5. The operative recipe

Once we have identified the optimal parameters for maximum SNR gain for faint point sources, we pass on to describe operatively the screening algorithm. From here on, we concentrate on PSPC data. Given a PSPC image, the application of our screening algorithm proceeds according to the following steps:

1. we build the MV rate light curve and reject observation intervals with MV rate >250 cnt s^-1 to screen out residual particle events as described in Sect. 3.1;
2. we determine the background count collection region. This is accomplished computing the 20% and 80% percentiles of the count-rate image, and retaining in the collection region the pixels which fall within this range. With this choice, we discard putative sources and possible shadows (i.e. PSPC ribs);
3. we build the background light-curve of the field summing up all pixel counts in the image region defined in step 2, and scale it to the detection cell size at 30' off-axis angle; then, we sum up the background light-curve to obtain the background counts B;
4. using the Wavelet trasmorm algorithm of Damiani et al. (1997a), we compute the detection threshold of our field at 30' off-axis angle and use it as the source count-rate in the screening procedure; we then multiply this count-rate by the exposure time in order to get the source counts S.

At this stage we have all the information necessary to run the screening procedure. We define a variable that we call background moving threshold and set it initially to the maximum value of the background light-curve. Step by step, we decrease this variable by small amounts (say 1% of the initial value). At each step we evaluate x as the fraction of exposure time where the background count-rate is in excess of the moving threshold, and also evaluate B(x) subtracting from B the background counts in the above "excess" background intervals. Figure 4 (click here) gives a pictorial view of the quantities calculated at each step. Given x and B(x), the efficiency function E can be calculated. When this function reaches its maximum, the screening procedure is stopped and the time intervals to be screened out are identified.

  
Figure 4: In this simulated background light curve we put in evidence some of the relevant quantities involved in the screening procedure. The straight line is the background moving threshold. The shaded areas represent the intervals (and thus background counts) that are screened out at this stage of the procedure. B(x) is the integral of the light curve out of the shaded areas