the problems with the b-field that i reported earlier this month were
not due to the fieldmap nor its extension with with biot-savart
calculations; the map was not extended far enough. as you can see in
the attachment, the strange effects of the edge are due to quadratic
interpolation on the 5cm x 5cm grid (black).
chi has gone through and extended the range in phi with a simple
quadratic extrapoltion on the boundaries.
while the fit isn't perfect (see the attachment, red), it should be
adequate for our purposes. it is extended even past the physical
boundaries of the coils in order to give the newton fitter a larger
radius of convergence. the changes have been checked into cvs, and
should be used in the next recrunch.
--chris
-------- Original Message --------
Subject: bgrid2.blast
Date: Sun, 8 Aug 2004 04:30:52 -0400 (EDT)
From: Chi Zhang <zhangchi@MIT.EDU>
To: Chris Crawford <chris2@lns.mit.edu>
References:
<Pine.LNX.4.44.0407271559270.1361-200000@nucleon2.lns.mit.edu>
<1091751179.4112cd0b1cc01@webmail.asu.edu> <4113A884.6030609@lns.mit.edu>
Hi Chris,
I made a new bgrid file with a lot of the 0's plugged. I basically
quadratically extended the outmost 3 non-zeron points in Y for each slice
of fixed X and Z.
The file is too big to send via email, please get it from
~zhangchi/bgrid2.tgz or bgrid2.dump2.bz2. they both extract to an ascii
called bgrid2.dump2. You ll need to replace yoru bgrid2.blast with it.
I wonder if you can test it with you codes to see if the bump in Bfield is
fixed. If it works, please check it in and release for blast account.
if it does not work, let me know.
If this thing works, I plan to recrunch those seg-faulted runs with head
version and the new bfield file. I think if things work well, we should
update to head.
Chi
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST