arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1412.5153cs.CG
收藏

随机点的凸包的面积和周长

Area and Perimeter of the Convex Hull of Stochastic Points

Pablo Pérez-Lantero

摘要:

给定平面上$n$点的集合$P$,研究了$P$的随机子集$S$的凸包的面积和周长的概率分布函数的计算.随机子集$S$通过以给定的有理概率$π_p$独立地绘制$P$的每个点$p$来形成。对于这两个措施的凸包,我们表明,它是P-难计算的概率,该措施至少是一个给定的绑定$w$。对于$\varepsilon\in(0,1)$,我们提供了一种算法,该算法在$O(n^{6}/\varepsilon)$时间内运行,并返回一个介于区域至少为$w$的概率和区域至少为$(1-\varepsilon)w$的概率之间的值。对于周长,我们显示了一个类似的算法运行在$O(n^{6}/\varepsilon)$时间。最后,给定$\varepsilon,δ\in(0,1)$和任何措施,我们展示了一个$O(n\log n+ (n/\varepsilon^2)\log(1/δ))$时间蒙特卡罗算法,返回一个值,成功的概率至少$1-δ$,不同的概率,最多$\varepsilon$的措施至少是$w$。

英文摘要:

Given a set $P$ of $n$ points in the plane, we study the computation of the probability distribution function of both the area and perimeter of the convex hull of a random subset $S$ of $P$. The random subset $S$ is formed by drawing each point $p$ of $P$ independently with a given rational probability $π_p$. For both measures of the convex hull, we show that it is \#P-hard to compute the probability that the measure is at least a given bound $w$. For $\varepsilon\in(0,1)$, we provide an algorithm that runs in $O(n^{6}/\varepsilon)$ time and returns a value that is between the probability that the area is at least $w$, and the probability that the area is at least $(1-\varepsilon)w$. For the perimeter, we show a similar algorithm running in $O(n^{6}/\varepsilon)$ time. Finally, given $\varepsilon,δ\in(0,1)$ and for any measure, we show an $O(n\log n+ (n/\varepsilon^2)\log(1/δ))$-time Monte Carlo algorithm that returns a value that, with probability of success at least $1-δ$, differs at most $\varepsilon$ from the probability that the measure is at least $w$.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑