Gregory & Loredo Algorithm
This implements the algorithm described by
Loredo (1992, ApJ, vol. 398, pp. 146-168) to evenly divide a
lightcurve, and determine whether this description is better than a
- sitar_glvary( t, [; tmin=#, tmax=#, mmax=#, thresh=#,
nbins=#, texp=array, frac_exp=array]);
Use the Gregory-Loredo algorithm to find the odds ratios that
even divisions of a lightcurve are better descriptions of the
data than a constant lightcurve. A 'best estimate' lightcurve
can also be output for a lightcurve with nbins.
isis> fold = sitar_glvary(t; tmin=#, tmax=#, mmax=#, thresh=#,
nbins=#, texp=array, frac_exp=array);
Variables in  are optional qualifiers.
Omitted qualifier variables take on default values.
Optional Qualifier Inputs:
tmin : The minimum time to consider (Default=min(t)-1)
tmax : The maximum time to consider (Default=max(t)+1)
mmax : Consider lightcurve divisions from 2 to mmax evenly spaced
thresh : Truncate the maximum number of partitions considered by ignoring
partitionings for which--
\sum_m(odds ratio) < max(\sum_n(odds ratio))/exp(thresh) ,
where n = 2 -> mmax. I.e., the more partitionings you have,
the less significant the results tend to become. thresh
sets a minimum probability: (1-p_min) ~ exp(thresh)*(1-p_peak).
nbins : Create an output lightcurve with nbins. (Default=mmax)
texp, : A pair of arrays that give values for the fractional exposure
frac_exp as a function of time. *** Only non-zero values of exposure
will be retained ***, which will then be interpolated and used
to correct the lightcurve rates. These arrays must have a
minimum of five entries. (Default is for no correction.)
gl.p : Total probability that some evenly partitioned lightcurve, with up
to gl.mmax bins, is a better description than a constant lightcurve
gl.ppart : The probability for an individual evenly partitioned lightcurve
that it is a better description than a constant lightcurve
gl.lodds_sum : The natural logarithm of the sum of the odds ratios comparing
lightcurves with two or more partitions to a constant lightcurve.
gl.p == exp(gl.lodds_sum)/[1+exp(gl.lodds_sum)]
gl.mpeak : The number of partitions for the evenly partitioned lightcurve
with the maximum probability.
gl.mmax : The maximum number of partitions actually used (influenced by
the setting of the thresh parameter)
gl.m : The number of partitions corresponding to each evenly partitioned
lightcurve considered (=[2:mmax])
gl.pm : Total probability that some evenly partitioned lightcurve
is a better description than a constant lightcurve for each
maximum number of partitions considered ([2:mmax])
gl.nj : The counts histogram corresponding to each partitioning above
gl.aj : The integrated fractional exposure for each partitioning above
gl.a_avg : The averaged fractional exposure.
gl.tmin : The value of tmin actually used (maximum of [tmin,min(texp)]);
gl.tmax : The value of tmax actually used (minimum of [tmax,max(texp)]);
gl.tlc : The output lightcurve times (an array with input nbins bins)
gl.rate : Best estimate of the lightcuve rates at the above times
gl.erate : Best estimate of the lightcuve rate errors at the above times