让下一篇重要论文，不再错过。

原文摘要

Under the self-referential $K$-SAT exhaustive-search hypothesis (SRK-ESH), no uniform deterministic algorithm computes exact longest common subsequence or standard edit distance on binary strings in $O(n^{2-η})$ time for any fixed $η>0$, where $n$ is the longer sequence length. The same exclusion holds for one-dimensional dynamic time warping on $\{0,1,2,4,8\}$ with absolute-difference cost and for five-symbol dynamic time warping with the discrete metric. We derive Orthogonal Vectors hardness in polylogarithmic dimension by splitting the variables of the logarithmic-width source formula into two halves. Existing sequence reductions have polynomial overhead in the dimension. Their output lengths and construction times allow a fixed saving in the target exponent to yield a fixed proportional saving in the source exponent.