User blog comment:Vel!/Ordinal BEAF/@comment-25418284-20130315024619/@comment-5529393-20130315132150

That's an interesting question. I originally thought that a -> phi(a, 0) was a pretty good estimation of the array of operator, but I'm starting to think it may be stronger than that.