%0 Journal Article
%J Optimization Methods & Software
%D 2012
%T Randomized Heuristics for Exploiting Jacobian Scarcity
%A A. Lyons
%A I. Safro
%A Jean Utke
%X We describe a code transformation technique that, given code for a vector function F, produces code suitable for computing collections of Jacobian-vector products F\'(x)[x dot] or Jacobian transpose-vector products F\'(x)[sup T][y bar]. Exploitation of scarcity � a measure of the degrees of freedom in the Jacobian matrix � means solving a combinatorial optimization problem that is believed to be hard. Our heuristics transform the computational graph for F, producing, in the form of a transformed graph G\', a representation of the Jacobian F\'(x) that is both concise and suitable for the evaluation of large collections of Jacobian-vector products or Jacobian-transpose-vector products. Our heuristics are randomized in nature and compare favorably in all cases with the best known heuristics.
%B Optimization Methods & Software
%V 27
%8 04/2012
%G eng
%U http://www.tandfonline.com/doi/abs/10.1080/10556788.2011.577774?journalCode=goms20#preview
%N 2
%1 http://www.mcs.anl.gov/papers/P1716.pdf
%& 311-322