class PrepfoldSigmaRater(base.BaseRater):
short_name = "prepfoldsigma"
long_name = "Prepfold Sigma Rating"
description = "Re-compute the sigma value reported in the " \
"prepfold plots."
version = 1
rat_cls = profile.ProfileClass()
def _compute_rating(self, cand):
"""Return a rating for the candidate. The rating value is
the prepfold sigma value.
Input:
cand: A Candidate object to rate.
Output:
value: The rating value.
"""
prof = cand.time_vs_phase.get_profile()
pfd = cand.pfd
prof_avg = np.sum(pfd.stats[:,:,4][:pfd.npart])
prof_var = np.sum(pfd.stats[:,:,5][:pfd.npart])
chisqr = presto.chisqr(prof, pfd.proflen, prof_avg, prof_var)
df = pfd.proflen - 1
#print "DEBUG: prof_avg, prof_var, chisqr/df", prof_avg, prof_var, chisqr/df
prob = sp.stats.chi2.sf(chisqr, df)
sig = -sp.stats.norm.ppf(prob)
return sig
class DutyCycleRater(base.BaseRater):
short_name = "dutycycle"
long_name = "Duty Cycle Rating"
description = "Compute the duty cycle, that is, the fraction of " \
"profile bins in which the value is more than " \
"(max+median)/2."
version = 4
rat_cls = profile.ProfileClass()
def _compute_rating(self, cand):
"""Return a rating for the candidate. The rating value is an
estimate of the profile's duty cycle.
Input:
cand: A Candidate object to rate.
Output:
value: The rating value.
"""
profile = cand.get_from_cache('profile')
nbin = float(len(profile))
thresh = (np.amax(profile) + np.median(profile)) / 2.0
dutycycle = np.sum(profile > thresh) / nbin
return dutycycle
class RescaledPrepfoldSigmaRater(base.BaseRater):
short_name = "rescaledpfdsigma"
long_name = "Rescaled Prepfold Sigma Rating"
description = "Rescale the sigma value reported in the " \
"prepfold plots using a better estimate of" \
"the off-pulse chi-square."
version = 1
rat_cls = profile.ProfileClass()
def _compute_rating(self, cand):
"""Return a rating for the candidate. The rating value is
the prepfold sigma value.
Input:
cand: A Candidate object to rate.
Output:
value: The rating value.
"""
tvph = cand.get_from_cache('time_vs_phase')
prof = tvph.get_profile()
pfd = cand.get_from_cache('pfd')
prof_avg = np.sum(pfd.stats[:, :, 4][:pfd.npart])
prof_var = np.sum(pfd.stats[:, :, 5][:pfd.npart])
#chisqr = presto.chisqr(prof, pfd.proflen, prof_avg, prof_var)
# Below is adapted for presto v2.1
chisqr = presto.chisqr(prof, prof_avg, prof_var)
df = pfd.proflen - 1
redchisqr = chisqr / df
off_redchisqr = pfd.estimate_offsignal_redchi2()
rescaled_redchisqr = redchisqr / off_redchisqr
#print "DEBUG: prof_avg, prof_var, chisqr/df", prof_avg, prof_var, chisqr/df
prob = sp.stats.chi2.sf(rescaled_redchisqr * df, df)
sig = -sp.stats.norm.ppf(prob)
return sig
class PeakOverRMSRater(base.BaseRater):
short_name = "peakoverrms"
long_name = "Peak Over RMS Rating"
description = "Compute the peak amplitude of the profile divided" \
"by the RMS amplitude. Specifically, compute " \
"(max(profile)-median(profile))/std(profile)."
version = 3
rat_cls = profile.ProfileClass()
def _compute_rating(self, cand):
"""Return a rating for the candidate. The rating value is
the profile peak divided by its RMS.
Input:
cand: A Candidate object to rate.
Output:
value: The rating value.
"""
prof = cand.get_from_cache('profile')
return (np.amax(prof)-np.median(prof))/np.std(prof)
class ubc_pfd_ai(base.BaseRater):
short_name = "pfd_AI"
long_name = "UBC pfd AI"
description = "compute the prediction from the pulsar classifier " \
"based on pfd files."
version = 5
rat_cls = profile.ProfileClass()
def _compute_rating(self, cand):
"""Return a rating for the candidate. The rating value is
the prepfold sigma value.
Input:
cand: A Candidate object to rate.
Output:
value: The rating value.
"""
pfd_fn = cand.get_from_cache('pfd').pfd_filename
pred = classifier.report_score([pfdreader(pfd_fn)])
return pred
