The cube-root fluctuation conjecture for the LCS concatenation gain

About 7 years old · traced to

Let VV and WW be independent random words of length 2n2n, written as concatenations V=V1V2V=V_1V_2 and W=W1W2W=W_1W_2, where each component has length nn. Define

Δ(V,W)=LCS⁡(V,W)−LCS⁡(V1,W1)−LCS⁡(V2,W2),\Delta(V,W)=\operatorname{LCS}(V,W)-\operatorname{LCS}(V_1,W_1)-\operatorname{LCS}(V_2,W_2),

and write Δ(2n)=Δ(V,W)\Delta(2n)=\Delta(V,W). Cube-root fluctuation conjecture. There are constants c1,c2c_1,c_2 such that

E⁡Δ(n)∼c1n1/3\operatorname{\mathbb{E}}\Delta(n)\sim c_1n^{1/3}

and

Var⁡Δ(n)∼c2n1/3.\sqrt{\operatorname{Var}\Delta(n)}\sim c_2n^{1/3}.

The conjecture is suggested by the paper’s computational experiments on the excess LCS obtained by allowing matches to cross the two concatenated blocks. The cited prior estimate gives only the upper-order control E⁡Δ(V,W)=O(nlog⁡n)\operatorname{\mathbb{E}}\Delta(V,W)=O(\sqrt{n\log n}), so the cube-root asymptotics remain open.

References

Primary source

Boris Bukh and Christopher Cox, “Periodic words, common subsequences and frogs”, arXiv:1912.03510 (2021).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.