Working on making sure that the clustering method in hcana matches what we get from the clustering method in the VTPs.
One reason that the x and y values for clusters in hcana are off because the step size for each block is too big. It's currently set to have each block be 9x9cm. I believe the blocks are about 2x2cm each, but I'm not positive.
I took 2 events from run 1467 which had only 1 cluster in the VTP and had a large cluster energy. I tried to compare the properties of these clusters according to hcana and according to the VTP output.
Event 3431
Event 3527
Both of these are 3x3 clusters according to the PMTs listed by hcana. It's reassuring that the block listed by the VTP is included in the blocks in the hcana clusters, but for some reason hcana doesn't see them all as one cluster even though they are all neighbor blocks (different cluster ID #s). The x and y coordinates of each "cluster" are indicated in the histograms by red lines. The x values of the clusters seem reasonable, being neighboring columns, but the y values of the clusters don't make sense, particularly in event 3527.
I also enabled the debug output (nps_dbg_clusters_cal) and tried doing a short replay with it on. It crashes after 7 events, I think because there's a hit listed with a negative energy. However, I was still able to confirm that the "center of gravity" calculations to get the cluster position and time variables work correctly. This makes me think that the values for individual hits are incorrect, not the clustering calculations.
Event 5
Debug output
Output from root file
Here we see that the debug output and what's saved in the root file are not consistent. Cluster 2 from the debug output corresponds to cluster 1 from the root file. Both have 4 hits and the same total energy, and the position and time are consistent, but the blocks listed as part of the cluster are not. The debug output lists the blocks in rows 6 through 9, column 16, which are blocks 196, 226, 256, and 286. The hcana output in the root file lists blocks 545 through 548, i.e. row 18, columns 5 through 8. This almost looks like the coordinates are flipped, but even if they are, they're still off by 2 rows/columns in one coordinate and 1 column/row in the other.
Other things to note:
Energy values for hcana are very small. I'm not sure what the units are supposed to be for this, but a hit's energy in hcana is calculated as the hit's pulseInt*gain. Right now the gain is set to 0.001 for all blocks.
Need to implement time clustering. The debug output for event 5 shows a hit at 172ns and a hit at 313ns being grouped into the same cluster.