Module streaming_depth 

Streaming / online depth computation for functional data.

This module decouples reference-set construction from query evaluation, enabling efficient depth computation in streaming scenarios:

• SortedReferenceState pre-sorts reference values per time point for O(log N) rank queries.
• StreamingMbd uses a rank-based combinatorial identity to compute Modified Band Depth in O(T log N) per query instead of O(N² T).
• StreamingFraimanMuniz computes Fraiman-Muniz depth via binary search on sorted columns.
• StreamingBd computes Band Depth with decoupled reference and early-exit optimisation.
• RollingReference maintains a sliding window of reference curves with incremental sorted-column updates.

Structs

FullReferenceState

Full reference state that keeps per-curve values alongside sorted columns.

RollingReference

Sliding window of reference curves with incrementally maintained sorted columns.

SortedReferenceState

Pre-sorted reference values at each time point for O(log N) rank queries.

StreamingBd

Streaming Band Depth estimator.

StreamingFraimanMuniz

Streaming Fraiman-Muniz depth estimator.

StreamingMbd

Rank-based Modified Band Depth estimator.

Traits

StreamingDepth

Trait for streaming depth estimators backed by a pre-built reference state.