• (Neil 6/11/06) Use a cadence of one sample every 15 seconds. This will handle all of the sources we have in the early challenges, and it will keep the data files down to a more reasonable size. (Need 15 rather than 30 second cadence to properly handle high frequency sources in galactic challenge sets)
• (Neil/Michele) Define a “year” as 2^21 * 15 = 31457280 s, which is about 1.16 days short; this has the advantage of setting the length of data sets to an exact power of 2, which simplifies FFTs

• The latest version (2.1) of LISA Simulator can go through a full XML pipeline for Galactic binaries. It should also be able to read and process the SMBH and EMRI xml files (à la Stas) that are produced by the Synthetic LISA pipeline. Michele is checking this (6/11/06).
• The latest version (1.3.0) of synthetic LISA can go through a full XML pipeline for Galactic binaries and SMBH binaries (à la Stas): read XML files for LISA and sources, and output XML TDI, with the LISA and source specification. See the Python scripts.
• This is the resulting lisaXML file for a single binary; the large linked binary file is here.
• Reading/writing noise is also completed (but need to document it better), while the GWDB background will need a little testing.
• Needed the source-parameter-generation codes!

The known verification binaries parameters were taken from LISA verification binary wiki that Gijs Nelemans maintains. The selected binaries were chosen strictly based on their computed SNRs; this was done using The LISA Calculator, and the results agree well with a study by Alexander Stoerr and Alberto Vecchio (astro-ph/0605227).

The inclination, polarization angle, and orbital phase have been randomly generated uniformly over the ranges: INC(0,180), PSI(0,360) and PHIo(0,360). These parameters are pure projection parameters related to the observer; there is no astrophysical history associated with their values, so a uniform distribution over the full range is suitable. For actual verification binaries (ones which have been electromagnetically observed prior to the LISA flight era), the values of these projection parameters may be subject to selection effects associated with the vagaries of electromagnetic detection.

The non-verification binaries in the list were drawn from a galactic population synthesis model by Matt Benacquista; his models are described in several papers (astro-ph/0308041; astro-ph/0504135). The particular galaxy realization used for this data comes from a set of galaxies Matt and I call the “BLT Galaxies” (Benacquista-Larson-Taylor), which are a set of galactic realizations with seven different sets of galactic shape parameters, three realizations each. The particular galaxy file used for these binaries is called bet_2.0_200_30MB.dat (filename included here for posterity, so we can dig it up if necessary), which contains a 30 million binary population synthesis galaxy.

The signals for these binaries will be generated using the built in binary signal generators from the simulators. As agreed, we will set all fdot terms == 0 (i.e., the sources for the challenges at this level are monochromatic).

For the second set of binaries (discussed on the 5 June 2006 telecon) with lower SNR, all above f ~ 2mHz, I simply drew from the same BLT galaxy as the one noted above and selected 20 binaries. The SNR selection was a complete hack, simply taking the ratio of the estimate spectral amplitude of the source to the standard sky-averged sensitivity + HB background from the Online Sensitivity Curve Generator. Curiously, you’ll note the SNRs computed post-facto using The LISA Calculator are significantly higher than the range of 1 to 10 I used in my simple hack selection. I find this curious and possibly disturbing; I’m uncertain to the internal workings of the Calculator, but if you look at the last several binaries in this list, they appear to have extrememly similar masses and distances, yet wildly varying SNR outputs from the LISA Calculator. It is possible the projection parameters are affecting this, but I’m surprised to see such a variation. In any event, the actual computed SNRs may be higher than what we desire for this set. If so, another set can be generated in the same fashion, but cut a different way to achieve the desired effect.