The method used is a fit of hyperplane through the cloud of points formed by the number of counts versus PIF fraction for every source. In the case of one source this is just a linear regression. The intercept gives the background, while the slope gives the flux of the source in one fully illuminated pixel. Normalization is a number of counts in a perfect detector if source is on axis (half of all detector pixels illuminated).
There are two methods to extract the flux for one source:
1. the source flux is extracted considering the source alone.
2. the source flux is extracted taking into account that other sources
are in the FOV.
For a faint source in the PCFOV it is important to take into account
all bright (typically larger than 0.1 Crab) sources in the entire FOV, whereas to analyse a faint source in the
FCFOV it is necessary to consider only bright sources in the
PCFOV.
The sources that need to be considered simultaneously for the lightcurve extraction are the ``essential sources'' (the DOL to the list is stored in source_selectDol parameter). The list however should include the lower possible number of sources (defined through the maxessential parameter). This is due to the fact that extracting simultaneously the flux of many (N) sources takes a longer (of order ) computation time than extracting a source alone.
Care has been taken so that the output structure is compatible with HEASARC tools.
There are no limits on the size of time bin (up to 0.1s), and number of energy bands.
ii_light is not the official lightcurve extraction tool and should be used mainly to check relative variability of bright sources within a given Science Window, rather than for a long term absolute flux estimate. An extended comparison between the official lightcurve extraction tool (ii_lc_extract) and ii_light can be found in Grinberg et al. 2011 [15]