Struct PprResult 

pub struct PprResult {
    pub scores: FxHashMap<FragmentId, f64>,
    pub truncated: bool,
    pub forward_pushes: usize,
    pub backward_pushes: usize,
    pub seeded: bool,
}

Public PPR result. truncated is logical-OR of forward + backward push truncation flags. When true, downstream renormalization (sum-to-1 across nodes) hides the fact that the iteration was cut short by max_pushes_cap; the absolute relevance scores are biased toward seeds and not directly comparable across instances. We surface this to Python so calibration / final-eval rows can be filtered in post-analysis (PolyBench / Multi-SWE-bench instances with >20k fragments are the primary suspects).

Fields

scores: FxHashMap<FragmentId, f64>

truncated: bool

forward_pushes: usize

backward_pushes: usize

seeded: bool

False when no seed mass ever entered the push (no seed matched a graph node, or seed_weights summed to <= 0 across matched seeds). Lets the caller tell “nothing to seed from” apart from “seeded and converged to a genuinely empty scores map” – both produce the same empty map and forward_pushes == 0 otherwise.