IU and TSIU Algos: Procedural Comparison


IU Algorithm

  1. Initialize "seed floors" to zero
  2. Starting with strip 3 and going until 3 before the last strip
    1. Find seed strips, i.e. a strip above (nominal threshold + seed floor energy)
    2. For each seed, raise the seed floor energy for the strips nearby the seed before looking for another seed
  3. Sort the array of seed strips according to energy
  4. Loop through the seed strips, starting with the highest energy strip
    1. Ensure that no seed strips are within given number of strips from another seed (start w
    2. Make clusters around each seed by including +/- 3 strips
    3. Require a minimum number of strips in the cluster to have non-zero energy (i.e. >3sigma above pedestal)
  5. For clusters that share strips, readjust the cluster energies to account for the shared energy in the shared strips.
  6. Cluster mean position is the energy weighted position of the 7 strips in the cluster, and the cluster energy is the sum of the strip energies

TSIU Algorithm

Copied from blog in July.

  1. Apply Tukey-smooter with mNumSmoothIters (10) iterations. [TSP]
  2. Let all strips with energy above threshold, mSeedAbsThres (2 MeV) [IU], and which have higher energy than their neighbors [TSP], be defined as seed strips .
  3. Associate mNumStripsPerSide (3) strips on either side of the seed as part of the cluster. [IU]
  4. Require that the cluster meet the following cuts:
    1. The number of non-zero strips (before smoothing) is greater than mMinStripsPerCluster (5 strips) [IU]
    2. The total energy in the cluster is greater than mMinEnergyPerCluster (3 MeV) [TSP]
    3. The energy of the strips (after smoothing) monotonically decreases from the seed.

Procedures in IU but not in TSIU

Seed floors

There are no seed floors.  The purpose of this was to discourage multiple clusters too close together.  Keith tried modifying the seed floor parameters, and found very little effect.  Varying the cluster size (+/- 3 strips) had a large effect.

The seed floors were also the only place Weihong did things differently depending on whether the SMD layer occured 1st, 2nd or 3rd in the SMD--the energy floor is double in one case verses the other.  The studies to determine these seed floor shapes were done with the iron/air simulations, and so may not represent the true case.

In the TSIU algorithm, the requirement for monotonically decreasing +/- 3 strips from the peak reduces the amount of clusters found close by, and in fact, requires the seeds to be at least 6 strips apart.  There has been no indication amoung all the studies performed suggesting any need to consider varying the parameters of the algorithm based on the layers position in the SMD (1st, 2nd, 3rd, before or after spacer, etc.).

Sorting the strips and requiring them not to be too close

The sorting was only used by the IU algorithm to faciliate dropping the lower energy seeds that are too close to other seeds.  In the TSIU algorithm, clusters are only made around seed strips that meet the monotonically decreasing requirement, which enforces a 6 strip seperation for seeds.  The IU miniumum seed distance cut was three strips, so that cut is automatically enforced by the monotonically decreasing cut.


Procedures in TSIU but not in IU

  • Smoothing
  • Seeds have higher energy than their neighbors
  • Cluster energy requirement
  • Strip energy monotonically decreases from the seed strip