go to next pagego upgo to previous page

4.5 Low frequency expansion of the hypergeometric expansion

So far, we have considered exact solutions of the Teukolsky equation. Now, let us consider their low frequency approximations and determine the value of n. We solve Equation (136View Equation) with n = 0,
n n n b0 + a 0R1 + g0L -1 = 0, (171)
with a requirement that n --> l as e --> 0.

To solve Equation (171View Equation), we first note the following. Unless the value of n is such that the denominator in the expression of n an or n gn happens to vanish, or n b n happens to vanish in the limit e --> 0, we have ann = O(e), gnn = O(e), and bnn = O(1). Also, from the asymptotic behavior of the minimal solution f n n as n --> ± oo given by Equation (134View Equation), we have Rn(n) = O(e) and L-n(n) = O(e) for sufficiently large n. Thus, except for exceptional cases mentioned above, the order of an n in e increases as |n | increases. That is, the series solution naturally gives the post-Minkowski expansion.

First, let us consider the case of Rn(n) for n > 0. It is easily seen that ann = O(e), gnn = O(e), and bn = O(1) n for all n > 0. Therefore, we have R (n) = O(e) n for all n > 0.

On the other hand, for n < 0, the order of L -n(n) behaves irregularly for certain values of n. For the moment, let us assume that L -1(n) = O(e). We see from Equations (124View Equation) that n a 0 = O(e), gn0 = O(e), since n = l + O(e). Then, Equation (171View Equation) implies bn0 = O(e2). Using the expansion of c given by Equation (110View Equation), we then find n = l + O(e2) (i.e., there is no term of O(e) in n). With this estimate of n, we see from Equation (128View Equation) that L (n) = O(e) -1 is justified if L (n) -2 is of order unity or smaller.

The general behavior of the order of L- n(n) in e for general values of s is rather complicated. However, if we assume s to be a non-integer and l > |s|, and t = (e - mq)/k = O(1), it is relatively easily studied. With the assumption that n = l + O(e2), we find there are three exceptional cases:

These imply that L-2l-1(n) = O(1/e), L-l-1(n) = O(1), and 2 L -l(n) = O(e ), respectively. To summarize, we have

fn Rn(n) = -nn--= O(e) for all n &gt; 0, fn-1 fn-l L -l(n) = -n----= O(e2), f-l+1 fn-l-1 L -l-1(n) = --n---= O(1), (172) f-l fn-2l-1 L -2l-1(n) = --n--- = O(1/e), f-2l -fnn-- Ln(n) = fn = O(e) for all the other n &lt; 0. n+1
With these results, we can calculate the value of n to O(e), which is given by
( ) 1 s2 [(l + 1)2- s2]2 (l2 - s2)2 2 3 n = l + ------ -2 - --------+ ------------------------- ------------------ e + O(e ).(173) 2l + 1 l(l + 1) (2l + 1)(2l + 2)(2l + 3) (2l - 1)2l(2l + 1)
Now one can take the limit of an integer value of s. In particular, the above holds also for s = 0. Interestingly, n is found to be independent of the azimuthal eigenvalue m to O(e2).

The post-Minkowski expansion of homogeneous Teukolsky functions can be obtained with arbitrary accuracy by solving Equation (123View Equation) to a desired order, and by summing up the terms to a sufficiently large |n|. The first few terms of the coefficients fnn are explicitly given in [33Jump To The Next Citation Point]. A calculation up to a much higher order in O(e) was performed in [55Jump To The Next Citation Point], in which the black hole absorption of gravitational waves was calculated to O(v8) beyond the lowest order.



go to next pagego upgo to previous page