Dear ASIS members,                                 5 May
1999

as I shall not be able to participate in the upcoming
conference-call ASIS meeting ( 6 May 99) , I wish to
bring up some points (which, I hope, will be further
discussed by Sathyaprakash and Bala Iyer during the telecon).

The minutes of the last ASIS meeting contain the following
statements: "They [ = Droz, Knapp, Poisson and Owen] showed
that the stationary phase approximation (SPA) is more accurate
than had been thought. ... The loss in SNR is acceptably small."

Me and my collaborators Sathyaprakash and B. Iyer were surprised
by this statement because we were the first ones (in our paper
"Improved filters for GW from inspiralling binaries",PRD57(1998)885)
to warn, after some careful analysis and thought, that the SPA was 
not accurate enough for defining Post-Newtonian filters for 
MASSIVE binary systems. 

Indeed, in our DIS paper we showed that the overlaps, for 10 M_sun , 
10 M_sun systems, between the SPA approximate PN waveform and its 
corresponding exact FFT transform, ranged between 0.967 (for the Newtonian
 waveform) and 0.918 (for a fiducial "exact" test-mass PN waveform). 
We have recently CONFIRMED these results. 
We find that the results of Droz et al are mathematically correct,
but are :(i) too limited (only the Newtonian waveform is considered),
and (ii) physically misleading (the overlaps > 0.997 they obtain
are mathematically correct but physically irrelevant). 
Our recent work shows again that the construction of accurate
templates is a delicate matter. We are not aware, in the literature,
of any work constructing templates of accuracy comparable to the
P-approximants we have proposed, and studied in detail, in DIS.

Let me reiterate: P-approximants are excellent, and (until they are 
superseded) necessary tools for defining template waveforms.  We are now
 extending the results of DIS to study in more detail the performance
of SPA-waveforms.


 Thibault Damour