Radiation Physics and Chemistry 59 (2000) 281±285
www.elsevier.com/locate/radphyschem
Technical note
On a generalized secant integral L.A.-M. Hanna, S.L. Kalla* Department of Mathematics and Computer Science, Faculty of Science, Kuwait University, P.O. Box 5969, Safat 13060, Kuwait Received 18 May 1999; accepted 2 February 2000
Abstract In this paper, we introduce a generalization of the secant integral in the following form
c Ia
c, b, l ba e ÿ bsec j
sec j a
tan j 2lÿ1 dj 0
with ar0, b > 0, 0 < cR p2 , and l > 0: This new generalization of the secant integral might have some applications in radiation ®eld problems of dierent source-shield con®gurations. We have obtained two series representations for Ia
c, b, l in terms of incomplete gamma function. The complete generalized secant integral is expressed in terms of generalized hypergeometric functions. Some recurrence relations are given. 7 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction
integral, de®ned as
For an isotropic line source of uniform strength s with a slab shield, the detector response at a particular point P is given as (Michieli, 1998)
Ia
c, b ba
(
R
3 X rs I0
c, mt Ai
1 ÿ b ÿaÿi Iai c, mt
1 4ph i0
ÿ b
)
1
where r is the appropriate ¯ux to detector response conversion factor, and Ia
c, b is the generalized secant
* Corresponding author. Tel.: +965-532-7901; fax: +965481-7201. E-mail addresses:
[email protected] (L.A.-M. Hanna),
[email protected] (S.L. Kalla).
c 0
e ÿb secj
sec j a dj
2
with ar0, b > 0, and 0 < cR p2 : This generalized secant integral has been considered recently by Michieli (1998), and he has obtained some series representations. Hungerford (1962) has considered similar integrals, but only for integral values of a. For a 0, Eq. (2) is known as Sievert integral. For a 0 and c p2 , Eq. (2) reduces to a special case of Bickley function (Abramowitz and Stegun, 1972; Sievert, 1921, 1930; Wood, 1982). In this paper, we introduce another generalization of the secant integral in the following form Ia
c, b, l ba
c 0
e ÿb sec j
sec j a
tan j 2lÿ1 dj
with ar0, b > 0, 0 < cR p2 , and l > 0:
0969-806X/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 9 - 8 0 6 X ( 0 0 ) 0 0 2 6 4 - 4
3
282
L.A.-M. Hanna, S.L. Kalla / Radiation Physics and Chemistry 59 (2000) 281±285
For l 12 , Eq. (3) reduces to Eq. (2), that is Ia
c, Ia
c, b: This new generalization of the secant integral might have some applications in radiation ®eld problems of dierent source-shield con®gurations. We have obtained two series representations for Ia
c, b, l in terms of the incomplete gamma function. The complete generalized secant integral is expressed in terms of generalized hypergeometric functions. Some recurrence relations are given. b, 12
On expanding
x 2 ÿ 1lÿ1 in powers of Eq. (6), by using Eqs. (4) and (7)
1 x
we get from
ÿ I^a ce , c, b, l ba
sec c sec ce
e ÿbx x aÿ1
2iÿ2l2 1 X 1 ki dx, x i0
8
p rc > ce > 0, 2 which after integration leads to ÿ I^a ce , c, b, l
2. Generalized secant integral Using the substitution sec j x, Eq. (3) can be written as Ia
c, b, l ba
sec c 1
1 X ÿ ki b 2iÿ2l2 G a ÿ 2i 2l ÿ 2, b sec ce i0
e ÿbx x aÿ1
x 2 ÿ 1
lÿ1
dx:
1 X ki
cos c 2iÿ2l2 E2iÿ2l3
b sec c
ÿ G
a ÿ 2i 2l ÿ 2, b sec c :
4
For a 0 and l 12 , Eq. (4) becomes the Sievert integral, I0
c, b, 12 I0
c, b: Using exponential integrals, 1 the En
y ynÿ1 y e ÿt t ÿn dt, which satisfy the x relation En
y ynÿ1 G
1 ÿ n, y, where g
a, x 0 e ÿt taÿ1 dt, 1 and G
a, x x e ÿt taÿ1 dt, a > 0 are the incomplete and the complementary incomplete gamma functions, satisfying g
a, x G
a, x G
a: Expanding
x 2 ÿ 1lÿ1 , Eq. (4) can be written for a 0 as ÿ I0
c, b, l I0 p2 , b, l ÿ
5
i0
lÿ1 i where ki
ÿ1i
1ÿl and I0
p2 , b, 12 I0
p2 , i! i b is the Bickley function,
an G
a n=G
a a
a 1
a n ÿ 1: The disadvantage of such a representation is the slow convergence of the expansion for small values of c (Michieli, 1998). For simplicity, we introduce the auxiliary de®nition ÿ ÿ ÿ I^a c1 , c2 , b, l Ia c2 , b, l ÿ Ia c1 , b, l ,
For a 0, the use of exponential integrals leads to ÿ I^0 ce , c, b, l
nÿ 1 X 2iÿ2l2 ÿ ki cos ce E2iÿ2l3 b sec ce i0
o ÿ
cos c 2iÿ2l2 E2iÿ2l3
b sec c :
x aÿ1
x 2 ÿ 1
lÿ1
1 X i pa, i
x ÿ 1 ,
where, i 1 X j a ÿ 1
1 ÿ l j
ÿ 1 iÿj 21ÿl j0
2j!!
12a
i
1 ÿ l iÿj 1 X iÿj a ÿ 1
: ÿ 1 1ÿl j 2 2
i ÿ j !! j0
12b
pa, i
The generalized secant integral from Eq. (3) can be presented as ÿ ÿ Ia
c, b, l Ia ce , b, l I^a ce , c, b, l ,
Hence, from Eq. (7) we obtain an expression
where from Eq. (4) as ÿ Ia ce , b, l ba
sec ce 1
ÿ Ia ce , b, l ba
sec ce 1
1 X pa, i
x 1ÿl
e ÿbx
x ÿ 1
i
ÿ 1 dx, e
ÿbx
x
aÿ1
x ÿ 1 2
lÿ1
dx:
7
11
i0
6
10
Let a 0, c p2 and ce c, then from Eq. (6) we get, I^0
c, p2 , b, l I0
p2 , b, l ÿ I0
c, b, l: So, Eq. (9) reduces to the Sievert integral, as given in Eq. (5). Consider Taylor's expansion about x 1;
c2 > c1 > 0:
c > ce > 0,
9
which after integration leads to
i0
13
L.A.-M. Hanna, S.L. Kalla / Radiation Physics and Chemistry 59 (2000) 281±285 1 X ÿ ÿ Ia ce , b, l e ÿb pa, i baÿlÿi g i l, b sec ce
jp1, iÿj j <
i0
ÿ1 :
We now consider the following cases for dierent values of a: (i) For a 0 (Sievert integral) From Eq. (12a), we have p0, i
i
1 ÿ l j
ÿ 1 i X 21ÿl j0
2j!!
i
1 ÿ l
2 ÿ l
j ÿ l 1 j
ÿ 1 i X 1ÿl , 2 j! 2 j0 which when compared with the expansion of 1 1 lÿ1
1 ÿ xlÿ1 , with x 12 we get jp0; i j < 21ÿl
2 1, and hence, jp0, i jR1: (ii) For a 1 From Eq. (12a) we have
ÿ 1 i
1 ÿ l i , 21ÿl
2i !!
jpn, i jR
<
for irn:
20
nÿ1 X nÿ1 jp1, iÿj j j j0
nÿ1 ÿj nÿ1 1 X 1 nÿ1
1 2iÿn1 21ÿl j0 j 2 1
1
1
2iÿn1
21ÿl
1
1 2
nÿ1
j
3nÿ1 : 2i1ÿl
can
ÿ 1
i ÿ l p1, iÿ1 i 2
1 ÿ l
ÿ 1 i
i ÿ l
i ÿ 1 ÿ l p1, 0 : i iÿ1 i 1 2 So,
17
(iii) For a 2,3, . . . ,n From Eqs. (16) and (12b) we get a recurrence relation to express p1, iÿj in terms of p1, iÿk where k satis®es jRkRi, 0RjRi, that is 0RjRkRi
p1, iÿj
19
(vi) For a is a noninteger and 0 < a < 1 Following Michieli (1998), we can show that jpa; i j < jp0; i jR1: (v) For a is a noninteger and a > 1 Following Michieli (1998), and using Eq. (17), it
ÿ 1
i 1 ÿ l p1, i and p1, i 2 i1
jp1, i1 j < jp1, i jR2 ÿiÿ1l :
1 : 2iÿj1ÿl
nÿ1 X nÿ1 p1, iÿj j j0
16
ÿ 1 i1
1 ÿ l i1 21ÿl
2
i 1 !!
2 ÿinlÿ2 2nÿ1ÿj
So, from Eq. (19),
From Eq. (16), p1, i1
jp1, iÿn1 jR
From Eqs. (12b) and (16) we have pa; i Pi a ÿ 1 p1; iÿj , and hence j0
j i X nÿ1 pn, i p1, iÿj for i < n, j j0
pn, i
15
p1, i
14
1 2nÿ1ÿj
283
be easily shown that jpa; i jRra; i Pi a ÿ 1 1 iÿj j
2 : It is also evident that ra, i j0 j
j 21ÿl satis®es the relation 1
ra, i
1 1 aÿ1 ra, iÿ1 1ÿl j j: i 2 2
The following series representations of Ia
c, b, l were used as input to the mathematica computer package to generate a number of tables for the generalized secant integral as in Eq. (3) for dierent values of the parameters. For lack of space the tables are not reproduced here.
ÿ ÿ ÿ
ÿ 1 kÿj
i ÿ j ÿ l
i ÿ j ÿ 1 ÿ l
i ÿ j ÿ 2 ÿ l
i ÿ k 1 ÿ l p1, iÿk ,
i ÿ k 1 2kÿj
i ÿ j ÿ 1
i ÿ j ÿ 2
i ÿ j
1 and so, jp1; iÿj j < 2kÿj jp1; iÿk j: For k n ÿ 1, by using Eq. (17) we have
For 0 < cRce < p3 , from Eq. (14) we have
18
284
L.A.-M. Hanna, S.L. Kalla / Radiation Physics and Chemistry 59 (2000) 281±285
Ia
c, b, l e ÿb
4n 3 p, b, l , Ia
c, b, l Ia c 2
1 X pa, i baÿlÿi g i l, b
sec c ÿ 1 : i0
21 For ce < c <
p 2,
3. Complete secant integral
i0
2l ÿ 2, b sec ce ÿ G
a ÿ 2i 2l ÿ 2, b sec c :
22
As studied by Michieli (1998), it turns out that ce p4 is a good choice for both series in Eqs. (21) and (22) to converge fast enough for all practical cases. For c p2 , Eq. (22) leads to (for, sec p2 1 and G
x, 1 0)
X 1 ÿ p p ki b 2iÿ2l2 G a ÿ 2i 2l , b, l Ia , b, l 2 4 i0 p ÿ 2, b 2
where ce p4 : Using Eq. (21), and replacing c by ce p4 , we get a
p , b, l Ie ÿb 2
1 X
h i ÿp pa, i baÿlÿi g i l, b 2 ÿ 1
i0
1 ÿ X p ki b 2iÿ2l2 G a ÿ 2i 2l ÿ 2, b 2
23 i0
By using the functional relations, g
r 1, y rg
r, y ÿ yr e ÿy and G
r 1, y rG
r, y yr e ÿy , we can easily deduce that " # iÿ1 X 1 ÿy lk g
i l, y
l i g
l, y ÿ e
24 y
l k1 k0
and G
r ÿ i, y
G
r ÿ i G
r, y G
r ÿ e ÿy
iÿ1 X G
r ÿ i k0
G
r ÿ k
yrÿkÿ1
1 Ia2
c, b, l ÿ 1 ÿ Ia
c, b, l ÿ 1: b2
Ia
c, b, l Ia
c 2pn, b, l,
In Eq. (4), let c p2 , then Ia
p2 , b, l can be expressed in terms of the hypergeometric function as follows (Prudnikov et al., 1993, p. 326, Eq. (3)) ( p a a 1 a a 1 Ia , b, l b B l, 1 ÿ l ÿ ; , 1 F2 2 2 2 2 2 2 b2 1 1ÿa l; ÿ l 1 F2 ÿ bB l, 2 2 4 1a 3 1a b2 ; , l; 2 2 2 4 2ÿaÿ2l
b G a 2l ÿ 2 1 F2 1 ÿ l; 2 ÿ l ) a lÿa b2 ÿ , ÿ l; , 2 2 4 Re
l > 0, Re
b > 0: For l 1, we have ( p a a 1 2 a b2 a 1 Ia , b, 1 b B 1, ÿ ; , ; 1 F2 2 2 2 2 2 2 4 1 1a ÿ bB 1, ÿ 2 2 ) 1 a 3 3 a b2 ÿa ; , ; b G
a , 1 F2 2 2 2 4 Re
l > 0, Re
b > 0:
28
Eq. (28) can be written as: 82 < X 1 p b Ia , b, 1 ba 4 : k0
2k 1
2k 1 a 2 9 3 = 2 k
b 5 1 ÿ b ÿa G
a ; 2k a
2k !
25
where i 1, 2, 3, . . . : The following recurrence relations can be easily veri®ed: Ia
c, b, l
n is an integer:
we get from Eqs. (9) and (6)
1 ÿ ÿ X Ia
c, b, l Ia ce , b, l ki b 2iÿ2l2 G a ÿ 2i
Ia
26
n is an integer:
G
a ÿ
1 1 X X
ÿ 1 2k1 b 2k1a
ÿ 1 2k b 2ka ÿ
2k !
2k a
2k 1 !
2k 1 a k0 k0
G
a ÿ g
a, b G
a, b:
L.A.-M. Hanna, S.L. Kalla / Radiation Physics and Chemistry 59 (2000) 281±285
In Eq. (4) let c p2 and a n 2, then we get (Prudnikov et al., 1993, p. 326, Eq. (4)), In2
1 lÿ 2
n n2
ÿ1 b p 2 p , b, l 2 p Re
l > 0, Re
b > 0:
G
l
n
@ @ bn
1 b 2 ÿl K
b , 1 l 2
I2
which should be corrected from Eq. (20) as pn, i
i X nÿ1 p1, iÿj : j j0
Also, Eq. (12) in Section 3.1.5., that is
where Kl
x is the modi®ed Bessel function of the second kind (Abramowitz and Stegun, 1972; Prudnikov et al., 1993). For n 0, we get
285
1 3 p 2lÿ 2 b 2 ÿl p G
l K 1
b , , b, l l 2 2 p
Re
l > 0, Re
b > 0:
Ia
c, l e ÿb
1 X 1 1 pa, i b1ÿiÿ 2 g i , b
sec c ÿ 1 , 2 i0
should be read as Ia
c, l e ÿb
1 X 1 1 pa, i baÿiÿ 2 g i , b
sec c ÿ 1 : 2 i0
Further in Prudnikov et al., 1993, p. 326, Eq. (3), there seems to be a misprint in the last term:
2p 2ÿaÿ2b should be replaced by
p 2ÿaÿ2b :
Acknowledgements The authors acknowledge the support of Kuwait University (Project: SM. 181). Appendix A
References
In Michieli's (1998) paper, there are some misprints which we would like to point out here:Eq. (9), that is i 1 X
ÿ 1 Pa, i p 2 j0
j
a ÿ 1
2j ÿ 1!! , 1 ÿ j
2j!!2 j
should be corrected to i 1 X
ÿ 1 pa, i p 2 j0
j
aÿ1 iÿj
2j ÿ 1!! :
2j!!2 j
In Eq. (12a) put l 12 , and notice that !!
12 j
2jÿ1 2 j :The equation in Section 3.1.3. in Michieli (1998) is Pn, i
nÿ1 X nÿ1 u0
j
p1, iÿj ,
Abramowitz, M., Stegun, I.A., 1972. Handbook of Mathematical Functions. Dover, New York. Hungerford, H.E., 1962. Tables of secant integrals of the ®rst and second kinds. Nuclear Science Engineering (14), 312. Michieli, I., 1998. Point kernel calculation of dose ®elds from line sources using expanded polynomial form of buildup factor data: generalized secant integral-series representation. Radiat. Phys. Chem. 51 (2), 121±128. Prudnikov, A.P., Brychkov, Yu.A., Marichev, O.I., 1993. Integrals and Series, vol. 1. Gordon and Breach, London, UK. Sievert, R.M., 1921. Die IntensitaÈtsverteilung der primaÈren gStrahlung in der NaÈhe medizinischer radiumpraÈparate. Acta Radiologica 1, 89±128. Sievert, R.M., 1930. Die g-StrahlungsintensitaÈt an der Ober¯aÈche und in der naÈchsten Umgebung von Radiumnadeln. Acta Radiologica 11, 249±267. Wood, J., 1982. Computational Methods in Reactor Shielding. Pergamon Press, New York.