stephenbrooks.orgForumMuon1GeneralMisc.
Username: Password:
Search site:
Subscribe to thread via RSS
excaliber[Free-DC.org]
2003-09-29 17:13:16
Firstly, cool project.  Smile

Secondly, the new client doesnt work with MuonMonitor, because of the switch thing.  Not a big deal, just thought i'd let you know.  Actually, it should be the writer of MuonMonitor's job to change it...I'll email him.

How many entries should I have in my results.dat?  Im running the newest one . Should I erase my old results to use the newer Best100 from the website?  Or append them together?  More the merrirer, or less?

I've read the manual page, but Im still confused.  Frown

Thanks.

Humph...still getting only 2% or so muon yield.  Others are getting 8%!  Frown

[This message was edited by excaliber[Free-DC.org] on 2003-Sep-30 at 1:34.]
Pollock[Romulus2]
2003-09-30 21:11:53
excaliber, Muonmonitor was almost worthless even on the 4.31 version.  Half of the functions wouldn't work.  If you just want to filter or merge the results files, [AMD Users] Michal Hajicek posted this tool a while back and it works well on the 4.32 version for me.  One function would need to be updated for the new version, but it is not a critical one. 
There are a few other ones available and even some in different languages if needed.  Not sure if the others work with the current version.
Have you downloaded the 'best 100' file on the top of the stats page?  That would improve your results almost immediately.  Just right-click it and 'save target as' then rename it to results.dat and drop it in your Muon folder.  The zip works the same way but must be unzipped first.  It will overwrite your current .dat file when you do that.  If you want to merge it in with your current results, the tool above does that just fine.
BTW: Some of your results are in the best 100 list and look promising.  They just need some tweaking.
excaliber[Free-DC.org]
2003-10-01 07:51:52
Thanks for the reply.

Didnt know that MuonMonitor was obsolete.  I might take it upon myself to create a clone, but current...may be fun Smile

As to the Best100, yes, I downloaded it.  I merged it with my older one, and now have 500 some results inside the results.dat.

Im reaching about 8.96-ish on the SolenoidsTo15cm.  Making the top 10 on the stats.  Smile


How do you "tweak" the results?  Can that be done manually?  Or do you have to rely on the computer to tweak it for you?  Thanks.  Smile
Pollock[Romulus2]
2003-10-01 15:56:02
It can easily be done manually with the 4.32 version.
The older versions were a real challenge to tweak, but this one was designed for it.

Just find a result that you want to work with and COPY/paste it into Notepad or Word.  DO NOT CUT/paste or the result will be removed for good.
In Notepad, remove everything from the second line except < SolenoidsTo15cm > and add the letters TEST before it.  The second line should look like this:
TEST< SolenoidsTo15cm >
with no spaces.  Then, at the end of the first line, make sure to remove this:
#runs=5;
it may or may not be there, depending on whether it was a re-checked result.
After that you can change any of the parameteres you wish.  You must be sure that there are exactly three nubers after every = character and there are no extra spaces between characters.  DO not alter anything before the = characters, only the three numbers after it. 
When you have changed what you wish, save the file with the name:
queue.txt
in your Muon folder.  After the current run is completed, the client will look for a queue file and start running it.  It will be re-checked five times and if it turns out to be a bad one, you will not get credit (pts.) for it if you stop it before five re-checks are completed.  It will be refused when you send it for a bad checksum.  If it runs through all five checks, it will automatically be entered in you results and sent just like the rest. 
WARNING: If you create the queue.txt file and rop it in the Muon folder while the client is running, it may be overwritten and lost.  If the result that is running is the highest score at the time, the client will trash your queue and create one for the result that it created. 
If you want to start it immediately, shut down the Muon client.  Delete any files named auto.sav or autocli.sav if there are any, or they will restart the current run.  Drop your queue.txt in the Muon folder and restart the client and it should run immediately.  If not, you did something wrong like leaving a space or entering more or less than three numbers for a value. 
If you use the command-line client, it will read 'Rechecking quarantined result' if it is running properly.  That's good.
The background client will not give any indication until after the first run is completed.  At that point, if you open the queue.txt file, the TEST will be removed and the score and checksum will take it's place.  It is much easier to be sure with command-line, but slightly slower.
This may sound complicated but after trying it once or twice it will be very simple.

[This message was edited by Pollock[Romulus2] on 2003-Oct-02 at 0:06.]

[This message was edited by Pollock[Romulus2] on 2003-Oct-02 at 0:06.]
Pollock[Romulus2]
2003-10-01 16:01:50
Sorry but apparently the proper text will not appear on the message board.
TEST< SolenoidsTo15cm >
I had to alter this line because the characters < and > will not show up on the board when used as brackets.  Omit the spaces before and after SolenoidsTo15cm or the spaces will cause a read error in your que.
excaliber[Free-DC.org]
2003-10-01 17:57:36
Sounds fun.  may have to start tweaking.

How do i know what values to edit, and what to change them to?  I have no idea what any of this information means.

Wouldnt the computer do a better job of tweaking the results?

Thanks
Pollock[Romulus2]
2003-10-01 21:25:04
If you open the results.dat file, you can look stright down the columns and see which values are changing as the results climb.  The software-generated changes are usually only minor changes.  You will notice that about 95 of the 'best 100' are very similar.  The tweaking allows for larger changes to speed up the climb.  You will also see that most of the highest results at this point are 'maxxed out' or the values are close to 999 or 000. That was likely done by tweaking.  It can push the values to the limits and then gradually refine them.  At the moment, I am working on maxing out some other configurations.  Some have produced some wild results, but not the kind that show on the stats.  Hopefully, they can grow to mix in with the current breed to generate some variations.  If you want to try it, look for some results in the 2.00 or 3.00 range in your .dat file.  Try maxing them out to see if it helps or hurts.  The smaller ones won't take as long to run and produce the most noticable improvement.
Examples:
If the current value is 810, change it to 999
If the current value is 180, change it to 000
The first two tantalumrod values seem to work best in about the 500 and 200 range, but that could change.

I hope some of us come up with some different stuff rather than everybody running the same stuff over and over.  It's more fun with two or three competing scenarios.
Pollock[Romulus2]
2003-10-01 21:32:37
Another possibility that I use at times;
If you happen to see a result that appeals to you, just use only that result as your results.dat file.  You can just rename your current .dat and it will be ignored if you want to keep it.  With only one result in your .dat file, you will generate a lot of zeros and a few similar results to the interesting one.  In a short time, they will multiply along the same parameters as the original.  If you like, the zeros can be removed as they progress.
excaliber[Free-DC.org]
2003-10-02 08:57:12
Heh...that sounds really interesting.  Why are maxed out values better?  Or do they just produce more dramatic results, which I can then fine tune later?

What exactly are tantle-whatchacallits rods?  Thanks
David
2003-10-02 17:10:22
quote:
Originally posted by excaliber[Free-DC.org]:
[snip]
What exactly are tantle-whatchacallits rods?  Thanks


You probably should read the "Background Information" page on this website.

In summary, the tantalum rod is the initial target, hit by another particle beam, which is used to generate the particles that we simulate here.  (Which themselves decay...)
Pollock[Romulus2]
2003-10-02 22:00:55
quote:
Heh...that sounds really interesting.  Why are maxed out values better?  Or do they just produce more dramatic results, which I can then fine tune later? 

That's it.  It just speeds up the development if you hit it right.  If not, it will be a junker and will run faster. 

David, I'm glad that you explained that.  I understand how it works and what it does, but don't ask me to explain it.
[AMD Users] Michal Hajicek
2003-10-03 01:55:39
OK.  Tool updated.
Michal
David
2003-10-03 10:38:19
quote:
Originally posted by Pollock[Romulus2]:
[snip]

David, I'm glad that you explained that.  I understand how it works and what it does, but don't ask me to explain it.


Understand it?

Not at all - I may have scored as much as 0.2 % on my particle physics finals...

David
Pollock[Romulus2]
2003-10-03 22:48:15
Thanks, Michal, that worked well.  Cool

David, I must have been cheating from your paper in school.  Big Grin

Looks like the tweaks might be worth it.  I have one running 5.40 wide open and haven't even started on the main section yet.  Should be ready soon if I find the time to pay with it.
[DPC]TeamNWW - Huub
2003-10-09 01:06:29
Ok.. tried some manual tweaking last night.. but i ran into a bit of a problem.. when i checked this morning what the result was of my manipulations i found the queue.txt file still existed!!  It had run only two times!!  Muon just continued with normal runs ignoring the queue.txt.. anybody an idea how this happened??

Oh.. some details.. first line now ends with "#runs=2;#queued=1;"

Edit:
I put more info this thread: http://www.stephenbrooks.org/groupee/forums?a=tpc&s=724606111&f=3126012111&m=1256057844 in the 'bugs'section..

[This message was edited by [DPC]TeamNWW - Huub on 2003-Oct-09 at 13:24.]
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25164033 accesses.