Answers are grounded in this paper and cite their sources.
Exact sizes of Levenshtein spheres are hard to compute because different edit scripts can produce the same string. We study run strings u=x1n1…xmnm in the stable regime, where every run length is at least the radius r. For r=2 we prove a closed formula for |SL(u,2)| for every m≥2: the layers of output length n−1, n and n+1 are identified with explicit sets of strings described through longest common prefixes and suffixes, and the layers of length n±2 follow from deletion and insertion counts. For r=3 we give a computer-assisted proof of a closed formula for 2≤m≤7. The proof rests on a certification method: a banded edit-distance automaton shows that the sphere size is a polynomial of bounded degree in the run lengths and in Q=q−1, where q is the alphabet size, so a candidate formula is proved by exact evaluation on an explicit finite set of profiles. In all proved cases the coefficients of the polynomial in Q are polynomials in the invariants S1, P2, P3, B1, E1, E2, and n1nm of the run profile. We state conjectured formulas, supported by exact enumeration, for r=3 with m≥8 and for r=4 with five runs, formulate a general template conjecture, and give an algorithm, based on exact enumeration, interpolation, and invariant fitting that discovers such formulas.
No takes yet. Share an insight, caveat, or question.
Said Algarni (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: