exp(a*log(b)) is usually faster than b^a (in fact, b^a internally usually does something similar, but with some extra tricks to avoid loss of precision).
You might be able to use one of the vectorized math libraries: AppleAccelerate.jl (if you're on OS X), VML.jl (if you have access to Intel MKL: you can download the community licensed version for free), or Yeppp.jl. -Simon On Wednesday, 29 June 2016 23:08:35 UTC-4, Anonymous wrote: > > I have an algorithm where the bulk of the computation time is spent > calculating an operation of the form > > Matrix .^ transpose(Vector) > > I was able to get a 33% speed up by re-writing this as > > exp(transpose(Vector) .* log(Matrix)) > > however this is something of a hack and I don't really approve of it, is > there a more elegant way to speed up this operation? >