[plt-scheme] Statistics (V301.5 Speed Up)

On Feb 11, 2006, at 10:11 AM, Williams, M. Douglas wrote:

> Actually, this exercise is very much a "plug and chug" endeavor on  
> my part.
> I am just curious - in a wholly unscientific way - about the speed  
> up from
> V301 to V301.5 for a continuation entensive application.  In this  
> case, the
> application domain is models developed using the simulation  
> collection I
> developed.  It was meant as a thanks for the work that went into the
> run-time improvements and as I way that I could guage for myself some
> measure of the speed up.  I am quite satisfied with simple  
> statistics on the
> results - the only variable between runs of the same model has been  
> the
> version of the PLT Scheme system used.  Some of us were also  
> curious about
> the shape of the distributions of run times - another example of  
> which is
> attached here.
>
> If anyone would like to do some statistically meaningful ananysis  
> on the
> data.  All of the code is available on PLaneT.
>
> Doug

It's certainly taking awhile under 301 (344 dots so far).

What intrigues me is the interplay between long tails on the  
individual distributions and the convergence rate.

===
Gregory Woodhouse
gregory.woodhouse at sbcglobal.net

"Education is a progressive discovery
of our own ignorance."
--Will Durant