Cross-correlating EPIC source lists for either merging or for correlating earlier pipeline source lists is done in two stages. The first stage is similar for both the merging of source lists and cross-identifying sources with earlier pipeline source lists. In this stage (subroutine compare_srcs) correlations are investigated and pointers between matching sources are set. Always the closest companion (in terms of sigmas) is chosen. Only sources within 3 to 5 sigmas will be correlated (the exact number can be set using input parameter maxerr). In order to avoid problems due to boresight effects an input parameter systerr (for systematic error) has been introduced.
In the second stage, the cross identifications are either used to get the pointers to the source numbers from previous detection stages (e.g. the PN_ML_ID_SRC field), or the two lists are merged together. The first option works for the data sets provided with the crossidsets command line parameter, whereas merging is done if more than one set is provided using the inputlistsets parameter. Both options can be used simultaneously.
Merging means that new sources will be added to the list and parameters of cross-identified sources will be averaged using weights based upon the statistical errors. The maximum likelihood parameters will be all listed separately using columns with dimension larger or equal to one, in the same manner as the cross-identification columns. The reason is that maximum likelihood values cannot be simply averaged or added without losing its meaning as a probability related parameter.
Both for merging and for cross-identification, the source lists are read one by one and are correlated separately. This ensures that not more than 2 source lists will be stored in memory. Upon execution the number of sources that can be stored in the summary source list is three times the number of sources in the first input set. If during the execution this turns out not to be enough, the source list will be copied to a new array with a three times larger capacity.
The output EPIC observation source lists are provided both in FITS and in HTML format, where the HTML list is intended to give a quick overview for each source detected, but does not provide instrument-specific data.
As of version 3.10.2, the output FITS table contains the columns inst_b, inst_b_ERR, inst_EXP, with inst=[PN,M1,M2] and b=[1,2,3,4,5,TOT,XID]. These columns contain count rates, (inst_b), count rate errors (inst_b_ERR), and exposure map values (inst_n_EXP, n=[1..5]) from the individual instruments and up to 5 energy bands. In the case of several source lists from one instrument, count rates are averaged and exposure map values are added.
As of version 3.10.5, the strings from the input columns VAL_FLAG, VAL_FLAG, VER_COMM are copied into the columns of the output file inst_VAL_FLAG, inst_VAL_FLAG, inst_VER_COMM (inst=[PN,M1,M2]).
Fluxes and flux errors from the input source lists are copied to the columns inst_b_FLUX and ERR_inst_b_FLUX (inst=[PN,M1,M2], b=[1,2,3,4,5,TOT,XID]). The columns EP_b_FLUX and ERR_EP_b_FLUX contain all-EPIC flux values and errors, calculated from the available single camera values using error-weighted averages.
The hardness ratios and errors from the input source lists are copied to
output columns inst_HRn and inst_HRn_ERR (inst=[PN,M1,M2], n=[1,2,3,4]). In addition, weighted
averages of the single camera hardness ratios are written to
columns EP_HRn, EP_HRn_ERR.
Note: Since the EPIC cameras have different energy dependent effective areas, the averaged HR values depend on the relative exposures and/or presence of the (sources in the) individual cameras.