Simulations are performed on the scan 044 of the night 991024s.They consist in adding a component to each star shifted by 2'' and with 1 magnitude fainter than the primary star. A second simulation is done for a 2.8'' shift.
This scan has been chosen for its very good seing and represents the best case.
 

I. Fraction of deblended sources

9110 sources are found in the J band and the magnitude vary from 6.93 to 25.7. Obviously faint source are not reliable
and will not been deblended.  206 and 301 sources are actively deblended for the 2.0'' and 2.8'' simulations, respectively.
These numbers seem small but the fraction of deblended sources versus the magnitude presented in Figures 1a and 1b shows that the fraction in the useful magnitude range (SNR > 10) is actually important. The magnitude used in Figures 1 is the magnitude of the primary; the secondary is 1 mag fainter.
 
 

Figure 1a	Figure 1b

Figure 2 shows the evolution of the chi2 before and after the deblending.
 

Figure 2

 

II.  Distance / Delta(mag) between the two components

For both the 2'' and 2.8'' simulations prophot found the correct distance between the primary and the secondary and the correct delta(mag) between the two components.
 

Figure 3a	Figure 3b

 

III.  Comparison between the sum of the primary and secondary flux with the true flux

The true magnitude is simply obtained by adding the flux of a source with the same source 1 magnitude fainter. This value is compared to the sum of the flux of the two components found by prophot.
 

Figure 4a	Figure 4b