Dice Question Streamline Icon: https://streamlinehq.com

Close the approximation gap for directed weighted replacement paths

Determine the exact randomized round complexity of computing (1+ε)-approximate replacement path distances for a given s–t path in directed weighted graphs in the CONGEST model by closing the current gap between the O~(n^{2/3}+D) upper bound and the Ω~(√n+D) lower bound.

Information Square Streamline Icon: https://streamlinehq.com

Background

The paper proves a tight Θ~(n{2/3}+D) bound for exact replacement paths in unweighted directed graphs and gives an O~(n{2/3}+D) algorithm for (1+ε)-approximate replacement paths in directed weighted graphs.

However, the lower-bound technique used for exact computations does not extend to approximations, leaving a gap between the current upper bound and the best-known Ω~(√n+D) lower bound from prior work.

References

Our lower bound proof inherently does not apply to approximation algorithms, so closing the gap between our upper bound of \widetilde{O}(n{2/3} + D) and the lower bound of \widetilde{\Omega}(\sqrt{n} + D) by \citet{manoharan2024computing} remains an intriguing open question.

Optimal Distributed Replacement Paths (2502.15378 - Chang et al., 21 Feb 2025) in Abstract