feat(postgres): Add 53 SQL function definitions for all advanced modules #46
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Comprehensive implementation of advanced AI capabilities: ## New Modules (23,541 lines of code) ### 1. Self-Learning / ReasoningBank (`src/learning/`) - Trajectory tracking for query optimization - Pattern extraction using K-means clustering - ReasoningBank for pattern storage and matching - Adaptive search parameter optimization ### 2. Attention Mechanisms (`src/attention/`) - Scaled dot-product attention (core) - Multi-head attention with parallel heads - Flash Attention v2 (memory-efficient) - 10 attention types with PostgresEnum support ### 3. GNN Layers (`src/gnn/`) - Message passing framework - GCN (Graph Convolutional Network) - GraphSAGE with mean/max aggregation - Configurable aggregation methods ### 4. Hyperbolic Embeddings (`src/hyperbolic/`) - Poincaré ball model - Lorentz hyperboloid model - Hyperbolic distance metrics - Möbius operations ### 5. Sparse Vectors (`src/sparse/`) - COO format sparse vector type - Efficient sparse-sparse distance functions - BM25/SPLADE compatible - Top-k pruning operations ### 6. Graph Operations & Cypher (`src/graph/`) - Property graph storage (nodes/edges) - BFS, DFS, Dijkstra traversal - Cypher query parser (AST-based) - Query executor with pattern matching ### 7. Tiny Dancer Routing (`src/routing/`) - FastGRNN neural network - Agent registry with capabilities - Multi-objective routing optimization - Cost/latency/quality balancing ## Docker Infrastructure - Dockerfile with pgrx 0.12.6 and PostgreSQL 16 - docker-compose.yml with test runner - Initialization SQL with test tables - Shell scripts for dev/test/benchmark ## Feature Flags - `learning`, `attention`, `gnn`, `hyperbolic` - `sparse`, `graph`, `routing` - `ai-complete` and `graph-complete` bundles 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
- Fix simsimd Option/Result type mismatch in scaled_dot.rs - Fix f32/f64 type conversions in poincare.rs and lorentz.rs - Fix AVX512 missing wrapper functions by using AVX2 fallback - Fix Vec<Vec<f32>> to JsonB for pgrx pg_extern compatibility - Fix DashMap get() to get_mut() for mutable access - Fix router.rs dereference for best_score comparison - Update Dockerfile to copy pre-written SQL file for pgrx - Simplify init.sql to use correct function names - Add postgres-cli npm package for CLI tooling All changes tested successfully in Docker with: - Extension loads with AVX2 SIMD support (8 floats/op) - Distance functions verified working - PostgreSQL 16 container runs successfully 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Added from claude/ruvector-lfm2-llm-01YS5Tc7i64PyYCLecT9L1dN branch: - examples/ruvLLM: Complete LLM inference system with SIMD optimization - Pretraining, benchmarking, and optimization system - Real SIMD-optimized CPU inference engine - Comprehensive SOTA benchmark suite - Attention mechanisms, memory management, router Enhanced postgres-cli with full ruvector-postgres integration: - Sparse vector operations (BM25, top-k, prune, conversions) - Hyperbolic geometry (Poincare, Lorentz, Mobius operations) - Agent routing (Tiny Dancer system) - Vector quantization (binary, scalar, product) - Enhanced graph and learning commands 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
- Change createVectorTable to use ruvector type (native RuVector extension) - Add dimensions column for metadata since ruvector is variable-length - Update index creation to use simple btree (HNSW/IVFFlat TBD) - Tested against Docker container with ruvector extension 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Enable all advanced PostgreSQL extension functions by adding their SQL definitions to the extension file. This exposes all Rust #[pg_extern] functions to PostgreSQL. ## New SQL Functions (53 total) ### Hyperbolic Geometry (8 functions) - ruvector_poincare_distance, ruvector_lorentz_distance - ruvector_mobius_add, ruvector_exp_map, ruvector_log_map - ruvector_poincare_to_lorentz, ruvector_lorentz_to_poincare - ruvector_minkowski_dot ### Sparse Vectors (14 functions) - ruvector_sparse_create, ruvector_sparse_from_dense - ruvector_sparse_dot, ruvector_sparse_cosine, ruvector_sparse_l2_distance - ruvector_sparse_add, ruvector_sparse_scale, ruvector_sparse_to_dense - ruvector_sparse_nnz, ruvector_sparse_dim - ruvector_bm25_score, ruvector_tf_idf, ruvector_sparse_normalize - ruvector_sparse_topk ### GNN - Graph Neural Networks (5 functions) - ruvector_gnn_gcn_layer, ruvector_gnn_graphsage_layer - ruvector_gnn_gat_layer, ruvector_gnn_message_pass - ruvector_gnn_aggregate ### Routing/Agents - "Tiny Dancer" (11 functions) - ruvector_route_query, ruvector_route_with_context - ruvector_calculate_agent_affinity, ruvector_select_best_agent - ruvector_multi_agent_route, ruvector_create_agent_embedding - ruvector_get_routing_stats, ruvector_register_agent - ruvector_update_agent_performance, ruvector_adaptive_route - ruvector_fastgrnn_forward ### Learning/ReasoningBank (7 functions) - ruvector_record_trajectory, ruvector_get_verdict - ruvector_distill_memory, ruvector_adaptive_search - ruvector_learning_feedback, ruvector_get_learning_patterns - ruvector_optimize_search_params ### Graph/Cypher (8 functions) - ruvector_graph_create_node, ruvector_graph_create_edge - ruvector_graph_get_neighbors, ruvector_graph_shortest_path - ruvector_graph_pagerank, ruvector_cypher_query - ruvector_graph_traverse, ruvector_graph_similarity_search ## CLI Updates - Enabled hyperbolic geometry commands in postgres-cli - Added vector distance and normalize commands - Enhanced client with connection pooling and retry logic 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
#[pg_extern]functions to PostgreSQLNew PostgreSQL Functions
Hyperbolic Geometry (8 functions)
Distance metrics and operations for Poincaré ball and Lorentz hyperboloid models:
ruvector_poincare_distance,ruvector_lorentz_distanceruvector_mobius_add,ruvector_exp_map,ruvector_log_mapruvector_poincare_to_lorentz,ruvector_lorentz_to_poincareruvector_minkowski_dotSparse Vectors (14 functions)
Text-based sparse representations with BM25 scoring:
sparse_create,sparse_from_dense,sparse_to_densesparse_dot,sparse_cosine,sparse_l2_distancesparse_add,sparse_scale,sparse_normalize,sparse_topkbm25_score,tf_idfGNN - Graph Neural Networks (5 functions)
gnn_gcn_layer,gnn_graphsage_layer,gnn_gat_layergnn_message_pass,gnn_aggregateRouting/Agents - "Tiny Dancer" (11 functions)
Agent routing with FastGRNN acceleration:
route_query,route_with_context,adaptive_routecalculate_agent_affinity,select_best_agent,multi_agent_routeregister_agent,update_agent_performance,get_routing_statsLearning/ReasoningBank (7 functions)
Adaptive search parameter optimization:
record_trajectory,get_verdict,distill_memoryadaptive_search,learning_feedback,optimize_search_paramsGraph/Cypher (8 functions)
Graph storage with Cypher query support:
graph_create_node,graph_create_edge,graph_get_neighborsgraph_shortest_path,graph_pagerank,cypher_querygraph_traverse,graph_similarity_searchTest plan
CREATE EXTENSION ruvector🤖 Generated with Claude Code