Renting a Supercomputer

[John Q Public]

Quote from Te':

Not all Tampa

If that was directed to me, I don't understand your answer. Are you just trying to be disagreeable again?

 

[canyonman00]

Quote from Katrina Johns:



http://www.sciam.com/article.cfm?articleID=000E238B-33EC-1C6F-84A9809EC588EF21

Ok, you made my point. That was posted by Ninja. I saw that one, but not one from the other poster (Metooxx) as Alphie's reply had credited.

 

[nitro]

Quote from aphexcoil:


Perhaps I do need a better algorithm. I don't have the time to refine an algorithm, though -- so if it isn't precise than hopefully the brute force will just cover any small inefficiencies.

Yes you do...It's called genetic algorithms. As long as the optimizer has some way to "unstukk" itself from a valley, this should work for most purposes.

At the end, if the parameters in the "neighborhood" are stable, then you can zone in on an area of the topological space with brute force.

This is probably one of the most important areas of CS currently.

I will be posting a picture of my 4 node cluster later...

nitro

 

[DTK]

I couldn't agree more.

Quote from plumlazy:

Aphie,

As you once told Gordon, " JUST TRADE"

 

[TSaimoto]

Aphie, are you in search of the holy grail?

 

[trader99]

Quote from vhehn:

dont you think the big mutual funds with unlimited money have already done this?and yet most cant even beat the s&p.something to think about.

Actually, NO! Mutual funds who managed a few hundred billions(like Vanguaard, Fidelity, Janus,etc.) don't use that much computing power. They are NOT quants. They use traditional fundamental analysis which any simple PC with spreadsheet and some fundamental database can do.

Quant hedge funds and proprietary trading desk on the Street on the other hand are rather data and computing intensive. But since their horizons are much longer(position trading), they dont' nee dto analyze down to the last tick.



trader99

 

[axeman]

Well then.... stop programming in LOGO

peace

axeman

Quote from aphexcoil:

My first shot is to call up the University of Maryland and ask them if they have any open time-sharing slots on their system.

Unfortunately, even a couple linked Pentiums won't be able to work for me. I want to do various types of number crunching on a few billion ticks worth of data.

My computer at work is a 2.0 Gigahertz Pentium and it just grinds to a halt when I ask it to start calculating just a week's work of tick data -- not to mention 5 years worth.

 

[miniTrdr]

my .0002 cents
sounds like sneaker net distributed computing would be easier/faster - since you already have the programming done - divide your data up - probably by contract expirations - and run the program on all your PCs - log all the statistics to a central database

cluster would just add a lot of complexity- speed advantages probably would be nil if you dont know what your doing

keep it simple

besides does it take a supercomputer to figure out how fast your going to lose money just a joke

 

[saxon]

Here are 2 commercial sites that might be what you are looking for.

http://www.tsunamictechnologies.com/

http://www.pioneertechnology.com/index.php

good luck,
saxon

 

[nitro]

Quote from canyonman00:


...I built my own Beowulf cluster for a pet project (handicapping thoroughbred's) about a year ago with this OS...

:eek:

What were you testing here that required a cluster! Did it make money after the 17% take and the rounding down?

When I go to the track now (harness) I use Dr. Z's method. However, even if I wait until the last few seconds of a race to wager, the "hole" is filled by 3/4 of the people at the track at the same time

Fascinating...

nitro