From: ptkwt@...1.aracnet.com (Phil Tomson) Date: 2002-02-06T03:00:42+09:00 Subject: Re: looking for an example problem to demonstrate TaskMaster In article , wrote: >-----BEGIN PGP SIGNED MESSAGE----- > >In article , >Phil Tomson wrote: >>In article , >>Dave Thomas wrote: >>>ptkwt@shell1.aracnet.com (Phil Tomson) writes: >>> >>>> I'm looking for suggestions here... >>> >>>The good ol' Mandelbrot calculation? >>> >>Actually, now that I think about it doing the Mandelbrot on several >>machines and then sending back all the data points to the central 'server' >>would require too much information transfer vs compute time. >> > >- - You aren't looking deep enough then, the closer you get to the > actual set, the more iterations you need to make. I wasted a > bunch of Cray time doing exactly this kind of distributed > caculation years ago, at the limits of floating pt resolution > you need about 32k iterations per pixel. > >- - You can see the results in this mpeg, it's a funnier with > the sound track though... > >http://www.ifs.hr/ifs/fun/fractals/ > > Look for Mand. > >- - Actually, another useful example would be a webcrawler, but I'm >not sure we want to inflict yet another webcrawler on the world. > >- - There are lot's of interesting examples from the physics > world. They are under the general rubric of "Monte Carlo" > simulations. You could compute the mass of the proton. I've actually settled on estimating the value of pi using the Monte Carlo method - more machines either gets you more accuracy in the same amount of time or the same number of iterations in less time. Do you have any links or references to using Monte Carlo simulations to compute the mass of the proton? > >- - Lastly, you could adopt some simple examples from genetic > algorithms. I've thought of this. Got any pointers to some simple examples? Phil