Preliminary Conclusions

The analysis of these results is still going-on, but we can set some preliminary conclusions:

• The best results overall correspond to linefit, both when we compare the first guesses and the results from the log-likelihood algorithm fed by any of the first guesses.

• The processing time of linefit and dipole fit are shorter (linefit is a bit faster) than in JAMS and DW, as expected. Moreover, the time for each Cluster JAMS call is much longer (> x10 wrt direct walk).

• Likelihood reconstructions take a factor 100 more time than linefit or dipole fit (but actually less than Cluster JAMS).

• Efficiencies at trigger level are similar among the first guesses (almost 100%), except for direct walk, which is 65% (even less at low energies).

• The best results with fake rates correspond to direct walk, but the differences (0.99% vs 1.04% in linefit) is not compensated by the differences in efficiency.

• The tuning of jams and direct walk is still on-going, which could improve thier performance.
  

• In order to tune the data rate sent via satellite, we are studing the effect of a cut in the zenith angle and the threshold in the number of DOMs to trigger the detector. A maximum rate of 160 kB/s of satellite bandwidth can be tranlated into ~16 ev/s for a 8-DOM threshold. This would require a cut of 65 deg in the zenith angle (maybe a bit larger, when the AMANDA data is taken into account). This cut at 65 deg would keep  ~95% of a E-2 cosmic signal.
  

• For detailed comparisons among the strategies, please, look at the plots!