graph LR
PathGenerator["PathGenerator"]
IntervalProcessor["IntervalProcessor"]
LeafNodeEvaluator["LeafNodeEvaluator"]
PathState["PathState"]
BrownianIncrementGenerator["BrownianIncrementGenerator"]
LevyIncrementContainer["LevyIncrementContainer"]
PathGenerator -- "delegates to" --> IntervalProcessor
IntervalProcessor -- "relies on" --> BrownianIncrementGenerator
IntervalProcessor -- "updates" --> PathState
IntervalProcessor -- "relies on" --> LeafNodeEvaluator
LeafNodeEvaluator -- "uses" --> LevyIncrementContainer
LeafNodeEvaluator -- "updates" --> PathState
The diffrax._brownian.tree subsystem orchestrates the generation of Brownian paths through a hierarchical, tree-based evaluation. The PathGenerator serves as the primary entry point, delegating the core recursive processing to the IntervalProcessor. The IntervalProcessor is responsible for breaking down time intervals, coordinating the generation of Brownian increments using the BrownianIncrementGenerator, and updating the PathState as the path evolves. At the smallest time steps (leaf nodes), the IntervalProcessor relies on the LeafNodeEvaluator to compute the final Brownian values, which in turn utilizes the LevyIncrementContainer for fundamental random components and updates the PathState. This design ensures a clear separation of concerns, with PathGenerator managing the overall flow, IntervalProcessor handling recursive decomposition, and LeafNodeEvaluator performing terminal computations, all while maintaining the path's dynamic state via PathState.
Serves as the public entry point and orchestrator for the entire stochastic path generation process. It initiates the recursive/iterative evaluation of the Brownian path across the tree structure and may perform final adjustments like increment normalization.
Related Classes/Methods:
Handles the processing of intermediate time intervals within the tree-based path generation. It recursively divides larger intervals into smaller ones and coordinates the generation of corresponding Brownian increments.
Related Classes/Methods:
Computes the specific Brownian values at the terminal (leaf) nodes of the tree, representing the smallest time steps in the path. It utilizes pre-generated or computed Levy increments for this purpose.
Related Classes/Methods:
A data structure (likely a PyTree, given JAX integration) responsible for maintaining the dynamic state during the path generation, such as the current time, path values, and other contextual information.
Related Classes/Methods:
Generates the raw Brownian increments based on a specific archival or generation algorithm. These are the fundamental random components that build the stochastic path.
Related Classes/Methods:
A data structure that encapsulates the Levy increments, which are the fundamental random components used at the leaf nodes of the Brownian path tree.
Related Classes/Methods: