Username: Password:
Search site:
Subscribe to thread via RSS
Stephen Brooks
2004-06-28 01:16:03
Nice result in PhaseRotB from [TA]Wonga: now 7% over the baseline, which is getting pretty significant now.
2004-06-28 01:36:23
Well done Wonga!

Stephen- Is there much more growth left in PhaseRotB, do you think?  It'd be good if we got to 10 or 15% over.  Also, are you going to throw the baseline data into the mix just to see what happens?
Stephen Brooks
2004-06-28 04:52:24
I think there might be a lot of growth left - like 25% - but we might only get it fairly slowly right now.  I'll be improving the optimiser a bit in August and that _might_ speed things up, but you can never tell until you test things.
2004-06-28 10:44:07
Almost 10% better now.  Cool
2004-06-28 16:12:46
Good work guys.
Stephen Brooks
2004-06-29 13:48:07
Now it's LostinTime [] with 12.0% above the baseline.  Looks like the current version of the optimiser is still doing its stuff. 
I think whatever Wonga found has set off this new cascade of better results.
2004-06-29 21:15:45
Actually, my results started improving when I change from 4.41a to 4.41b.

Note: Its not stopping at 12 percent already dumped my forth straight number one result.
2004-07-01 01:06:16
By some weird accident, my client decided to produce exactly the same parameters with gen=8 while the previous value was found with gen=4 and the result is a whopping 1.911300 (2362.8 Mpts) Big Grin
2004-07-03 07:52:21
This is the first time I've had a look in this discussion forum and I was shocked to find my name up there!

Did I do something good?  Smile I'm afraid I'm pretty new around here, so don't really understand the fine details of what I've done.
2004-07-03 08:16:17
Well, took me a while but made it back up to the number 0ne spot.  Actually had a few runs with an even higher percentage, but over a five run didnt make the table.

Wonga ,
You had a number one result for a while.  (That is very good).  Welcome aboard.
Stephen Brooks
2004-07-03 14:24:55
HAHAHA.  Well I can take a break from packing to congratulate Wonga in person.  And also tell "It'd be good if we got to 10 or 15% over" Nexus to have a look at the current PhaseRotB statistic.

What you did (or rather your Muon1 client did) was to locate a design for the accelerator that was superior to all the other (nearly 100`000) results submitted up to that point.  Unless you manually submitted a design to the client, this was simply a matter of chance and the particular set of results your installation was learning from.  However, the consequences of getting a best result are that it is rapidly sent out via the "sample files" to most of the other computers on the network.  So the other project members then had your design in their 'results.dat' database, meaning it set off a cascade of variations on that theme, improving the score still further.

Of course all this happens seamlessly even if you go leave your internet-connected computer in the garage for a month while you go on holiday.

I just hope my stats generator is going to be equally forgiving of my conference trip...
2004-07-03 17:21:07
Cheers for explaining that Stephen - makes me pleased I joined now Smile

Thanks for the 'welcome aboard' too, LostinTime Smile
2004-07-03 22:41:16
We're up to 16.2% better than baseline now.  The growth here doesn't seem to stop Smile Almost at 2% muon yields.  I wonder how much more this thing has left in it...
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel

Site has had 17725566 accesses.