...$s$^10.1

Note that when $s=0$ the values computed are coefficients of the intrinsic frequency and thus must be computed for the value corresponding to the index $b$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... T010095^16.1

Note that we write insted of $S_k$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... plane.^22.1

These form a right-handed basis, providing an answer to the age-old question ``What's Up?'': ``East cross North.''

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... is^27.1

It is important to make the limits of integration symmetric about $f_k$ to maintain reality conditions when dealing with Fourier transforms of real quantities.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... as^35.1

Note that the $P$th frequency bin is not treated specially, as would be expected for the Nyquist frequency. This is the appropriate behavior if $M$ is an odd number (so that there is no Nyquist bin) or if, as a result of coarse-graining, the Nyquist bin has been removed from $\widetilde{Q}$. At any rate, if there's a significant contribution to the cross-correlation statistic from the Nyquist frequency, something is wrong.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... series^35.2

The values are compared to a series generated by Matlab, correcting for the differing sign convention.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... series^35.3

The values are compared to a series generated by Matlab, correcting for the differing sign convention.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... order^38.1

We use units in which $c=G=1;$ thus $1M_\odot=5\times 10^{-6}$ s.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (or^38.2

In what follows we will only mention $\tau_3$; however, the algorithm is itself valid, and has been implemented, in the case of $(\tau_0, \tau_2)$ too.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (or^38.3

In what follows we will only mention $\tau_3$; however, the algorithm is itself valid, and has been implemented, in the case of $(\tau_0, \tau_2)$ too. However, we recommend that the space $\tau_0$-$\tau_3$ be used.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.