total_spanning_tree_weight (#7098).
    • Image (3D RGB data) segmentation by spectral clustering with 3D illustrations (#7040).
    • 📚 update triadic_census documentation for undirected graphs - issue 4386 ([#7141](https://github.com/ne...