Polynomial-time trace reconstruction in the smoothed complexity model

SODA(2021)

引用 24|浏览70
暂无评分
摘要
In the \emph{trace reconstruction problem}, an unknown source string $x \in \{0,1\}^n$ is sent through a probabilistic \emph{deletion channel} which independently deletes each bit with probability $\delta$ and concatenates the surviving bits, yielding a \emph{trace} of $x$. The problem is to reconstruct $x$ given independent traces. This problem has received much attention in recent years both in the worst-case setting where $x$ may be an arbitrary string in $\{0,1\}^n$ \cite{DOS17,NazarovPeres17,HHP18,HL18,Chase19} and in the average-case setting where $x$ is drawn uniformly at random from $\{0,1\}^n$ \cite{PeresZhai17,HPP18,HL18,Chase19}. This paper studies trace reconstruction in the \emph{smoothed analysis} setting, in which a ``worst-case'' string $x^{\worst}$ is chosen arbitrarily from $\{0,1\}^n$, and then a perturbed version $\bx$ of $x^{\worst}$ is formed by independently replacing each coordinate by a uniform random bit with probability $\sigma$. The problem is to reconstruct $\bx$ given independent traces from it. Our main result is an algorithm which, for any constant perturbation rate $0<\sigma < 1$ and any constant deletion rate $0 < \delta < 1$, uses $\poly(n)$ running time and traces and succeeds with high probability in reconstructing the string $\bx$. This stands in contrast with the worst-case version of the problem, for which $\text{exp}(O(n^{1/3}))$ is the best known time and sample complexity \cite{DOS17,NazarovPeres17}. Our approach is based on reconstructing $\bx$ from the multiset of its short subwords and is quite different from previous algorithms for either the worst-case or average-case versions of the problem. The heart of our work is a new $\poly(n)$-time procedure for reconstructing the multiset of all $O(\log n)$-length subwords of any source string $x\in \{0,1\}^n$ given access to traces of $x$.
更多
查看译文
关键词
smoothed complexity model,trace,polynomial-time
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要