search

JohannesBuchner / nway

nway -- Bayesian cross-matching of astronomical catalogues
Other
56 stars 14 forks source link

p_any #8

Closed lib-j closed 5 years ago

lib-j commented 5 years ago

Hi, Could you clarify if a good p_any should be a high or low value? The manual on page 11 suggests that the p_any value should be low. Whilst the calibrate cuttoff example (below) suggests that only large values should be kept.

python3 ~/.local/bin/nway-calibrate-cutoff.py example1.fits example1.fits-fake.fits created plot "example1.fits_p_any_cutoffquality.pdf" created plot "example1.fits_p_any_p_i.pdf"

For a false detection rate of <5% --> use only counterparts with p_any>0.66 (0.51% of matches)

Thanks.

JohannesBuchner commented 5 years ago

Thank you, the manual is inaccurate there from a previous version.

p_any should be high. It is the probability that one of the association is the correct one.