Proposal for naming of Bessel functions



Raymond Toy wrote:

> So, both bessel_j(a,x) and bessel_j[a](x) are recognized and produce
> the same results?

Correct. I can't find any differences between the way Macsyma
evaluates bessel_j(a,x) and bessel_j[a](x).  They both even TeX to
$$ J_{a}\left(x\right) $$.

Further, the Macsyma documentation says that the names jo(x), j1(x),
and jn(x,n) "are still supported, but their use is discouraged."

>Using this syntax, can I take the derivative wrt to a still?  This is
>a well-defined operation.

Macsyma (circa 2.2) doesn't know the derivatives of the
Bessel functions with respect to the order.  I don't think
it is possible to use gradef to define the derivative of
bessel_j[a](x) with respect to both a and x.  Maybe this
matters, maybe it doesn't to you; Macsyma doesn't appear
to use gradef to differentiate the Bessel functions.

One more thing: I'm guessing that some versions of Macsyma
used Miller's algorithm for numerical evaluation of the Bessel
functions (see A&S).  After evaluating J_n(x), for a float x,
an array bessel_j_array held the values of  J_0(x), J_1(x), ...
J_n(x).  I wasn't able to test this;  my Macsyma doesn't
seem to do this.

Barton