On Mon, Jun 2, 2008 at 12:02 PM, Harald Schilly <[EMAIL PROTECTED]> wrote: > > > > On Jun 2, 7:46 pm, "William Stein" <[EMAIL PROTECTED]> wrote: >> Hi, >> >> I've posted a first very very (very) minimal "quantitative finance" >> piece of code here: > > I have briefly looked at the code and I'm curious: How significant is > the speed difference between R, using an ts() object and fSeries, and > Python?
One sample benchmark I tried was computing the standard deviation. The new code I just wrote was 500 times faster than rpy (and that's *without* considering any communications overhead), and 15 times faster than numpy. Every single function in the time series class that has the same functionality as R or numpy/scipy is significantly faster than those systems -- sometimes not nearly 15 times faster, but often 5 times. I was *really* surprised by how a few hours with cython could do much better than R (and numpy in some cases) for very basic things. If the speed factor were "only" 5 over R, then I still think it would be fully worth redoing everything. That the speed difference is much larger is truly shocking. > And how does it relate to the finance packages in R? Do you > want to implement something similar to rmetrics.org or is this > different/more general? Different and probably less general. The focus is on being blazingly fast and doing a few things very well. -- William --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
