July 15, 2004:

These files were updated on July 15, 2004 at 2pm EDT. The earlier files were 
replaced by these with the same names.

Neil Cornish writes:

(1) The tutorial on the website is correct. The LISA Simulator produces fourier
components that have been normalized so that they are independent of the number of
data points in the time series. This means that you can go ahead and add together
signals of different cadence in the frequency domain exactly as described in the
tutorial I wrote.

(2) The galactic backgrounds that we produced are 100% correct in the frequency
domain. I spent quite a bit of time in the last few days making sure of that. But
the time domain signals that we posted were not properly normalized. They are out
by a factor of N/sqrt(T). Since we used N=2^23 and T = 1 year, this gives a factor
of 1493.26. We never noticed this problem as we do all our analysis on the
backgrounds in the frequency domain. The offending code has been fixed and the
re-scaled backgrounds are now avaliable for download at
http://www.physics.montana.edu/faculty/cornish/files/ Don't expect our signals to
look just like your Michelson signals. We output the TDI variables X(t), Y(t), Z(t).
These differ quite significantly from Michelson signals. In the frequency domain the
X-type signals differ from their Michelson counterparts by a factor of |2
sin(f/2f_*)|, where f_* ~ 0.01 Hz is the transfer frequency.