radiative transfer in plane-parallel media and cauchy integral

analytic in the right complex half-plane, we must have in particular f(+1/k) < +∞, where k = k(a) is the unique root in ]0,1[ of the transcendental equation T(a,1/z) ...
149KB taille 8 téléchargements 267 vues
TRANSPORT THEORY AND STATISTICAL PHYSICS, ( ), 1–18 ( )

RADIATIVE TRANSFER IN PLANE-PARALLEL MEDIA AND CAUCHY INTEGRAL EQUATIONS II. THE H-FUNCTION B. Rutily, J. Bergeat, L. Chevallier Centre de Recherche Astronomique de Lyon (UMR 5574 du CNRS), Observatoire de Lyon, 9, Avenue Charles Andr´e, 69561 Saint-Genis-Laval Cedex, France

Abstract In the central part of this paper, we revisit the classical study of the H-function defined as the unique solution, regular in the right complex half-plane, of a Cauchy integral equation. We take advantage of our work on the N -function published in the first article of this series. The H-function is then used to solve a class of Cauchy integral equations occurring in transfer problems posed in plane-parallel media. We obtain a concise expression of the unique solution analytic in the right complex half-plane, then modified with the help of the residue theorem for numerical calculations. Key Words: Radiative transfer equation; Plane-parallel geometry; Isotropic scattering; Cauchy 1 Copyright

C

2000 by Marcel Dekker, Inc.

www.dekker.com

2

RUTILY, BERGEAT AND CHEVALLIER

integral equations

I.

INTRODUCTION

This article is the second in a series dedicated to the Cauchy integral equations arising in radiative transfer problems in plane-parallel geometry. We adopted in Ref. [1] (hereafter I) the simplest scattering law–monochromatic and isotropic–with volumic albedo 0 < a < 1 and assumed that light propagates through a homogeneous and stationnary slab of optical thickness 0 < b ≤ +∞. The resulting integral equations take the form Z 1 a dv T (a, z)f (z) − z f (v) = c(z), (1) 2 0 v−z where T (a, z) is the dispersion function defined for any z ∈ C \ {±1} by Z +1 dv a . (2) T (a, z) = 1 + z 2 −1 v − z The integrals in Eqs. (1) and (2) are Cauchy principal values (symbol f ) for z ∈]0, 1[ and z ∈] − 1, +1[ respectively. Thus, Eq. (1) is both a Cauchy integral equation over [0, 1] and the definition of the extension of f outside this interval. Given a function defined by a Cauchy-type integral, the same symbol is used for the values of the function outside the cut and on the cut. This writing convention allows to extend, on the cuts, the relations satisfied by those functions which include Cauchy-type integrals in their definition. For instance the definition (2) of the T -function becomes, for any z = u ∈] − 1, +1[ a +1 dv a 1+u T (a, u) = 1 + u f = 1 − u ln[ ], 2 −1 v − u 2 1−u

(3)

where u denotes the z-variable while exploring the real axis. The right-hand side c(z) of Eq. (1) is a given function satisfying the following properties:

CAUCHY INTEGRAL EQUATIONS. II

3

(i) it is defined in C except possibly at z = 0 and, for finite media, at z = −1, (ii) it is analytic in the half-plane 0). The surface value of the B-function was calculated for the first time by Ambartsumian, who found B(a, 0, u) = H(a, u) .[18] The solution within the medium is[19] R(a, k) ku B(a, τ, u) = exp(−kτ )H(a, u) H(a, +1/k) 1 − ku + (gT )(a, u) exp(−τ /u) a 1 dv + H(a, u) uf (g/H)(a, v) exp(−τ /v) . (46) 2 0 v−u As expected, the H-function is retrieved on the τ = 0 layer from

CAUCHY INTEGRAL EQUATIONS. II

13

this expression and Eq. (38), which illustrates the role played by the latter equation in this particular case.

IV.

BACK TO THE GENERAL CASE: FOR ANY c

Obviously, the H-function is the proper auxiliary function to write the unique solution to Eq. (1) analytic in the right complex halfplane. We are then tempted to express this latter solution, as given by Eqs. (9)-(13), in terms of the H-function with the help of relations (19)-(23). Actually, it is possible to solve Eq. (1) following a more direct approach, leading to its solution as a unique functional of c(z) and H(a, z). This relation is valid in the whole complex plane, and it once again yields relations (9)-(13) (expressed in terms of H) when the z-variable covers their validity domain. It is established hereafter from Eq. (1) through a residue calculation, taking into account the properties (i)-(iii) in Sec. I of the free term c(z). Replace z 6= 0 by −1/z 0 in Eq. (1), multiply the resulting equation by (1/2iπ)H(a, 1/z 0 )/(1 + zz 0 ) and integrate with respect to z 0 along the imaginary axis in the sense of the principal value at infinity, i.e., R +i∞ R +iX by writing −i∞ = limX→+∞ −iX . We obtain Z +i∞ 1 dz 0 (α) 0 0 0 H(a, 1/z )T (a, 1/z )f (−1/z ) 2iπ −i∞ 1 + zz 0 Z +i∞ Z 1 dz 0 (β) 1 a dv + H(a, 1/z 0 ) f (v) 2iπ −i∞ 2 0 1 + vz 0 1 + zz 0 Z +i∞ 1 dz 0 = . (47) H(a, 1/z 0 )c(−1/z 0 ) 2iπ −i∞ 1 + zz 0 The (α)-term on the left-hand side can be calculated by replacing H(a, 1/z 0 )T (a, 1/z 0 ) by 1/H(a, −1/z 0 ), observing that the function z 0 → f (−1/z 0 )/H(a, −1/z 0 ) is analytic in the left complex half-plane. Applying the theorem of residues to an obvious contour in the left complex half-plane, we find for all z ∈ C∗ (α) =

1 f (z) 1 {Y[ 0]. Owing to this extension, (48) is still valid on the imaginary axis without zero (denoted iR∗ ), the integral in the (α)-term of Eq. (47) becoming a Cauchy principal value. The (β)-term on the left-hand side of Eq. (47) is calculated by inverting the order of integration, writing 1 1 1 1 1 =[ 0 − ] . 1 + vz 0 1 + zz 0 z + 1/v z 0 + 1/z v − z It follows from the residue theorem, applied to a contour in the right complex half-plane, that Z +i∞ dz 0 1 1 H(a, 1/z 0 ) = Y[−