** Next:** Additional contributions to the
** Up:** GRASP Routines: Gravitational Radiation
** Previous:** Example: filters program
** Contents**

##

Practical Suggestion for Setting Up a Large Bank of Filters:

0
We have carefully explained (how to avoid)
a number of the pitfalls in computing post-Newtonian chirps.
Before using the chirp generators to
spit out hundreds or thousands of chirps needed for
a bank of filters and farming out the computations out to dozens
of parallel processors in a massive coalescing binary search,
we strongly suggest that you edit the examples already given
and check the routine against
the **three** extreme cases you will encounter in your search.

- Try the example with both masses set to the minimum mass in your
proposed search,
*i.e.* compute the phase and frequency evolution
and the chirps for the template in the upper right hand corner in
figure . This is the template of longest duration. If
you are going to have a memory allocation problem you will have it with
this template. Also, knowing the duration of the longest template
in your search will help you decide the length of the segments of
data which you filter. In general, you want the length of these data
segments to be at least several times longer than the longest chirp.
See Section for further details.
- Try the chirp generator with both masses set to the maximum mass in
your search,
*i.e.* compute the phase and frequency evolution
of the template in the lower left corner of figure .
This is the shortest duration template and the one least likely to make
it to the upper cut off frequency before going out of the region of
post-Newtonian viability. This case will be the most demanding test
of the ``chirp-termination'' logic in `phase_frequency()`. It is
also possible in the case of extremely large masses that there really
is no chirp at all in the frequency regime requested. For example a
binary composed of two 100 object will coalesce long before
it reaches the initial chirp frequency of the 60Hz we are using as our
a lower cutoff frequency in our example. Don't worry. The routine `phase_frequency()` will warn you that the root finder was unable to
find a viable solution for the initial time. You may have to adjust
the search range accordingly.
- Try the chirp generator with one mass at the minimum
allowed value and the other mass at the maximum allowed value,
*i.e.* compute the phase and frequency evolution for the template
in the upper left corner of figure .
This is the template which is
most dominated by post-Newtonian terms in the evolution.

If the routine gives satisfactory results for these three
cases, it should work for all the cases shown
in figure ; you are now ready for wholesale production.

** Next:** Additional contributions to the
** Up:** GRASP Routines: Gravitational Radiation
** Previous:** Example: filters program
** Contents**
Bruce Allen
2000-11-19