Re: LegendreP -- up to which order?

*To*: mathgroup at smc.vnet.net*Subject*: [mg53591] Re: LegendreP -- up to which order?*From*: "A. E. Siegman" <siegman at stanford.edu>*Date*: Wed, 19 Jan 2005 02:00:02 -0500 (EST)*Organization*: Stanford University*References*: <cscicl$5lp$1@smc.vnet.net> <csiou4$np4$1@smc.vnet.net>*Sender*: owner-wri-mathgroup at wolfram.com

I did a lengthy calculation once involving Legendre and other polynomials of high order (up in the 60s) and got remarkably accurate results (IMHO, anyway). However I discovered that it made an observable difference at high enough orders whether the arguments were integers or reals. If the arguments were reals I could see observable differences between the results from two calculations that should theoretically have given identical results; but if I entered an argument of 1.001 Pi in the form 1001 Pi/1000, for example, the results of the two alternative calculations would agree exactly.