nilsen [Sat, 5 Oct 2002 00:12:39 +0000 (00:12 +0000)]
Added material to simulate services in front of the SPD, SDD, and SSD support
structures. Modified folumes I212, I200, and I099 so that they do not go
beyond the volume they are creating holes for. This was nessesary so that
the material added to these volumes would be correct.
vestbo [Fri, 4 Oct 2002 14:05:24 +0000 (14:05 +0000)]
Some changes resulting from last week work. The mc stuff has been removed.
Added possibility to not save the width when compressing file; this is one
by setting the flag AliL3Compress::fWriteShape.
The status of the code is that it seems to be more or less working. The
efficiency drops a little bit when running it through offline, but this
is mainly because overlapping clusters are not deconvoluted, and 2 clusters
found by the offline cluster finder will be 1 cluster after compressing/
decompressing. The only way to get rid of this, is to include some
deconvolution in the AliL3Modeller::FindClusters.
vestbo [Wed, 2 Oct 2002 15:24:15 +0000 (15:24 +0000)]
Perform check on cos(crossing_angle_with_padrow) in GetCrossingAngle().
Every once in a while this number is equal to one, which results in
return value being NaN.
vestbo [Wed, 2 Oct 2002 14:29:15 +0000 (14:29 +0000)]
Serious bugfix in Local2Raw function. This function failed to transform
corretly on the z<0 side of the tpc, because of a "wrongly" check. I have
no idea why I did not see this bug before, but probably because the
function is seldom used.
vestbo [Tue, 1 Oct 2002 12:19:40 +0000 (12:19 +0000)]
Added new function AliL3Hough::MergeEtaSlices which merges tracks which
appear in 2 neighbouring eta slices. The tracks are matching if they have
identical kappa and psi; they appear in the same bin. The one with the
lower threshold are removed from the track array.
There is also a bugfix in AliL3HoughTransformer::GetEta, the sign should
not be swicthed on the other half of TPC, because the sign is already included
in the (GetEtaMax() - GetEtaMin()).
Also the DoMC flag in the AliL3HoughTransformer ctor.
Fixing the previous unresolved bug before I was going on vacation.
When using the AliLevel3 tracking class and reading from a rootfile
with the patched AliTPC (ASV) version, one can only read 1 whole slice
at once (npatches=1). This because in AliLevel3 class only 1 instance
of the AliL3FileHandler is taking care of reading the data from all
patches, and therefore the AliL3FileHandler::fLastIndex method fails when
reading patch by patch. This problem did not show up in the AliL3Hough,
because there you have one AliL3FileHandler per patch.
Check if the provided path is a rootfile or a directory by using the
IsZombie() function; the file pointer is not zero if the provided
string is a directory.
Use global pointer gSystem instead of creating a new instance of the
TUnixSystem class. This in order to get rid of a error message during
runtime.
Changed the row numbering in the patches to match the latest technical
layout as much as possible. The numbers are taken from mapping tables
provided by GSI (Dr.Frankenfeld). The implementation is still not identical
to the actual foreseen layout, as there are some pads (FEC) which are
physically located in the area we have been calling patch 1 which actually
belong to patch 0. But this is only concerns 1 or 2 FECs as far as I could
see from the tables.
More changes as a consequence of changing the pad geomtry in the new
AliROOT. Some hardcoded values which were set to 176 has now been changed
to AliL3Transform::GetNRows() etc. Also the remaining hardcoded AliTPCParam
and digits tree object names has been changed in AliL3FileHandler.
There were and still are a bug in the AliL3Merger class in FillNTuppel,
which I was not able to figure out, so I switched off. This only affects
the SlowMerging methods in the track mergers, and should be looked at more
if needed in the future.
git://git.uio.no / u / mrichter / AliRoot.git / log