derivatives of bessel_j A&S 9.1.27 vs 9.1.30



maxima-bounces at math.utexas.edu wrote on 01/24/2007 02:54:42 PM:

> >>>>> "Robert" == Robert Dodier <robert.dodier at gmail.com> writes:
>
>     Robert> On 1/24/07, Barton Willis <willisb at unk.edu> wrote:
>     >> Would it be OK to change the derivatives of the Bessel functions
>     >> to use A&S 9.1.27 instead of 9.1.30?
>
>     Robert> OK by me.
>
> Is this related to the bug Barton reported about taylor not being able
> to expand bessel functions?

Yes.

> If so, we should also tell limit about bessel functions.  (I think that
was the issue.)

Sure, telling limit about bessel is a good thing, but I don't think that
will fix the taylor(bessel_j(0,x),x,0,4) bug. I think taylor evaluates
by substitution, not via limit. I tested the change from A&S 9.1.27 to
9.1.30:

(%i22) load("bessel.lisp")$
(%i23) taylor(bessel_j(0,x),x,0,5);
(%o23) 1-x^2/4+x^4/64+...

(%i24) taylor(bessel_j(5,x),x,0,5);
(%o24) x^5/3840+...

(%i25) taylor(bessel_j(5,x),x,1,5);
(%o25) bessel_j(5,1)+((bessel_j(4,1)-bessel_j(6,1))*(x-1))/2 + junk

That's OK. But the same change for beseel_y is a problem:

(%i28) taylor(bessel_y(5,x),x,0,1);
(%o28) bessel_y(5,0)+((bessel_y(4,0)-bessel_y(6,0))*x)/2+...

I'm not sure how to handle that... yeechs:

(%i32) bessel_y(0,0);
(%o32) bessel_y(0,0)   <--- could be an error or

> But otherwise, I have no problems with changing the derivatives.
> Perhaps the derivative in 9.1.30 makes it easier for maxima to deduce
> that the bessel functions satisfy Bessel's differential equation?

No, I don't think it makes it easier, just different.

Barton