Function viterbi_decode 

pub fn viterbi_decode(
    position_candidates: &[Vec<CandidateNetElement>],
    position_probabilities: &[Vec<f64>],
    netelements: &[Netelement],
    netelement_index: &HashMap<String, usize>,
    graph: &DiGraph<NetelementSide, f64>,
    node_map: &HashMap<NetelementSide, NodeIndex>,
    cache: &mut ShortestPathCache,
    config: &PathConfig,
) -> Result<ViterbiResult, ProjectionError>

Decode the most probable netelement sequence using the Viterbi algorithm.

Arguments

• position_candidates — Per-position candidate netelements (from Phase 1).
• position_probabilities — Per-position emission probabilities: position_probabilities[t][candidate_idx] = emission_prob. candidate_idx is an index into position_candidates[t].
• netelements — Full netelement set (for geometry lookups).
• netelement_index — Map from netelement ID → index in netelements.
• graph — Directed topology graph with distance-weighted edges.
• node_map — Map from NetelementSide → NodeIndex in graph.
• cache — Mutable shortest-path cache (lazily populated).
• config — Path configuration (provides beta, edge_zone_distance).

Returns

A ViterbiResult with one or more sub-sequences covering all time-steps that have at least one candidate.