| J. Phys. A: Math. Gen. 39 No 18 (5 May 2006) L293-L296 |
| doi:10.1088/0305-4470/39/18/L06 |
| PII: S0305-4470(06)18839-3 |
Analytical results for a Bessel function times Legendre polynomials class integrals
A A R Neves L A Padilha, A Fontes, E Rodriguez C H B Cruz L C Barbosa and C L Cesar
Instituto de Física Gleb Wataghin, Universidade Estadual de Campinas, PO Box 6165, 13083-970 Campinas, Brazil
Email: aneves@ifi.unicamp.br
Received 13 February 2006
Published 19 April 2006
| Abstract. When treating problems of vector diffraction in electromagnetic theory, the evaluation of the integral involving Bessel and associated Legendre functions is necessary. Here we present the analytical result for this integral that will make numerical quadrature techniques or localized approximations unnecessary. The solution is presented using the properties of the Bessel and associated Legendre functions. |
PACS numbers: 41.20. - q, 42.25.Fx, 42.60.Jf
1. Introduction
In a recent work to calculate the optical force of the optical tweezers in a complete electromagnetic treatment for any beam shape focused at an arbitrary position, we encountered an integral involving Bessel and associated Legendre functions [1]. The same integral appears in fields related to vector diffraction theory where computationally intensive methods or approximations are employed [2]. This letter presents the analytical evaluation of this integral (1).
Typically, the first task in solving scattering problems is to decompose the incident beam into partial waves involving associated Legendre polynomials Pmn(cos θ) for the angular part and spherical Bessel functions jn(kr) for the radial part. The difficulty in doing this exactly is to determine an analytical expression for integrals of the class
which, so far, have not been reported in a closed form, as far as we know. This integral is not shown in any Integral Tables, nor in calculation packages such as Mathematica, and we do not know of any other report of this result. An analytical expression for this integral would be useful for many different fields in physics, especially for those that require partial wave decomposition. Without this result people have used all sorts of approximations to proceed forward and obtain results [3].
2. General analysis
We analysed the behaviour of such integrals for the limit of kr = R → 0 using Lock's result for the integral [4]
and found that the lowest order term would be compatible with the closed form and simple expression given by
valid for any n ≥ 0 and - n ≤ m ≤ n. We then tested it numerically using Mathematica software (version 5.2) for randomly generated n, m, α and R and showed that this result is indeed true. It would also allow one to access a whole family of integrals by taking any number of derivatives with respect to R or α, such as
After discovering the expression for this integral, we proceed to prove it using the same induction procedure used in the Lock integral paper. We started by proving that Imn follows the same recurrence relations as Fmn = 2in - mPmn(cos α)jn(R). We then proved that the results hold for n = m = 0.
The recurrence relations for Imn can be obtained by the associated Legendre polynomials recurrence relations
where we adopted the sign convention for the associated Legendre polynomials of Abramowitz and Stegun [5], followed by the Mathematica software. The Bessel functions recurrence relation
Using relations (5) and (6), it can be readily shown that
is the desired recurrence relation for Imn.
Now, for Fmn = 2in - mPmn(cos α)jn(R) we used the associated Legendre polynomials recurrence relation
and the spherical Bessel functions relation
to prove that
This assures that both sides of the identity follow the same recurrence relation. Therefore, the only task remaining is to prove that
holds for n = m = 0. Now, the series expression for J0 is
therefore,
The Poisson integral representation of the spherical Bessel function is
which, added to the parity null term
can be rewritten as
This leaves the initial integral as
That is still not good because the argument of the spherical Bessel function is Rcos α instead of R. To change the argument, we use the multiplication theorem for Bessel functions
To obtain
we can now make ν = 1/2, n = 0, z = Rcos α and λ = 1/cos α. Therefore,
and finally
Comparing this result with the obtained series, it turns out that
and the proof is complete, validating the following integral for all n and m:
3. Concluding remarks
In this letter, we have shown the analytical solution to an integral involving Bessel and associated Legendre functions (23). We have obtained the solution through recurrence relations and the multiplication theorem. We believe that this result, analytical and simple, is of interest to the general community especially for problems involving electromagnetic vector diffraction of arbitrary beams.
Acknowledgments
This work was partially supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) through the Optics and Photonics Research Center (CePOF). We thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for financial support of this research.
ReferencesA A R Neves et al 2006 J. Phys. A: Math. Gen. 39 L293
D. Fong et al. 2006 ApJ 652 1626
C. D. Fassnacht et al. 2008 ApJ 681 1017
David Ehrenreich et al 2007 ApJ 668 L179
Daniel Kocevski et al. 2007 ApJ 663 1180
Jonathan Langton and Gregory Laughlin 2007 ApJ 657 L113
J. A. Eisner et al. 2007 ApJ 669 1072
Mitsuhiko Honda et al. 2006 ApJ 646 1024
Kirsten von Bergmann et al 2007 New J. Phys. 9 396
H. Lammer et al 2003 ApJ 598 L121