We investigate iterative methods with randomized preconditioners for solving overdetermined least-squares problems, where the preconditioners are based on a random embedding of the data matrix. We consider two distinct approaches: the sketch is either computed once (fixed preconditioner), or, the random projection is refreshed at each iteration, i.e., sampled independently of previous ones (varying preconditioners). Although fixed sketching-based preconditioners have received considerable attention in the recent literature, little is known about the performance of refreshed sketches. For a fixed sketch, we characterize the optimal iterative method, that is, the preconditioned conjugate gradient as well as its rate of convergence in terms of the subspace embedding properties of the random embedding. For refreshed sketches, we provide a closed-form formula for the expected error of the iterative Hessian sketch (IHS), a.k.a. preconditioned steepest descent. In contrast to the guarantees and analysis for fixed preconditioners based on subspace embedding properties, our formula is exact and it involves the expected inverse moments of the random projection. Our main technical contribution is to show that this convergence rate is, surprisingly, unimprovable with heavy-ball momentum. Additionally, we construct the locally optimal first-order method whose convergence rate is bounded by that of the IHS. Based on these theoretical and numerical investigations, we do not observe that the additional randomness of refreshed sketches provides a clear advantage over a fixed preconditioner. Therefore, we prescribe PCG as the method of choice along with an optimized sketch size according to our analysis. Our prescribed sketch size yields state-of-the-art computational complexity for solving highly overdetermined linear systems. Lastly, we illustrate the numerical benefits of our algorithms.
翻译:我们用随机设置的先决条件来调查迭代方法,用随机设置的先决条件来解决最不固定的平方问题,在这种方法中,先决条件是以随机嵌入数据矩阵为基础。我们考虑两种截然不同的方法:草图或者计算一次(固定先决条件),或者随机投影在每次迭代中刷新,即独立于先前的迭代(变式先决条件)抽样。尽管固定的草图前导者在最近的文献中受到相当重视,但对于更新的草图的性能却知之甚少。对于固定的草图,我们描述最佳的迭代方法,即:预设的同位梯度,以及它以随机嵌入的子空间属性为条件的趋同率。对于更新的草图,我们提供了一种封闭式公式公式,对于迭代图的迭代图(IHS)的预期误差(IHS,a.k.a.a.a.a.a.a.a.a.,对于基于高空嵌入特性的固定前置精度,我们的公式与预想的随机预测过程。我们的主要技术贡献显示的是,我们最精确的趋趋和最接近的汇率,我们最接近的渐变的数值是,我们最接近的数值是我们最接近的我。