stephenbrooks.orgForumMuon1GeneralGood news and bad news
Username: Password:
Search site:
Subscribe to thread via RSS
Stephen Brooks
2004-05-20 02:29:12
The good news is: I've tested one of the recent ~10.25% designs from SolenoidsTo15cm and they are repeatable, not a fluke or an error.

The bad news is split into two parts.
The part you probably already knew was that v4.4 is slower and disagrees with v4.34 on the percentages for SolenoidsTo15cm because it lets the solenoid fields extend a large distance from the magnets rather than clipping them at the entrance of the adjacent solenoids, as v4.34 did.  The way I've got Muon1 to repeat the results properly is by changing the solenoid clipping essentially back to how it was (except less cludged) in the v4.41 development source.
The other part of bad news is that there's a stupid error in the "...A" lattice files themselves, so when v4.41 comes out we will be correcting that, with PhaseRotB and ChicaneLinacB, simulated using the proper clipping (so SolenoidsTo15cm will agree with v4.34 again) and also back to v4.34 speeds because of that.
[TA]z
2004-05-20 06:04:21
Thanks for the update Smile

Is it not possible to fix the *A.txt lattices simply by replacing them?  And have those lattices caused bad results to be produced?
nukefarmer
2004-05-20 08:11:52
thanks stephen,
any indication on when we can expect v4.41?
Stephen Brooks
2004-05-20 10:57:08
Replacing the lattice files would fix the phase-delay bug, but not the solenoid bounds problem, which is inherent in v4.4. Since I want to fix both at once, the ...B versions will include a "!MINVER 4.41" directive to make sure they are all run with the same assumptions.  There's still the [v4.4] version identifier so the SolenoidsTo15cm ones can be told apart, if possible.  The current stuff in the ...A optimisation section of the database is bad results, or at least results that would need some funny interpretation.

The week or so of optimisation will have to be restarted for the B's, but if [TA]z's estimate of how much slower v4.4's clipping made the simulation, we ought to be able to do the same work in 3 days with 4.41... which I'm estimating will be out next week.
[TA]z
2004-05-20 11:05:45
Good to hear Smile

Do you think you should pull the 4.4 download from the main page being as the 'A' results are hosed?
kitsura
2004-05-21 00:41:31
Not pleasant news for early upgraders it seems...
[DPC]TeamNWW - Huub
2004-05-21 04:07:42
Too bad.. but some people have to be first i guess.. or nobody would find the bugs Smile

I just hope the 4.4 results so far are of any use to Stephen..
kitsura
2004-05-21 04:20:30
True the early adopters are also those who uncover bugs first.  Anyway I will migrate back to 4.34 since 4.4 PhaseRot and ChicaneLinac won't be compatible with 4.41 so back to square one it seems.
Herb[Romulus2]
2004-05-21 05:05:58
That's no problem for me, I'll just take it as it comes Big Grin

Ok, now who's going to have a 10.3x at first Big Grin and who is going to be second in Monaco on Sunday ?  Wink
Stephen Brooks
2004-05-23 08:43:41
quote:
Originally posted by Herb[Romulus2]:
Ok, now who's going to have a 10.3x at first Big Grin and who is going to be second in Monaco on Sunday ?  Wink
I didn't see who got there first - maybe you did?  Anyway, forget 10.3, it seems to be heading for 10.4 at some speed now!

[edit] and Mr. O seems to have got more than O particles in PhaseRotB now too.
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25161682 accesses.