Associated Legendre Function Problem in mma?
- To: mathgroup at smc.vnet.net
- Subject: [mg4361] Associated Legendre Function Problem in mma?
- From: siegman at ee.stanford.edu (A. E. Siegman)
- Date: Mon, 15 Jul 1996 05:01:00 -0400
- Organization: Stanford University
- Sender: owner-wri-mathgroup at wolfram.com
Associated Legendre functions are a real bear. Trying to
cope with them, I note in Abramowitz and Stegun, p. 334,
Eq. (8.6.16), that one of these functions which I
particularly need to use, LegendreP[n,-n,x], has the special
case (in TeX notation):
P_n^{(-n}(x) = 2^{-n} (x^2-1)^{n/2} / \Gamma[n+1]
But when I try to confirm this with mma, I see that the
magnitudes are OK, but there is a residual confusion
about phase angles:
specialCase[n_,x_] := 2^(-n) (x^2-1)^(n/2) /
Gamma[n+1]
Table[{n, LegendreP[n,-n,x] /
specialCase[n,x] // Simplify},
{n,0,5}] // TableForm
0 1
2
Sqrt[1 - x ]
-------------
2
1 Sqrt[-1 + x ]
2 -1
2
Sqrt[1 - x ]
-(-------------)
2
3 Sqrt[-1 + x ]
4 1
2
Sqrt[1 - x ]
-------------
2
5 Sqrt[-1 + x ]
and unfortunately getting the phase angles right is
important in my problem. Who's correct here?
Addendum: The reason for worrying about this is that
I want to evaluate very high order polynomicals (n > 50)
using rational fraction values of x for accuracy (which
seem to work pretty well). But while LegendreP[2n, x],
which I also need to use, seems to run fine in this way,
LegendreP[n,-n,x] slows to a crawl for n > 20 or thereabouts
-- even though the polynomial expressions for the regular
and associated Legendre's are of the same order in the
two cases. Hence the search for an alternative for the
associated case.
--AES siegman at ee.stanford.edu
==== [MESSAGE SEPARATOR] ====