Association of Neighboring Hits into Clusters

This step is done only for hits in the OTR, since in this detector a track usually produces more than one hit in a layer (up to four in case of double layers). The main idea behind this preliminary clusterization is to resolve left-right ambiguity at least partially and suppress combinatorial pile-up already at this early step of reconstruction.

Figure 4.3: The method for hit clusterization in the OTR.

The clusterization process is illustrated in Fig. 4.3. In order to account for drift distance information, cluster coordinates $(u^*,\;z^*)$ are calculated as a weighted sum of the coordinates of the wires participating in the cluster:

$$u^*=\frac{\sum\limits^N_{i=1}w_i(q_i) u_i}{\sum\limits^N_{i=1}w_i... ...qquad z^*=\frac{\sum\limits^N_{i=1}w_i(q_i) z_i}{\sum\limits^N_{i=1}w_i(q_i)},$$

where $q_i$ is a measured drift distance for the $i$-th wire, and the weighting function $w(\cdot)$ is given by

$$w(q)=\frac{1}{q}.$$

This dependence simply reflects the fact that the smaller the drift distance, the closer to the wire a particle passed. Of course, in case of large occupancies, the clusterization can give clusters with noise and wrongly associated hits from nearby tracks. In CATS the problem of identification and rejection of such hits is solved at the final step of the track recognition process.