|Do you keep a copy of all the results?|
If so, would or could you give them out if a person asked for them?
I am keeping track of all the results I can just to try a little expermint
If you were able to give them out, what amount of file size we talking about? 3 or 4 CD's worth? Or is there a link to where it is kept.
|I think there are about 30GB of results, so more than a few CDs. I have no reason not to distribute them apart from that it's difficult to move that much data around. And did you want just one optimisation or several?|
|I'd be interested in supplementing my results.dat file for DecayRotA. Currently I am running ~43,000 of my own results with ~1,300 sample results.|
Would there be a way to grab a top 5,000/10,000 for DecayRotA?
|Maybe, though I'm not regularly uploading larger than 100 result samplefiles because the size of them just hammers the bandwidth if everyone downloads them automatically. I could do a one-off 10000 sample for DecayRotA just to see what it looks like though.|
That has about 1000 results in it.
|Thank you Stephen!|
I tried the link and it bounces to a 404.
|I had to regenerate it. Now it's there.|
|Sorry... didn't see the number of results you posted on the link. I was able to download the file.|
Any chance of grabing a larger number of top results?
|I personally was looking at the DecayRot and DecayRotA|
I was trying to figure out why they are not as good as the rest of them are.. points wise and %mouns......
I was going to dedicate my whole farm 14 systems 32+ Ghz's to see if my little farm could help the %mouns in those 2 tests......
Is there any reason why DecayRot and DecayRotA always seem to be behind? Or is it just the way the program selects?
Would it be wise to run just one on all my systems and then share the results.dat file to get the best performance?
Just trying to figure out what is best for the project......
|They've been running for less time (still increasing in some cases) but there's also a physics reason: I was testing to see if I could make the channel shorter (and therefore cheaper to build) by combining two parts, but I think that has meant the phase rotation has become less powerful as the particles have spread out by less when they get there.|
The optimisation also seems to be hampered by the statistical noise from the number of particles. Muon1 seems to wade through this "fog" quite well, but it would go faster if I could work out a way of reducing this. More particles is a very CPU-consuming way as you need 4x as many to halve the "noise". A fixed random seed might work but might also 'bias' the optimisation for that particular way the particles decay.
Aha... that gives me an idea actually. I could make the seeding behaviour a lattice option so I could run some experimental lattices to compare the performance.
|...and if wmikrut is wondering why he isn't getting any e-mails, it's because his mailbox is full and bouncing all incoming mail with a fatal error.|
|... which is wrong! I've fixed that problem... thanks Stephen. So much for having multiple MX records.|