Expand description
Causal graph DAG with typed/weighted directed edges.
Provides a concurrent, lock-free causal reasoning graph where nodes
represent events or observations and edges encode causal relationships
with weights and provenance metadata. Built on DashMap for safe
concurrent access from multiple agent threads.
Structs§
- Causal
Chain - A traced causal chain from source to target through the graph.
- Causal
Edge - A weighted, typed directed edge between two causal graph nodes.
- Causal
Graph - A concurrent directed acyclic graph for causal reasoning.
- Causal
Node - A node in the causal graph representing an event, observation, or concept.
- Change
Event - A temporal change event: a set of nodes that changed together.
- Change
Prediction - A prediction that a node will change soon.
- Coupling
Pair - Coupling between two nodes based on co-modification frequency.
- Shadowing
Config - Configuration for geometric shadowing (memory decay).
- Spectral
Result - Result of spectral analysis on the causal graph.
- VqCodebook
- Vector quantization codebook for bootstrapping new entity embeddings.
Enums§
- Causal
Edge Type - The kind of causal relationship an edge represents.
- Spectral
Method - Algorithm used to estimate algebraic connectivity (lambda₂).
Constants§
- SPECTRAL_
EML_ MIN_ NODES - Prefer pure EML (O(1)) once the graph has this many nodes.
- SPECTRAL_
RFF_ MIN_ NODES - Prefer RFF over Lanczos once the graph has this many nodes.
Functions§
- compute_
shadows - Compute shadow-adjusted relevance for each node in the graph.
- max_
volatility_ batched - Maximum edge-type volatility across an iterator of edges, using a
clawft_treecalc::triagedispatch on the edge-kind stream so a uniform-kind incident set collapses to a singlematchbranch (Sequence form) instead of one per edge. - select_
spectral_ method - Select the spectral path for a graph of
node_countnodes.
Type Aliases§
- NodeId
- Numeric identifier for causal graph nodes.