Abstract

We present an extension of an adaptive, partially matrix-free, hierarchically semiseparable (HSS) matrix construction algorithm by Gorman et al. (2019) which uses Gaussian sketching operators to a broader class of Johnson–Lindenstrauss (JL) sketching operators. We develop theoretical work which justifies this extension. In particular, we extend the earlier concentration bounds to all JL sketching operators and examine this bound for specific classes of such operators including the original Gaussian sketching operators, subsampled randomized Hadamard transform (SRHT) and the sparse Johnson–Lindenstrauss transform (SJLT). We discuss the implementation details of applying SJLT and SRHT efficiently. Then we demonstrate experimentally that using SJLT or SRHT instead of Gaussian sketching operators leads to up to 2.5 times speedups of the serial HSS construction implementation in the STRUMPACK C++ library. Additionally, we discuss the implementation of a parallel distributed HSS construction that leverages Gaussian or SJLT sketching operators. We observe a performance improvement of up to 35 times when using SJLT sketching operators over Gaussian sketching operators. The generalized algorithm allows users to select their own JL sketching operators with theoretical lower bounds on the size of the operators which may lead to faster runtime with similar HSS construction accuracy.

Keywords

Mathematical Subject Classification

Milestones

Authors