Dear Hirouyuki,

thanks a lot for your effort to study the results of my code. When you
analyze the data with my pattern recognition and my fitting code, you
have to take into account some hints in order to achieve the results:

1) The new pattern tries to reconstruct both the tracks coming from K-
stopping point and the tracks outside the targets (as pi- and p of
Lambda). If you are evaluating the track quality,  you have to select
only the tracks coming from K- stopping point (for this selection you
can use the minimum 3-D distance between the track and K- stopping
point). Otherwise, the comparison among different type tracks is not
fair. By the way, to study properly the tracks outside the target you
have to consider the secondary vertex. I mean the 3D crossing point
between two helices. In the v6.00 fidarc I have not inserted this
feature yet. Coming soon.

2) Using only three points to fit the helix could be dangerous if are
not using any Chi² selection. Tracks with 3 points have only one degree
of freedom. If you don't use Chi² and topology selection (K- stopping
point or secondary vertex, which is another point belonging to the
track), please, don't consider tracks with less than 4 points. I suggest
to see the Fig. 2,8, [9,10], 13,14 with a cut on Chi² and a cut on
minimum distance [see 1)] .

3) When a track "skips" a chamber, you have to take into account the
Chi² in order to eliminate tracks with high energy loss.

4) The fitting procedure for the Isim-Isim-Osim type is not working
(Fig. 17), I left this kind of track only for development purpose.
Therefore, the main selection on "TYPE" flag should be > 1000.

Let me done just few remarks about your conclusions:

5) Your 2nd conclusion is wrong. Adding ISIM hits means that your track
is backward (more energy loss), so the resolution is poorer than a
forward track (without ISIM hits). You have to compare the same backward
track sample fitted with and without ISIM. Adding ISIM hits improves the
fitting quality (you can comment out the connection of ISIM in the
pattern recognition code and test my statement).

6) Same as 5). To compare different TYPE resolution, you have to use the
same track sample (11112) fitted with different pattern 11112 - 11110 -
01110 and so on...

7) For your 3rd conclusion see 2).

cheers,
Stefano