SCORPIUS X-1 S2 ANALYSIS

Data Analysis Pipeline

The analysis follows the following steps:

1. Produce the ephemeris for Sco X-1 and define search area in parameter space

2. Identify the data segment to be analysed based on the optimal SNR that can be achieved. This is done through a Monte Carlo where Sco X-1 like signals are injected in the data set and detected using the analysis pipeline
   • The codes that are used are:
     • makefakedata.c to perform software signal injection
     • ComputeFStatisticBinary.c which is the search code used for the computation of the F-statistic (the same one used in the actual search)
   • There are also scripts to run the Monte Carlo and combine the output of it in order to produce the figure of merit on which the decision is based

3. Generate the metric and lay the filters in parameter space (relevant to the segment to be analysed) for the different IFOs. At this stage we also identify areas in parameter space - X, Y (orbital parameters) and f (frequency) - that are considered ``in coincidence''
   • The code that performs this is LayMesh.c, which takes as input values that are derived from the metric computation
   • The check for the effectiveness of the filter bank(s) - i.e. given a signal injected in the parameter space - is performed with Monte Carlo's that use the signal injection code makefakedata.c and the F-statistic code ComputeFStatisticBinary.c

4. Run the search, i.e. compute the F-statistic on the relevant parameter space area (frequency and orbital filters) for each detector
   • The code that performs this is ComputeFStatisticBinary.c
   • The key building blocks of the code are LALDemod.c and ComputeSkyBinary.c
   • A script is used to split up the search area in a set of subareas and allocate the jobs to different nodes

5. Compare the lists of candidates from L1 and H1 and produce the ``unvetoed'' list of coincidence events
   • A script is used
   • Input to this is the results obtained at point 2

6. Veto-out coincidence signals that do not pass the chi2-test
   • The codes used are cluster.c and FstatShapeTest.c (same as for isolated pulsars)
   • Monte Carlo's are run before the analysis in order to set the threshold for the chi2-cut. Codes already mentioned are used for this Monte Carlo's.

7. Produce the upper-limit
   • No new codes are used here
   • Scripts do the job
   
   

Proposed Schedule for Review

Friday 17th September at 17:15 BST

Overall structure of the search: walk through structure and components of ComputeFStatisticBinary.c (search code)
Link to lalapps_ComputeFStatisticBinary walkthrough 17th Sep 2004

Thursday 27th September (isolated pulsars review)

Partecipate to review of cluster.c

Wednesday 29th September

Review of ComputeSkyBinary.c (LALDemod.c, which is also used in search for binary systems, is reviewed on September 16th)

Wednesday 6th October (isolated pulsar review)

Participate to review of chi2 test

Week 4th - 9th October

Review of parameter space area:

Ephemeris
Metric and filter placement (and therefore coincident events)

Wednesday 20th October (isolated pulsar review)

Participate to review of makefakedata.c (there is only one code that does both binaries and isolated)

Wednesday 27th October

Review of scripts and Monte Carlo's used to produce upper-limits