10. Decontamination algorithm¶
A correct determination of the most probable true cluster members, allows to recover a much clearer picture of a cluster’s photometric structure. It is also essential to ensure a more accurate and all around better estimation of the cluster’s fundamental parameters.
Since it is not possible to unequivocally discriminate between true members and field region stars from photometry alone (even when proper motions or radial velocities are available), the decontamination algorithms (DA, also membership probability assignment algorithms) in the literature usually give each star in the cluster region a probability of being a true member.
The process by which these membership probabilities (MP) are assigned by the code, is explained in detail in the article where ASteCA was introduced. The algorithm is based basically on a Bayesian scheme, and the repeated estimation of a ‘clean’ cluster region by means of a random removal of stars.
The first four binning schemes that can be used by the DA are taken from the astroML package; the rest are well known rules:
blocks: Bayesian Blocks.
knuth: Knuth’s rule.
scott: Scott’s rule.
freedman: Freedman-Diaconis rule.
sturges: Sturge’s rule.
These same binning methods are made available to the Dolphin likelihood equation in the Best fit block, see Sect. 11.2.
Once the DA process has finished, every star inside the cluster region will
have a MP associated to it. These MPs are stored, along with the data on each
star in the cluster region, in a file named
is the name of the analyzed cluster).
10.1. Reduced membership¶
After MPs have been assigned to all the stars within the cluster region, what follows is the selection criteria, referred here as the “Reduced membership” algorithm. This process determines which stars, among all those located within the cluster region, will be used by the best fit process (see Sect. 11).
Several options are available in ASteCA, associated to this function. These
options are handled by the
Reduced membership block in the
local: combines the assigned MPs with a cell by cell removal of stars in the photometric diagram. The stars discarded are those with lower MPs and the number is equal to the excess of field stars present in each cell.
n_memb: use the N stars with the highest MPs, where N is the approximate number of members found via the cluster region vs field region star density comparison.
mp_05: selects only those stars with MP>=0.5, i.e.: stars with higher probability of being a cluster member than a field star.
top_h: selects the top half of all stars in the cluster region with the highest MPs.
man: the minimum probability value has to be set manually via the
prob_minparameter (0, 1).
mag: switches the
prob_minvalue to a
mag_minvalue so the user can select a minimum magnitude to be used in the best fit function, instead of a minimum probability (the “minimum magnitude” value is the value of the dimmest star wished to be taken into account in the fitting process).
skip: use all stars in the cluster region.