StoriesAt Lexmark, I wrote code for Moore-Penrose inverse more than once, having misplaced a first version.Significantly smaller than others, my algorithm was not embedded in products, being instead used to generate tables that were embedded. Consequently, filing for a patent was unnecessary, but it now occurs to me that publishing it might have yielded some professional visibility. Even that second derivation took a week or two, and recreating it now is probably beyond me. One night recently, I woke up thinking about inflection points while working at IBM and Lexmark. Before forgetting more of them, here are related anecdotes.
|