Close the implementation, replay, semantic, generalization, and causal-evidence gaps

Establish Java–Lean refinement, complete-trace replay, semantic completeness, industrial generalization, and causal attribution for the typed flexible-arity slotted e-graph construction and its Alloy evaluation pipeline.

Background

The paper distinguishes its abstract Lean-checked certificate and finite-unfolding results from the Java implementation and the bounded Alloy experiments. The implementation was not shown to refine the formal construction, the corpus was not independently replayed through complete exported traces, and the bounded evaluation does not establish semantic completeness or broader industrial validity. The authors also report that the measurements do not isolate causal sources of the observed performance and behavior.

These limitations are explicitly presented as unresolved rather than merely as scope restrictions: Java–Lean refinement, full-trace replay, semantic completeness, industrial generalization, and causal attribution are all identified as open. Resolving them would connect the formal specification to the concrete producer and strengthen the empirical claims beyond bounded structural-consolidation evidence.