The orthogonality of Bessel function for finite range is
where is the n-th root of the Bessel function . The above Bessel function is scaled to length with n-th node. For example,
Simply the integration by setting
The relation between Riccati-Bessel function, spherical Bessel function, and Bessel function are:
The orthogonal condition of the spherical Bessel function should be
From the Bessel Function orthogonality,
In term of Riccati-Bessel function, ,
Using the orthogonality, the spherical Bessel function can be basis that span the whole space with . We can also normalized the spherical Bessel function. For Riccati-Bessel function, it span the whole space with . Finite-range function expands as Bessel function is called Fourier-Bessel series.
The infinite range orthogonality of Bessel function J is (according to wiki and many reference, why?)
in term of spherical Bessel function
and using , in term of Riccati-Bessel function is
For example, the spherical Bessel J of order zero,
,
Let’s integrate it
which is infinite, and the right-hand side is a delta function .
The problem with the above formulae is, that the orthogonal condition is OK. but it should be also normal, i.e. orthonormal, to be a normalizable basis.
The Hankel Transform is
and this can change to spherical-Bessel function as,
The foundation for the transformation is the orthogonality of the Bessel function. Substitute as Bessel function.
So, what do I miss?
For example, is oscillating around zero, the square of it is always positive, and with a weighting , it approach to with a phase shift,
And the integrate of is . Thus, the integral of Bessel function with weight is an infinite sum for between each zeros, and each terms is ~1. How this infinite sum becomes 1?
Recall the plane wave expansion,
As we show it before, it is a special case of the Hankel transform, or belong the general Fourier Transform.