[mcstas-users] bug in pol_bender

[Dalgliesh, RM (Robert)]

The attached code illustrates a considerable set of bugs in the polarising =
bender code.

The components seems to be ok (ish) if the number of channels in the bender=
 is > 1 but it is still possible to get neutrons trapped when a large radiu=
s of curvature is used.
I've had to put in a check to kill neutron that bounce excessively otherwis=
e simulations stall.

I haven't managed to track the source of the bug as there are such a large =
number of call outs to other routines within the code particularly to the r=
outines that work out cylinder intercepts and gravity.

Basically there are a number of manifestations.

1) Neutrons can pass out of the guide without being reflected (run a trace =
to see this)
2) in an open guide neutrons can get trapped in open space
3) neutrons multi-bounce along walls for no good reason.

This is all accentuated when a large radius of curvature is used.
Sorry to be no help at all on this one.

Rob

---------------------------------------------
Dr. Robert Dalgliesh
ISIS
Rutherford Appleton Laboratory
Chilton
Didcot
Oxfordshire
OX11 0QX
Tel: +44(0)1235 445687
Fax: +44(0)1235 445720
e-mail: r.m.dalgliesh at rl.ac.uk =




-- =

Scanned by iCritical.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: bender_bug.instr
Type: application/octet-stream
Size: 3801 bytes
Desc: bender_bug.instr
Url : http://eagle.zigzak.net/pipermail/mcstas-users/attachments/20081113/e=
e7e8274/bender_bug.obj