We study the classic problem of minimizing the expected total completion time of jobs on $m$ identical machines in the setting where the sizes of the jobs are stochastic. Specifically, the size of each job is a random variable whose distribution is known to the algorithm, but whose realization is revealed only after the job is scheduled. While minimizing the total completion time is easy in the deterministic setting, the stochastic problem has long been notorious: all known algorithms have approximation ratios that either depend on the variances, or depend linearly on the number of machines. We give an $\widetilde{O}(\sqrt{m})$-approximation for stochastic jobs which have Bernoulli processing times. This is the first approximation for this problem that is both independent of the variance in the job sizes, and is sublinear in the number of machines $m$. Our algorithm is based on a novel reduction from minimizing the total completion time to a natural makespan-like objective, which we call the weighted free time. We hope this free time objective will be useful in further improvements to this problem, as well as other stochastic scheduling problems.
翻译:我们研究的是,在工作规模不同的情况下,用相同机器将预期工作总完成时间降低到最低的典型问题。 具体地说, 每份工作的规模是一个随机变量,其分布为算法所知道,但其实现只是在工作排期之后才显现出来。 在确定性环境下,将全部完成时间降低到最低是容易的, 随机问题长期以来一直臭名昭著: 所有已知的算法都有取决于差异或线性取决于机器数量的近似比率。 我们给出了 $\ 全局性{O}(\\\ qrt{m})$- 套用于有Bernoulli 处理时间的随机工作。 这是这个问题的第一个近似点, 与工作规模的差异无关, 并且是机器数量的次线性。 我们的算法基于一种新型的缩减, 从将全部完成时间最小化到自然造色目标, 也就是我们称之为加权自由时间。 我们希望这个自由时间目标将有益于这一问题的进一步改善, 以及其它的排序问题。