Skip to main content
The OCGraph libraries provide graph analytics for the Ocient® System through both Python® (ocient_graph using pyocient) and Java® (using the JDBC driver). Changes listed here apply to both libraries unless noted otherwise. Python function names use snake case snake_case and Java method names use camel case camelCase. For library-specific reference documentation, see OCGraph Python Library and OCGraph Java Library.

1.1.0

  • Added these 15 new graph algorithms:
    • a_star_shortest_path
    • cosine_similarity
    • cycle_detection
    • eigenvector_centrality
    • fast_random_projection
    • jaccard_similarity
    • k_core_decomposition
    • leiden_modularity_optimization
    • louvain_modularity_optimization
    • max_bipartite_matching
    • random_walk
    • stable_marriage
    • temporal_page_rank
    • time_respecting_shortest_path
    • yens_k_shortest_paths
    • For details, see OCGraph Python Library and OCGraph Java Library.
  • Added the ExecutionMode enum, the set_execution_config function (Python), and the setExecutionConfig method (Java). These items control whether graph algorithms run using the legacy JDBC or SQL batch implementation or, if the connected Ocient System supports it, an Ocient data flow implementation. The default, AUTO_FASTEST, automatically selects the faster option. For details, see OCGraph Python Library and OCGraph Java Library.
  • Added execution backed by data flows to the aggregate_messages (Python) and aggregateMessages (Java) functions.
  • Removed the max_iterations argument from the connected_components (Python) and connectedComponents (Java) functions. Existing code that passes this argument must remove it.
  • Removed the TriangleCount class and its methods. Use the new triangle_count and triangle_count_pre_canonicalized functions (Python) and the triangleCount and triangleCountPreCanonicalized methods (Java) instead. These replacements accept the same arguments in the same order, except that the Python triangle_count_pre_canonicalized function no longer accepts the canonical_edges_schema argument. The canonical edges table must be in the schema that the input_schema argument specifies. For details, see OCGraph Python Library and OCGraph Java Library.
  • Renamed the PageRank damping factor argument from reset_prob to damping_factor (Python) and from resetProb to dampingFactor (Java) in static, dynamic, and temporal PageRank functions. Ocient temporarily accepts the deprecated Python keyword argument reset_prob as an alias for damping_factor. For details, see OCGraph Python Library and OCGraph Java Library.
  • Increased the maximum MATRIX size from 16,384 elements to 67,108,864 elements. The maximum memory allocation remains 512MiB. For details, see Data Types for SQL Functions.

1.0.0

  • Initial release of the OCGraph libraries. The Python ocient_graph module and the Java OCGraph JDBC class provide graph analytics operations (similar to Apache® Spark™ GraphX) for the Ocient System. For details, see OCGraph Python Library and OCGraph Java Library.
Last modified on October 29, 2026