Many applications require randomly sampling binary graphs with fixed degrees, or randomly sampling binary matrices with fixed row and column marginals. Although several algorithms to perform this sampling have been proposed, the ``curveball'' algorithm is the fastest that has been proven to sample uniformly at random. In this paper, we introduce a more efficient version of curveball called ``fastball'' that is roughly 8 times faster.
翻译:许多应用程序都要求随机抽样使用固定度的二进制图表,或者随机抽样使用固定行边和柱边边边边的二进制矩阵。 虽然已经提出了进行这一抽样的几种算法,但“曲线球”算法是经证明可以统一随机抽样的最快速度。 在本文件中,我们引入了一个更高效的曲线球版本,叫做“快球 ”, 速度约为8倍。