Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Contents
0.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Preface
Chapter 1 Solutions
Chapter 2 Solutions
Chapter 3 Solutions
Chapter 4 Results Cited
Chapter 4 Solutions
Results from Chapter 5 used
Chapter 5 Solutions
Chapter 6 Solutions
Chapter 7 Solutions
Chapter 8 Solutions
Chapter 9 Solutions
Chapter 10 Solutions
Chapter 11 Solutions
Chapter 12 Solutions
Chapter 13 Solutions
Chapter 14 Solutions
Chapter 15 Solutions
Chapter 16 Solutions
Chapter 17 Solutions
Chapter 18 Solutions
Chapter 19 Solutions
Chapter 20 Solutions
Chapter 21 Solutions
2
4
9
17
23
24
39
40
48
64
67
76
77
98
102
110
122
140
146
153
156
170
171
184
0. Preface
Earl D. Rainville began giving lectures on Special Functions at the University of
Michigan in 1946. The course was well received, and his notes became the basis for
his book, Special Functions, published in 1960. Also in 1946, Sylvester J. Pagano
received his B.S. degree in Electrical Engineering at the Missouri School of Mines
and Metallurgy (MSM). That fall, Pagano was appointed Instructor in Mathematics
at MSM. In 1950, Pagano, now Assistant Professor, spent the summer at Michigan,
where he and Rainville presumably met. In the summers of 1962, 1963, and 1964,
Pagano was again at Michigan, this time as a National Science Foundation Science
Faculty Fellow. Rainville passed away in 1966, the same year Pagano was promoted
to Professor at the University of MissouriRolla (UMR, MSM under its new name).
In the spring of 1966, I was a sophomore at UMR and was taking Elementary Differential Equations; the textbook was the third edition or so of Rainvilles Elementary
Differential Equations. This would be a better story if Pagano had been the instructor in that class, but he wasnt. I did have a class from Pagano later, when
I was a beginning graduate student; it was Operational Calculus, and we used the
Operational Mathematics book by R.V. Churchill. Churchill was Rainvilles Ph.D.
advisor (Michigan 1939), but I knew nothing of any of these connections at the
time. Pagano was a member of both my M.S. and Ph.D. committees at UMR.
Problem 1. Show that the following product converges and find its value:
Y
6
.
1+
(n + 1)(2n + 9)
n=1
Solution 1. (Solution by Leon Hall) By Theorem 3, page 3, this product converges
X
6
converges absolutely.
absolutely because
(n + 1)(2n + 9)
n=1
1+
6
(n + 1)(2n + 9)
(n + 1)(2n + 9) + 6
(n + 1)(2n + 9)
2n2 + 11n + 15
=
(n + 1)(2n + 9)
(2n + 5)(n + 3)
=
.
(n + 1)(2n + 9)
n
Y
1+
So, if
Pn
k=1
n
Y
6
(k + 1)(2k + 9)
(2k + 5)(k + 3)
(k + 1)(2k + 9)
k=1
[7 9 (2n + 5)][4 5 (n + 3)]
=
[2 3 (n + 1)][11 13 (2n + 9)]
[7 9][(n + 2) (n + 3)]
=
[2 3][(2n + 7) (2n + 9)]
21 (n + 2)(n + 3)
=
2 (2n + 7)(2n + 9)
=
then
21 n2 + 5n + 6
21
=
.
n
n 2 4n2 + 32n + 63
8
Note: The use of Theorem 3 is not needed because finding the value of the infinite
product is sufficient itself to show convergence.
lim Pn = lim
Y
n=2
1
1 2
n
=
1
.
2
1
Let
Pn
=
=
n
Y
1
1 2
k
k=2
n
Y
(k + 1)(k 1)
k2
[3 4 (n + 1)][1 2 3 (n 1)]
=
(2 3 4 n)2
(n + 1)
.
=
2n
k=2
n+1
= lim
n 2n
1
=
2
Y
1
=
1 2 .
n
n=2
lim Pn
Y
n=2
1
n
diverges to 0.
Since
1
=0
n
the product diverges to 0. [The product does not converge to 0 because none of the
terms in the product are 0.]
lim Pn = lim
(1 + z 2 ) in |z| < 1.
n=0
n
Y
(1 + z 2 ). Then
k=0
P0 = 1 + z,
P1 = (1 + z)(1 + z 2 )
P2 = (1 + z)(1 + z 2 )(1 + z 4 ) = (1 + z)(1 + z 2 + z 4 + z 6 ).
Assume Pn = (1 + z)
(z 2 )k =
2n+1
X2
(z 2 )k . Then
k=0
Pn+1
k=0
n+1
= Pn (1 + z 2 )
n+1
= Pn + z 2h Pn
n+1
= (1 + z) 1 + z 2 + + z 2
= (1 + z)
n+1
+ z2
n+1
+ z2
+2
n+2
+ + z2
2n+2
X2
(z 2 )k
k=0
2n+1
X2
z 2k ,
k=0
1
n
, for |z| < 1. |1 + z 2 |
2
1z
Y
1
2n
.
and same process works. Thus,
(1 + z ) converges absolutely to
1
z
n=0
1
exp
Problem 5. Show that
diverges.
n
n=1
n
Y
exp
k=1
Sn = log Pn =
1
and let
n
n
X
1
.
k
k=1
Sn is the nth partial sum of the harmonic series, which diverges. As in the proof
of Theorem 2, page 3, Pn = exp Sn and
lim Pn = lim exp Sn = exp lim Sn .
Y
1
Problem 6. Show that
exp
diverges to 0.
n
n=1
Solution 6. (Solution by Leon Hall) Let
log Pn = Sn =
n
n
X
X
1
1
=
k
k
k=1
k=1
as in Problem 5. Then
n
X
1
Pn = exp Sn = exp
k
k=1
!
=
1
!.
n
X
1
exp
k
k=1
Because
X
1
diverges to we have
k
k=1
lim Pn = 0
and so
1
exp
n
n=1
diverges to 0.
Problem 7. Test
Y
n=1
z2
n2
.
Solution 7. (Solution by Leon Hall) The product diverges to 0 for any z such
z2
that z = m, m a positive integer. For all z such that 1 2 6= 0, we have by
n
Y
X
z2
z2
Theorem 3, page 3, that
1 2 is absolutely convergent because
2
n
n
n=1
n=1
is absolutely convergent. In fact,
X
z2
z2 2
.
2 =
n
6
n=1
Y
(1)n+1
converges to unity.
Problem 8. Show that
1+
n
n=1
Solution 8. (Solution by Leon Hall) Let Pn =
n
Y
(1)k+1
1+
.
k
k=1
Pn =
1
2
3
4
n1
n
n!
Rearranging, we get Pn =
= 1 for even n.
n!
Case 2: n is odd. Then
1+1
21
3+1
41
n1
n2
n+1
n+1
Pn =
=
.
1
2
3
4
n2
n1
n
n
In both cases lim Pn = 1.
n
Y
1
Problem 9. Test for convergence:
1 p for real p 6= 0.
n
n=2
Solution 9. (Solution by Leon Hall) For p > 1 the series of positive numbers
Y
X
1
1
is known to be convergent (e.g., by the Integral Test). Thus,
1 p
np
n
n=2
n=2
is absolutely convergent by Theorem 3.
1
> 0 and so convergence and absolute convergence are the
np
X
1
diverges for 0 < p 1, our product diverges by
same. Because the series
np
n=2
Theorem 3.
For p < 0, let p = q where q > 0. Then
1
1 p = 1 nq = 1 + (nq ).
n
But lim (nq ) 6= 0, so in this case our product diverges by Theorem 1.
n
Y
1
Summary:
1 p diverges when p 1 (an p 6= 0), and converges when
n
n=2
p > 1.
For 0 < p 1, 1
Y
sin( nz )
z
n
n=1
Then
an (z)
= 1 + an (z).
sin( nz )
z
n
1 z2
1 z4
1 z6
=
+
+ ...
2
4
3!n 2 5! n 7!
n6
1
1
z
= 2 +O
.
n
6
n2
Thus, there exists a constant M such that
|an (z)| <
M
,
n2
X
M
and because
converges, the product
n2
n=1
(1 + an (z)) =
n=1
sin( nz )
z
n
1 = 1.
n=1
z+
z +
z3 + . . .
n c+n
2!n2
n(c + n)
3!n3
2!n2 (c + n)
X
c
cn
c (k 1)n k
=1+
z+ 2
z2 +
z
n(c + n)
2n (c + n)
k!nk (c + n)
k=3
c
1
1
=1+
.
z
z2 + O
n(c + n)
2n(c + n)
n2
Thus, for c not a negative integer and for any finite z, there is a constant M such
M
that |an (z)| 2 and so by Theorems 3 and 4 the product converges absolutely and
n
uniformly.
2. Chapter 2 Solutions
0 (z)
1 X
Problem 1. Start with ()
=
(z)
z n=1
1
1
, prove that
z+n n
20 (2z) 0 (z) 0 (z + 21 )
= 2 log 2,
(2z)
(z)
(z + 12 )
and thus derive Legendres duplication formula, page 24.
Solution 1. Applying () three times and simplifying yields
20 (2z) 0 (z) 0 (z + 12 )
(2z)
(z)
(z + 12 )
X
X
X
1
1
1
1
2
1
2
2
1
++ ++
+
= 2
2z
z
z+n n
z + 21 n=1 2z + n n
z + 21 + n n
n=1
n=1
2n
n
n
X
X
X
2
2
2
1
1
1
2
=
lim
+ lim
+ lim
n
n
2z + 1 n
2z + k k
z+k k
2z + 1 + 2k k
k=1
k=1
k=1
"
#
2n
n
n
X
X
X
2
2
2
2
=
+ lim
+ 2H2n +
Hn +
Hn
2z + 1 n
2z + k
2z + 2k
2z + 2k + 1
k=1
k=1
"k=1
#
2n
2n+1
X
X
2
2
2
+ lim
+
+ 2H2n 2Hn
=
2z + 1 n
2z + k
2z + k
k=1
k=2
2
2
2
=
+
+ lim
+ lim (2H2n 2Hn )
2z + 1 2z + 1 n 2z + 2n + 1 n
= 0 + 0 + 2 lim [(H2n log 2n) (Hn log n) + log 2n log n]
n
= 2[ + log 2]
= 2 log 2.
= 2 log 2.
(2z)
(z)
(z + 12 )
Now let z =
1
to get
2
0 (1) 0
2
(1)
1
2
1
2
0 (1)
= 2 log 2,
(1)
1
0
But (1) = 1, (1) = , ( 2 ) = , hence
0 ( 12 )
= 2 log 2,
1
and by rearrangement,
1
0 ( ) = ( + 2 log 2) .
2
Z
Problem 3. Use Eulers integral form (z) =
et tz1 dt to show that
0
(z + 1) = z(z).
Z
u = tz
du = ztz1
= 0 + z(z),
where lim tz et converges for R(z) > 0
t
.
(z) = lim
but
B(z, n) =
(z)(n)
(z)(n 1)!
(n 1)!
=
=
.
(z + n)
(z)n (z)
(z)n
Hence
(z) = lim nz B(z, n).
n
(p)(q)
, so
(p + q)
(p + 1)q(q)
p(p)(q + 1)
=
= qB(p + 1, q).
(p + q + 1)
(p + 1 + q)
(p)(q)
(p + q)
(p)(q)
=
(p + q + 1)
p+q
(p + q)(p)(q)
=
(p + q + 1)
q(p)(q)
p(p)(q)
+
=
(p + q + 1) (p + q + 1)
(p + 1)(q)
=
(p)(q + 1)
(p + q + 1) +
(p + q + 1)
= B(p + 1, q) + B(p, q + 1).
=
(c)
(p + q)B(p, q + 1)
(p + q)(p))(q + 1)
(p + q + 1
(p + q)(p)(q + 1)
=
(p + q)(p + q)
(p)(q + 1)
=
(p + q)
(p)q(q)
=
(p + q)
= qB(p, q).
(d)
B(p, q)B(p + q, n)
(p)(q) (p + q)(n)
(p + q) (p + q + n)
(p)(q)(n)
=
(p + q + n)
(q)(n) (p)(q + n)
=
(q + n) (p + q + n)
= B(q, n)B(q + n, p).
=
11
n!
.
(p)n+1
=
=
=
=
=
(p)(n + 1)
(p + n + 1)
(p)(n + 1)
(p + 1)n (p + 1)
(p)n!
(p + 1)n p(p)
n!
p(p + 1)n
n!
.
(p)n+1
Problem 7. Evaluate
1
Solution 7. Let A =
1
1+x
, x = 2y
2
= ( + 1) . . . ( + n k 1)
( + 1) . . . ( + n k 1)[( + n k)( + n k + 1) . . . ( + n 1)]
=
( + n 1)( + n 2) . . . ( + n k)
()n
=
( + n k)k
()n
=
(1)k (1 n)k
(1)k ()n
=
.
(1 n)k
(1)k n!
.
(n)k
13
as desired.
(1 n)
()
(1 n)
=
()( 1)( 1)
(1 n)
=
()( 1) . . . ( n + 1)(1 n)
1
=
,
(1)n ()n
1
In the following problems, the function P (x) := x bxc .
2
Z x
Problem 10. Evaluate
P (y)dy.
0
1
. Let m be an
2
=
2
2
2
1
1
=
P 2 (x)
2
4
1 2
1
= P (x) .
2
8
Problem 11. Use integration by parts and the result of the above exercise to show
that
Z
P (x)dx
1
8(1 + n) .
1
+
x
n
Z
P (x)
dx and use integration by parts
Solution 11. Consider
1+x
n
1
1
2
2
Z
Z
1 P (x) 4
P (x)
1 P (x) 4
1
dv = P (x)dx
u = (1 + x)
dx =
dx.
+
1+x
2
1+x
2 n
(1 + x)2
n
v=
1 2
1
P (x)
2
8
du = (1 + x)2 dx
Now max P 2 (x)
Z
1
1
1
= and P 2 (n) = implies
4
4
4
P (x)
1
dx = 0 0 +
1+x
2
P 2 (x)
1
4 dx
(1 + x)2
and
Z
1
Z
P (x) 1
1
1
4
dx
=
dx
1+x
2 n (1 + x)2
8 1+x n
or
P (x) 1
1
dx
.
1+x
8 1+n
Z
Problem 12. With the aid of the above problem, prove the convergence of
0
Z
Solution 12.
0
cise 11,
P (x)
dx converges lim
n
1+x
Z
lim
n
Z
Hence
0
P (x)dx
.
1+x
P (x)
dx = 0 but from Exer1+x
1
P (x)
dx lim
= 0.
N 8(1 + n)
1+x
P (x)
dx < .
1+x
(y 21 )dy
1
=
1+n+y
12
P (x)dx
is convergent.
1+x
Solution 13. (Solution by Leon Hall) Because P (x) is periodic with period 1, it is
clear that
Z
Z n+1
X
P (x)
P (x)
dx =
dx.
1+x
1+x
0
n=0 n
Let x = n + y. Then
Z
n+1
P (x)
dx
1+x
=
Z0 1
=
Z0 1
=
0
P (n + y)
dy
1+n+y
P (y)
dy
1+n+y
y 21
dy.
1+n+y
15
+ ...
=1 n+
2
n + 1 2(n + 1)2
3(n + 1)3
4(n + 1)4
Z 1
X
(y 21 )
To determine the convergence of
dy we compare with the known
y+n+1
n=0 0
X
1
convergent series
using the limit comparison test.
2
n
n=1
Z 1
(y 12 )
dy
1
3 X
0 y+n+1
2
2
=
n
n
n
+
1
2
k(n
+
1)k
n2
k=1
3
1
1
1
1
= n2 n2 n +
+
+
O
2
n + 1 2(n + 1)2
3(n + 1)3
n + 1
3
2
2
3
2
6n (n + 1) n (n + 2 )[6(n + 1) 3(n + 1) + 2]
1
=
+O
6(n + 1)3
n +1
15 2
3
6n5 + 18n4 + 18n3 + 6n2 [6n5 + 18n4 + 37
1
2 n + 2 n ]
+O
=
3
n+1
6(n + 1)
1 3
2n
1
=
+O
.
6(n + 1)3
n+1
Thus,
lim n
and
Z
X
n=0
Z
0
(y 21 )
1
dy = ,
y+n+1
12
(y 21 )
dy =
y+n+1
Z
0
P (x)
dx
1+x
converges.
Problem 14. Apply Theorem 11, page 27, to the function f (x) = (1 + x)1 ; let
n and thus conclude that
Z
1
=
y 2 P (y)dy.
2
1
Solution 14. (Solution by Leon Hall) Let f (x) =
1
. Theorem 11, page 27
1+x
gives with p = 0, q = n,
Z n
Z n
n
X
1
1
1 1
1
=
dx + +
+
f 0 (x)B1 (x)dx.
1+k
2 2 1+n
0 1+x
0
k=0
So,
n
X
k=0
1
log(1 + n)
1+k
1 1
= +
2 2
1
1+n
+
0
B1 (x)
dx
(1 + x)2
Z n+1
1 1
1
B1 (y + 1)
dy
= +
+
2 2 1 + n Z1
y2
n+1
1 1
1
B1 (y)
= +
dy
2 2 1+n
y2
1
y := x + 1
1
[cos(x y) cos(x + y)]
2
to prove that
1
2 sin (c a) sin (c b)
2 sin(c) sin (c a b)
sin(c) sin (c a b) sin (c a) sin (c b)
=
.
sin(c) sin (c a b)
=1
Now using the given trig identity, we get, continuing the equality:
+ b) cos (2c a b)] 12 [cos (b a) cos (2c a b)]
1
2 [cos (a + b) cos (2c a b)]
1
2 [cos (b a) cos (a + b)]
= 1
2 [cos (a + b) cos (2c a b)]
sin(a) sin(b)
=
sin(c) sin (c a b)
0 sin(a) sin(b)
=
.
sin (c 1) sin (c a b)
=
1
2 [cos (a
17
3. Chapter 3 Solutions
Problem 1. With the assumptions of Watsons lemma, show, with the aid of the
X
k
convergence of the series F (t) =
an expt
1 in |t| a + , that for 0
r
k=1
t a, there exists a positive constant c1 such that
n
X
k
n+1
ak expt
1 < c1 expt
1 .
F (t)
r
r
k=1
Solution 1. We wish to show that there exists c1 such that for 0 t a (see
problem (2) for t > a),
n
X
n+1
k
1
ak t r < c1 t r 1
F (t)
k=1
an t r 1 = F (t)
n=1
in |t| a we write
n
X
k
1
ak t r
F (t)
k=1
X
k
1
ak t r
=
k=n+1
X
kn1
n+1
1
=t r
ak t r
k=n+1
X
kn1
n+1
1
t r
ak a r
k=n+1
< c1 t
n+1
r 1
X
kn1
where c1 >
ak a r . Remember from Watsons lemma
k=n+1
F (t) =
an t r 1
k=1
Problem 2. With the assumptions of Watsons lemma, page 41, show that for
t > a, there exist positive constants c2 and such that
n
X
k
n+1
ak expt
1 < c2 expt
1 et .
F (t)
r
r
k=1
Solution 2. Under the assumption of Watsons lemma we wish to show that for
t > a, there exist constants c2 , such that
n
X
n+1
k
ak t r 1 < c2 et t r 1 .
F (t)
k=1
n+1
< M1 ebt t r 1 + M2 t
n+1
< c1 et t r 1 .
n+1
r 1
ebt
Problem 3. Derive the asymptotic expansion (6) immediately preceding these exercises by applying Watsons lemma to the function
Z xt
te
0
f (x) =
dt
t + t2
0
and then integrating the resultant expansion term by term.
Solution 3. Consider Watsons lemma with F (t) =
F (t) =
Choose r =
Hence
t
. Hence
1 + t2
X
X
t
n+1 2n+1
=
(1)
t
=
(1)n t2n1 ; 0 t < 1.
1 + t2
n=0
n=1
1
1
t
1
1
, a = , = . Also for t 0, et 1 and
< .
2
2
2
3
1+t
2
|F (t)| < 1 et ; t
1
,
2
(1)n t 1/2 1
n=1
for |t|
1 1
5
+ = .
2 3
6
f 0 (x)
n
(1)
X
n=1
x2n
n
1/2
text
dt we have
1 + t2
=
X
(1)n (2n 1)!
.
x2n
n=1
19
Then
Z
f 0 (s)ds
Z
X
n=1
X
(1)n (2n 2)!
.
x2n1
n=1
f 0 (s)ds
()
x
Now
d
dx
ext
dt =
1 + t2
Z
0
text
dt f 0 (x).
1 + t2
Let A label the integral in the middle of the last formula. Hence
Z
f (x) =
0
ext
dt,
1 + t2
which we label as B. Also note that integrals A and B are uniformly convergent. Hence
Z
ext
dt = 0.
lim f (x) =
lim
x
x 1 + t2
0
Therefore
Z
f 0 (s)ds = f (s) = 0 f (x).
x
By (),
0 f (x)
X
(1)n (2n 2)!
; |x| , R(x) > 0.
x2n1
n=1
So
f (x)
ext
dt
1 + t2
0
x2n1
n=1
X (1)n (2n)!
n
X
(1) (2k)!x
k=0
2k1
= (1)
n+1
Z
0
ext t2n+2
dt,
1 + t2
and thus obtain not only (6) but also a bound on the error made in computing with
the series involved.
X
1
=
(1)k t2k we have
1 + t2
k=0
1
1 + t2
=
=
=
n
X
k=0
n
X
k=0
n
X
k=0
So
(1)k t2k +
(1)k t2k
k=n+1
k 2k
(1) t
+ (1)n+1 t2n+2
(1)k t2k
k=0
t2n+2
.
(1)k t2k + (1)n+1
1 + t2
ext
dt
1 + t2
Z0
Z
n
X
=
ext
(1)k t2k dt + (1)n+1
Z
f (x)
n
X
(1)k
k=0
Z
ext t2n+2
dt
1 + t2
ext t2n+2
dt.
1 + t2
k=0
Z
0
Then
n
X
k (2k)!
(1) 2k+1
f (x)
x
k=0
Z xt 2n+2
e t
=
dt
2
1+t
Z 0
<
|ext |t2n+2 dt
Z0
<
eRe(x)t t2n+2 dt
0
=
In the region |arg x|
|x|
(2n + 2)!
.
[Re(x)]2n+3
and so
f (x)
(1)k
n=0
(2k)!
x2k+1
21
X
et t1 dt ex
(1)n n!xn1 .
x
n=0
et
dt as x . Using integration by parts,
t
x
Z
1
f (x) = et
t2 et dt
t
x
x
Z
1 t
1
1
e dt
= ex + 2 et + 2
3
x
t x
t
0
Z
1 t
1 t
1
x 1
=e
2 3e
23
e dt
4
x x2
t
t
x
x
Z
1
t
u=
u=
dv = et dt
1
t2
v = et
1
t2
du = 2
dv = et dt
1
t3
v = et
= ....
so
ex f (x)
X
(1)n n!
k=0
xn+1
or
Z
t1 et dt ex
X
(1)n n!
as x .
xn+1
n=0
X
Hn (x)tn
n!
n=0
for all x and t, as in Chapter 11. Also let the complementary error function erfc x
be defined by
Z
2
erfc x = 1 erf x =
exp( 2 )d.
x
X
1
exp x s
exp( 2 )d
Hn (x)sn1 , s ,
1
2
sx
2
n=0
and thus arrive at the result
"
2 #
X
1 1
1 1
1 1
exp
t
t x
t x
erfc
Hn (x)tn , t 0+ .
2
2
2
n=0
Solution 6. Because expt (n 1) = tn1 , and if
X
Hn (x) n
F (t) = exp(2xt t2 ) =
t ,
n!
n=0
we can write
F (t) =
X
Hn1 (x)
expt (n 1)
(n 1)!
n=1
so the first condition in Watsons Lemma is satisfied for any fixed x with r = 1,
a = 1, and any > 0. Also,
2
X
Hn1 (x)(n)
est F (t)dt
(n 1)!sn
0
n=1
as s , or
Z
X
1
2
Hn (x)sn1 ,
exp t + 2 x s t dt
2
n=0
X
1
1
exp t + s x
dt
Hn (x)sn1
exp x s
2
2
0
n=0
1
as s , and if we let = t + s x we get
2
"
2 # Z
X
2
1
exp x s
e d
Hn (x)sn1 ,
1
2
sx
2
n=0
as s .
1
Using the definition of erf c, and then making the substitution t =
(not the
s
inverse Laplace transform) we get
"
2 #
X
1
1
exp x s
erfc
sx
Hn (x)sn2 ,
2
2
2
n=0
as s or
exp
2t
as t 0+ as desired.
"
1
x
2t
2 #
erfc
1
x
2t
X
n=0
Hn (x)tn ,
23
Problem 7. Use integration by parts to show that if R() > 0, and if x is real,
Z
X
(1)n ()n
et t dt x1 ex
, x ,
xn+1
x
n=0
of which Problem 5 is the special case = 1.
Solution 7. Integration by parts with u = t and dv = et dt yields
Z
Z
t
x
e t dt = e x
et t(+1) dt.
x
The same integration by parts with replaced by + 1 applied to the last integral
gives
Z
Z
h
i
et t dt = ex x 1
+ ( + 1)
et t(+2) dt.
x
x
x
Continuing, after n + 1 integrations by parts, we have
Z
Z
n
X
(1)k ()k
t
x +1
n+1
e t dt = e x
+ (1)
()n+1
et t(+n+1) dt
xk+1
x
x
k=0
or
x 1
e x
e t
x
dt
n
X
(1)k ()k
k=0
xk+1
x 1
=e x
(1) ()n+1
et t(+n+1) dt.
Thus,
the desired asymptotic series if the right side of the last equation is
we have
1
+ 1 as x . This is true because
O
xn
Z
x 1
t (+n+1)
n
e x
(1)
()
e
t
dt
n+1
x
x 1
Z
e x
()n+1 t
e
dt
+n+1
x x
x
e ()n+1 x
=
e
xn+2
|()n+1 |
=
xn+2
1
=O
xn+1
as x .
4. Chapter 4 Results Cited
By definition,
F (a, b; c; 1)
X
(a)n (b)n
.
(c)n n!
n=0
(c)(c a b)
.
(c a)(c b)
pg.66, (2):
a a+1
1
,
;
a,
a
b
+
;
2
4z
x2
(1 + z)a F
=F
1
(1 + z)2
b
+
;
1 + a b;
2
z
< 1,
Theorem 2. If |z| < 1 and
1 z
a, c b;
a, b;
z
z = (1 z)a F
F
.
1z
c;
c;
,
;
2
2
a, b;
2
y
=F
2y .
(1 y)a F
2
(1
y)
2b;
1
b+ ;
2
1
Theorem 5. Kf a + b + is neither zero nor a negative integer, and if |x| < 1 and
2
|4x(1 x)| < 1,
a, b;
4x(1 x) .
F
1
a+b+
2
Theorem 6. If c is neither zero nor a
|4x(1 x)| < 1,
a, 1 a;
x = (1 z)c1 F
F
c;
4x(1 x) .
c;
5. Chapter 4 Solutions
a, b;
d
F
dx
c;
a + 1, b + 1;
ab
x = F
c
c + 1;
x .
25
Solution 1. From
F (a, b; c; 1)
X
(a)n (b)n
,
(c)n n!
n=0
we get
d
F (a, b; c; x)
dx
=
=
X
(a)n (b)n xn1
(c)n (c 1)!
n=1
X
(a)n+1 (b)n+1 xn
(c)n+1 n!
X
(a + 1)n (b + 1)n xn
ab
=
c n=0
(c + 1)n n!
ab
= F (a + 1, b + 1; c + 1; x).
c
n=0
1
a+b+ ;
2
1
1
a+b+
1
2
2
.
2 = 1
1
1
1
1
c+ a
c a+
2
2
2
2
2
2a, 2b;
1
a+b+
2
2 . From Theorem 5,
1
a+b+ ;
2
2a, 2b;
x =F
a, b;
1
a+b+
2
4x(1 x)
1
1
for |x| < 1, |4x(1x)| < 1. We need to use x = , but note that R(a+b+ ab) >
2
2
0. Hence, by Theorem 1,
2a, 2b;
a+b+
1
2
2 =F
a, b;
a+b+
1
2
1
1
a+b+
2
2
1 =
.
1
1
b+
a+
2
2
a, 1 a;
F
c;
1
2 (c)
2
1
=
1
1
ca+1
2
c+ a
2
2
2
1c
a, 1 a;
Solution 3. Consider F
c;
c1 c+a1
,
;
2
x = (1x)c1 F 2
4x(1 x) .
c;
By Theorem 6, for |x| < 1, |4x(1 x)| < 1,
ca c+a1
a, 1 a;
,
;
2
x = (1 x)c1 F 2
F
4x(1 x) .
c;
c;
c a c a 1
Since R c + +
> 0, we may use Theorem 1 to conclude that
2 2 2 2 2
c1 c+a1
a, 1 a;
c1
,
;
1
1
2
F
F 2
=
1
2
2
c;
c;
1
(c)
2
,
=
c
+
a
ca+1
2c1
2
2
as desired.
Problem 4. Obtain the result
n, b;
F
c;
(c b)n
1 =
.
(c)n
(c)(c b + n)
(c b)n
=
.
(c + n)(c b)
(c)n
Actually the condition R(c b) > 0 is not necessary because of the termination of
the series involved.
Problem 5. Obtain the result
n, a + n;
F
c;
(1)n (1 + a c)n
1 =
.
(c)n
Solution 5.
n, a + n;
1 =
F
c;
(c)(c a)
.
(c + n)(c a n)
n, a + n;
F
a;
(1)n (1 c + a)n
1 =
,
(c)n
27
as desired.
Problem 6. Show that
n1 b n;
F
a;
Solution 6.
n, 1 b n;
F
a;
1 =
1 =
(a + b 1)2n
.
(a)n (a + b 1)n
(a)(a 1 + b + 2n)
(a 1 + b)2n
=
.
(a + n)(a 1 + b + n)
(a)n (a 1 + b)n
n
X
k=0
X n!()k ()k
(n)k ()k
=
.
k!(1 + n)k
n!(n k)!()n
k=0
X
()n gn tn
n!
n=0
n
X
X
()k ()nk tn
k!(n k)!
!
!
X ()n tn
()n tn
=
n!
n!
n=0
n=0
= (1 t) (1 t)
= 1.
n=0 k=0
= Dn
=
=
X
k=0
X
k=0
d
). We have
dx
X
(a)k (b)k xn+k+a1
k=0
(c)k k!
1 1
1
a,
a
+
;
n, a;
2 2
2
4x
x = (1 x)a F
F
.
(1 x)2
1 + a + n;
1 + a + n;
Solution 9. From (2) on page (66) we get
a a+1
,
;
2
2
a
4z
(1 + z) F
(1 + z)2
1 + a b;
Use z = x, b = n to arrive at
a a+1
,
;
2
2
a
(1 x) F
1 + a + n;
a, b;
=F
1
+
a
b;
.
4x
(1 x)2
a, n;
=F
1 + a + n;
x ,
as desired.
1
Problem 10. In Theorem 23, page 65, put b = , a = + , 4x(1 + x)2 = z and
2
thus prove that
1
, + ;
2
F
21
1
2
.
z = (1 z) 1 + 1 z
a, b;
4x
(1 + x)2a F
=F
(1 + x)2
ab;
Put b = , a = +
1
and
2
1
a, a b + ;
1
b+ ;
2
4x
= z.
(1 + x)2
Then
zx2 + 2(z 2)x + z = 0
p
zx = 2 3 z 2 4z + 4 32 = 2 z 2 1 z.
Now x = 0 when z = 0, so
zx = 2 z 2 1 z = 1 z + 1 2 1 z
or
x=
(1
1 z)2
(1 1 z[1 (1 z)]
=
.
z
z(1 + 1 z)
x2
.
Thus
29
1 1z
x=
1+ 1z
and
1+x=
.
1+ 1z
Then we obtain
4x
4(1 1 z) (1 + 1 z)2
= z,
(1 + x)2
4
1+ 1z
1
a check. Now with b = , a = + , Theorem 4 yields
2
1
21
+
,
;
2
F
=F
z
1+ 1z
2;
1
+ , 1;
x2
1
+ ;
2
1;
x2
=1F0
;
= (1 x2 )1 .
2
2 1z
and 1 + x =
,
Since 1 x =
1+ 1z
1+ 1z
4 1z
2
(1 x ) =
.
(1 + 1 z)2
Thus we have
1
, + ;
2
F
2+1
2
2
2
1
=
(1 z) 2
1+ 1z
1+ 1z
21
2
21
= (1 z)
,
1+ 1z
2;
1
1
, ;
, + ;
1
2
2
2
F
2 = (1 z) F
2;
2;
2
F
2;
as desired.
z =
1+ 1z
21
,
x = (12x)ac F
c;
a, 1 a;
(1x)1c F
c;
a, 1 a;
x
(1 x)1c F
a;
ca c+a1
,
;
2
2
=F
c;
ca c+a1
,
;
2
=F 2
4x(x 1
.
(1 2x)2
4x(1 x)
1 (1 2x)2
c;
ca
ca+1
;
a c 2 ,c
2
2
= (1 2x) 2 F
1 + (1 2x)2
(1 2x)2
c;
ca ca+1
,
;
2
2
ac
= (1 2x) F
c;
4x(x 1)
,
(1 2x)2
a, c b;
z
w = (1 z)a F
.
1z
c;
x(1 x)
31
u = 0,
(1x)4 2 + 2(1 x)3 (1 x)2
dx
x
dx
x
or
x(1 x)
d2 u
du
+ [2x {c(1 x) + (a b 1)x}]
a(c b)u = 0,
dx2
dx
or
d2 u
du
a(c b)u = 0.
+ [x (a b + c + 1)x]
2
dx
dx
Now (3) is a hypergeometric equation with parameters = c, + beta + 1 =
a b + c + 1, = a(c b). Hence = a, = c b, = c. One solution of (3) is
(3)x(1 x)
u = F (a, c b; c; x),
so one solution of equation (1) is
a, c b;
W = (1 z)a F
z
.
1z
c;
Problem 13. Use the result of Exercise 12 and the method of Section 40 to prove
Theorem 2.
z
< 1,
Solution 13. We know that in the region in common to |z| < 1 and
1 z
there is a relation
a, c b;
z
1c
(1z)a F
= AF (a, b; c; z)+Bz F (a+1c, b+1c; zc; z).
1z
c;
Since c is neither zero nor a negative integer, the last term is not analytic at z = 0.
Hence B = 0. Then use z = 0 to obtain 1 1 = A 1, so A = 1. Hence
a, c b;
z
F (a; b; c; z) = (1 z)a F
.
1z
c;
Problem 14. Prove Theorem 3 by the method suggested by Exercises 12 and 13.
Solution 14. (Solution by Leon Hall) Note that the first two parameters in F (a, b; c; z)
are interchangeable, so results involving one of them also apply to the other. By
Exercise 12,
z
F (a, b; c; z) = (1 z)a F a, c b; c;
.
1z
z
Let w =
so this becomes
1z
F (a, b; c; z) = (1 z)a F (a, c b; c; w).
Again, by Exercise 12, applied to the second parameter,
w
a
(cb)
F (a, b; c; z) = (1 z) (1 w)
F c a, c b; c;
.
1w
w
But 1 w = (1 z)1 , and
= z, so
1w
F (a, b; c; z)
as desired.
Problem 15. Use the method of Section 39 to prove that if both |z| < 1 and
|1 z| < 1, and if a, b, c are suitably restricted,
a, b;
a, b;
(c)(c
b)
1z
z =
F
F
(c a)(c b)
1 + b + 1
c;
c;
c a, c b;
cab
(c)(a + b c)(1 z)
1 z .
+
F
(a)(b)
c a b + 1;
Solution 15. (Solution by Leon Hall) We denote the hypergeometric differential
equation:
z(1 z)w00 (z) + [c (a + b + 1)z]w0 (z) abw(z) = 0
by HGDE. If we make the change of variable z = 1 y, then HGDE becomes
y(1 y)w00 (y) + [c (a + b + 1)y]w0 (y) abw(y) = 0
where c = a + b + 1 c. Thus, two linearly independent solutions are
F (a, b; c; y)
and
y 1c F (a + 1 c, b + 1 c; 2 c; y).
These solutions as function of z are valid in |1 z| < 1 and are
F (a, b; a + b + 1 c; 1 z)
33
and
(1 z)cab F (c a, c b; c a b + 1; 1 z).
Thus, in the region D where both |z| < 1 and |1 z| < 1,
F (a, b; c; z) = AF (a, b; a+b+1c; 1z)+B(1z)cab F (ca, cb; cab+1; 1z)
for some constants A and B.
Assume Re(c a b) > 0 and c 6= 0 or a negative integer and let z 1 inside
the region D to get
F (a, b; c; 1) = A 1 + B 0.
Thus, by Theorem 18, page 49, we get
A=
(c)(c a b)
.
(c a)(c b)
Now, let z 0 inside the region D and assume Re(1c) > 0 and neither a+b+1c
nor c a b + 1 is zero or a negative integer. Then
1 = AF (a, b; a + b + 1 c; 1) + BF (c a, c b; c a b + 1; 1)
and we get
1 AF (a, b; a + b + 1 c; 1)
.
F (c a, c b; c a b + 1; 1)
Again using Theorem 18, this becomes
B=
1
B=
(c)(cab)(a+b+1c)(1c)
(ca)(cb)(b+1c)(a+1c)
(cab+1)(1c)
(1b)(1a)
=
=
Show that
L1
1
s
a, b;
F
s + 1;
a, b;
z =F
1;
z(1 et ) .
1
s
a, b;
F
1 + s;
X
(a)n (b)n z n 1
1
A=
L
.
n!
s(1 + s)n
n=0
But
1
s(s + 1)n
(1 + s)
s(1 + s + n)
1 (1 + s)(1 + n)
=
sn! (1
+ s + n)(1)
n, s;
1
1
=
F
n!s
1 + s;
n
X
(n)
k (s)k
1
= n!s
.
k!(1 + s)k
=
k=0
Hence
n
1
1 X (n)k
.
=
s(s + 1)n
n!
k!(s + k)
k=0
Then
L 1
1
s(s + 1)n
=
1
1 X (n)k ekt
= (1 et )n .
n!
k!
n!
k=0
Therefore
a, b;
n
t n
X
(a)n (b)n z (1 e )
A=
=F
n!n!
n=0
1;
z(1 et ) .
There are many other ways of doing Exercise 16. Probably the easiest, but most
undesirable, is to work from right to left in the result to be proved. It is hard to see
any chance for discovering the relation that way.
Problem 17. With that notation of Exercise 16 show that
n 3+n
1+ ,
;
2
2
a(n
+
2)
a2
L{tn sin at} =
F
2
n+2
s
s
3
;
2
Solution 17. We wish to obtain the Laplace Transform of tn sin at. Now
n
t sin at =
X
(1)k a2k+1 tn+2k+1
k=0
(2k + 1)!
and
L {tm } =
(m + 1)
.
sm+1
Hence
L {tn sin at}
X
(1)k a2k+1 (n + 2k + 2)
(2k + 1)!sn+2k+2
k=0
sn+2
(2)2k s2k
n+2
n+3
k
(1)
a2k
2
2
a(n + 2) X
k
k
=
3
sn+2
2k
k=0
k!
s
2 k
n+2 n+3
,
;
2
2
a(n + 2)
a2
.
F
=
sn+2
s2
3
;
2
k=0
(1)n
n!
=
we know that
(n + 1)!
(2)n
log(1 + x)
X
(1)n xn+1
n+1
n=0
(1 y )
X
n=0
using
y 2n
n!
1
2
1
=
2n + 1
n+
1
2 n
3
2 n
1
2 n
1
2
=
x
to get
2 21
(1 y )
dy =
X
n=0
Thus we arrive at
arcsin x =
X
n=0
= xF
1
2 n
1
2 n
x2n+1
.
n!(2n + 1)
1
2n+1
2 nx
3
2 n n!
1 1 3 2
, ; ;x .
2 2 2
35
Finally form
(1 + y 2 )1 =
(1)n y 2n
n=0
to obtain
Z
x
2 1
(1 + y )
0
X
X
(1)n
(1)n x2n+1
dy =
=
2n + 1
n=0
n=0
or
arctan x = xF
1
2 n
3
2 n
x2n+1
(1)n
n!
1
3
2
, 1; ; x .
2
2
1 1
Show that K = F
, ; 1; k 2 .
2
2 2
Z 2
d
p
Solution 19. From K =
we obtain
0
1 k 2 sin2
2n 2n
Z 2 X
1
sin d
2 nk
.
K=
n!
0 n=0
But
Z
2n
sin
0
Hence
1
2 12 12 n
n + 21 12
1 1
2 n
1
=
=
=
.
d = B n + ,
2
2 2
2(n + 1)
2n!
2 n!
X
K=
2 n=0
1
2 n
1
2 n
k 2n
n!n!
= F
2
1 1
2
, ; 1; k .
2 2
1 1
2
, ; 1; k .
2 2
Z 2 p
Solution 20. From E =
1 k 2 sin2 d, we get
Show that E = F
2
X
12
k 2n sin2n d
n!
0 n=0
1 2n
1
X
2 n 2 n k
=
.
2 n=0
n!n!
Z
Hence
E=
F
2
1 1
, ; 1; k 2 .
2 2
Problem 21. From the contiguous function relations 1-5 obtain the relations 6-10.
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
37
(6)
(10)
which checks with (6). Easier method: in (6) interchange a and b. From (1) and
(3) we get
[a + (b c)z (a b)(1 z)]F = b(1 z)F (b+) c1 (c a)(c b)zF (c+),
or
(12)
Problem 22. The notation used in Exercise 21 and in Section 33 is often extended
as in the examples
F (a, b+) = F (a 1, b + 1; c; z),
F (b+, c+) = F (a, b + 1; c + 1; z).
Use the relations (4) and (5) of Exercise 21 to obtain
F (a) F (b) + c1 (b a)zF (c+) = 0
and from it, by changing b to (b + 1) to obtain the relation
(c 1 b)F = (c a)F (a, b+) + (a 1 b)(1 z)F (b+),
or
(c 1 b)F (a, b; c; z) = (c a)F (a 1, b + 1; c; z) + (a 1 b)(1 z)F (a, b + 1; c; z),
another relation we wish to use in Chapter 16.
Solution 22. From Exercise 21 equation (4) and (5) we get
(1)(1 z)F = F (a) c1 (c b)zF (c+),
(2)(1 z)F = F (b) c1 (c a)zF (c+).
From the above we get
F (a) F (b) + c1 (b a)zF (c+) = 0.
Now replace b by b + 1 to write
F (a, b+) F + c1 (b + 1 a)zF (b+, c+) = 0,
or
F (a, b; c; z) = F (a 1, b + 1; c; z) + c1 (b + 1 a)zF (a, b + 1; c + 1; z).
Problem 23. In equation (9) of Exercise 21 shift b to b + 1 to obtain the relation
(c 1 b)F = (c a)F (c, b+) + (a 1 b)(1 z)F (b+),
or
(c 1 b)F (a, b; c; z) = (c a)F (a 1, b + 1; c; z) + (a 1 b)(1 z)F (a, b + 1; c; z),
another relation we wish to use in Chapter 16.
39
Theorem 7. If z is nonintegral,
.
sin z
Theorem 8. (Dixons Theorem) The follow is an identtiy of a, b, and c are so
restricted that each of the functions involved exists:
(z)(1 z) =
3 F2
a, b, c;
1 =
1 + a b, 1 + a c;
Theorem 9. If R() > 0, R() > 0, and if k and s are positive integers, then
insdie the region of convergence of the resultant series
Z t
a1 , . . . , ap ;
cxk (t x)s
x1 (t x)1 p Fq
0
b1 , . . . , b q ;
+1
+k1 +1
+s1
, ,
,...,
;
a1 , . . . , a p , k , k , . . . ,
k
s
s
s
+ ++1
++k+s1
b1 , . . . , bq ,
,
,...,
;
k+s
k+s
k+s
Theorem 10. If z is nonintegral,
.
(z)(1 z) =
sin z
Theorem 11. (Dixons Theorem) The follow is an identtiy of a, b, and c are so
restricted that each of the functions involved exists:
a, b, c;
(1 + 21 a)(1 + a b)(1 + a c)(1 + 21 a b c)
1 =
.
3 F2
(1 + a)(1 + 12 a b)(1 + 21 a c)(1 + a b c)
1 + a b, 1 + a c;
Theorem 12. If R() > 0, R() > 0, and if k and s are positive integers, then
insdie the region of convergence of the resultant series
Z t
a1 , . . . , ap ;
cxk (t x)s
x1 (t x)1 p Fq
0
b1 , . . . , b q ;
k k ss ctk+s
(k + s)k+s
+1
+k1 +1
+s1
, ,
,...,
;
a1 , . . . , a p , k , k , . . . ,
k
s
s
s
+ ++1
++k+s1
b1 , . . . , bq ,
,
,...,
;
k+s
k+s
k+s
7. Chapter 5 Solutions
;
;
x
F
F
0 1
0 1
a;
b;
1 1
1
1
1
a + b, a + b ;
2
2
2
2
2
x = 2 F3
4x .
a, b, a + b 1;
=
=
xn+k
(a)k (b)n k!n!
n,k=0
n
X
X
xn
n=0 k=0
n
X
X
(1 b n)k (n)k
xn
(a)k k!
(b)n n!
n=0 k=0
n,
1
n;
X
xn
1
=
F
.
(b)n n!
n=0
a;
=
X
(a + b 1)2n xn
0 F1 (; a; x)0 F1 (; b; x) =
(b) (a)n (a + b 1)n n!
n=0 n
a+b1
X
2
n
=
(a)n (b)n (a + b 1)n n!
a+b
2n
n=0
2 xn
n
2
a+b a+b1
,
;
2
= 2 F3 F 2
4x .
a, b, a + b 1;
Problem 2. Show that
1 3
, ;
Z t
4 4
1
1
1
1
2
2
x 2 (t x) 2 [1 x (t x) ] 2 dx = t2 F1
2
0
1;
t4
.
16
k k ss ctk+s
(k + s)k+s
Now
41
1
; ; x2 (t x)2 dx,
2
0
3
1
1
so in Theorem 12 we pput |alpha = , = , p = 1, q = 0, a1 = , c = 1, k =
2
2
2
2, s = 2. The result is
1 3 5 1 3
, , , , ;
2 4 4 4 4
3 1
22 22 t2+2
,
A=B
,
t 5 F4
4
2 2
4
2 3 4 5
, , , ;
4 4 4 4
or
1 3
1 3
3
1
4, 4;
4, 4;
2
2
4
t
t4
A=
t 2 F1
= t2 F1
,
(2)
2
16
16
1, 1;
1;
Z
A=
x 2 (t x) 2 1 F0
as desired.
Problem 3. With the aid of Theorem 10, show that
(1 + 21 a)
cos 21 a(1 a)
=
(1 + a)
(1 21 a
and that
sin (b 12 a)(b 21 a)
(1 + a b)
.
=
sin (b a)(b a)
(1 + 12 a b)
a, b, c;
cos 21 a sin (b 12 a) (1 a)(b 21 a)(1 + a c)(1 + 21 a b c)
1 =
.
3 F2
sin (b a)
(1 21 a)(b a)(1 + 12 a c)(1 + a b c)
1 + a b, 1 + a c;
2 cos a sin a
2
2
=
2 sin a
a 2
= cos
.
2
a
1
1
1
cos
sin b a (1 a) b a (1 + a c) 1 + a b c
a, b, c;
2
2
2
2
1
F
=
,
3 2
1
1
sin (b a)
1 + a b, 1 + a c;
1 a (b a) 1 + a c (1 + a b c)
2
2
1
1
so long as a, a, b a, b a are not integers and the gamma functions involved
2
2
have no poles.
But now we have arrived at an identity (for non-integral values of certain parameters) which has the property that both members are well-behaved if a is a negative
integer or zero. It follows that the identity continues to be valid for a = n, n a
a
= 0.
nonnegative integer. If a = (2n + 1), an odd negativer integer, cos
2
Problem 4. Use the result in Exercise 3 to show that if n is a non-negative integer,
2n, , 1 2n;
(2n)!()n ( )n
1 =
.
3 F2
n!()2n ()n
1 2n, ;
Solution 4. If a = 2n in the identity of Exercise 3 above, and if we chose
b = , c = 1 2n , we obtain
2n, , 1 2n;
cos(n) sin ( + n) (1 + 2n)( + n)(1 2n 1 + + 2n)(1 n 1 + + 2n)
1 =
3 F2
sin ( + 2n)
(1 + n)( + 2n)(1 n 1 + + 2n)(1 2n 1 + + 2n)
1 2n, ;
(1)n (1)n sin (2n)!()n ()( + n)
=
sin
n!()2n ( + n)( )
(2n)!()n ( )n
=
,
n!()2n ()n
as desired.
Problem 5. With the aid of the formula in Exercise 4 prove Ramanujans theorem:
, ;
;
;
x2
x
x =2 F3
.
F1
1 F1
4
1 1
1
;
;
1
, , + ;
2 2
2
Solution 5. Consider the product
1 F1 (; ; x)1 F1 (; ; x)
X
n
X
(1)k ()k ()nk xn
()k ()nk k!(n k)!
n=0 k=0
X
n
X
(n)k ()k (1 n)k ()n xn
=
k!()k (1 n)k n!()n
n=0 k=0
n, , 1 n;
X
()n xn
1
=
.
3 F2
n!()n
n=0
, 1 n;
Since the product of the two 1 F10 s is an even function of x, we may conclude that
3 F2
2n 1, , 1 2n 1;
1 =0
, 1 2n 1;
43
and that
1 F1 (; ; x)1 F1 (; ; x)
3 F2
2n, , 1 2n;
, 1 2n;
(2n)!()n ( )n ()2n x2n
=
,
n!()2n ()n
(2n)!()2n
n=0
n=0
()2n x2n
1
(2n)!()2n
, ;
1 F1 (; ; x)1 F1 (; ; x)
= 2 F3
1 1
1
, , + ;
2 2
2
x
4
Problem 6. Let n =3 F2 (n, 1an, 1bn; a, b; 1). Use the result in Exercise 3
to show that 2n+1 = 0 and
2n =
, , ;
cos a
2 sin ( 2 ) (1 )( 2 )(1 + )(1 + 2 )
1 =
.
3 F2
sin ( )
(1 2 )( )(1 + 2 )(1 + )
1 + , 1 + ;
Consider
n n = 3 F2 (n, 1 a n, 1 b n; a, b; 1). We wish to use = n, but
cos
= 0 for n odd. Hence 2n+1 = 0 and
2
2n = 3 F2 (2n, 1 a 2n, 1 b 2n; a, b; 1).
Therefore in the result from Exercise ?? we put = 2n, = 1 a 2n, =
1 b 2n, and thus obtain
n
so that
n =
1
1
1
(a + b 1), (a + b), (a + b + 1);
3
3
3
0 F2 (; a, b; t)0 F2 (; a, b; t) =3 F8
1
1 1 1
1 1
1
a, b, a + , b, b + , (a + b 1), (a + b);
2
2 2 2
2 2
2
27t2
64
= 0 F2 (; a, b; t)0 F2 (; a, b; t)
X
n
X
=
n=0 k=0
X
n
X
tn
(n)k (1 a n)k (1 b n)k
k!(a)k (b)k
n!(a)n (b)n
n=0 n=0
n
X
t
=
n
n!(a)
n (b)n
n=0
X
2n t2n
=
,
(2n)!(a)2n (b)2n
n=0
we have
(t)
X
(1)n 33n a+b1
t
3
n 3
n 3
n
=
a+1
b+1
2n b
2n a+b1
2n a
2
2
n!(a)
(b)
2
n
n
2 n
2
2 n
2
2
n=0
n
n
n
=
,
a+b
2
n
or
F
(1,
a,
b;
x)
F
(;
a,
b;
x)
=
F
0 2
0 2
3 8
1
3 (a
1
1
+ b 1), (a + b), (a + b + 1);
3
3
27t2
64
k!(n k)!()k
3 F2
k, b, c;
3 F2
n!()n
1 + b k, 1 + c k;
1
1
1
n, n + , 1 n;
2
2
2
1 + b n, 1 + c n;
4x
(1 x)2
45
b, c;
Solution 8. Let = 2 F1
;
=
=
=
=
X
( b)n ( c)n tn [(1 x) + xt]n
()n n!
n=0
X
n
X
( b)n ( c)n (1 x)nk xk tn+k
k=0 k=0
[n
X
2]
X
n=0 k=0
[n
X
2]
X
k!(n k)!()n
( b)nk ( c)nk (1 x)n2k xk tn
k!(n 2k)!()nk
or
3 F2
n=0
n n1
,
, 1 n;
2
2
( b) ( c) (1 x)n tn
n
n
4x
.
n!()
n
(1 x)2
1 + b n, 1 + c n;
But also, since 1 t(1 x + x) = (1 t)(1 + xt),
b, c;
= (1 t)b+c (1 + xt)b+c 2 F1
t(1 x + xt) .
;
Hence
= (1 t)b+c (1 + xt)b+c
= (1 t)b+c1 (1 + xt)b+c
X
(b)n (c)n tn [1 x(1 t)]n
n!()n
n=0
n
X
k
X
(1) (b)n (c)n xk (1 t)k tn
n=0 k=0
k!(n k)!()n
or
X
(1)k (b)n+k (c)n+k xk (1 t)x tn+k
k!n!()n+k
n,k=0
b + k, c + k;
X
1
= (1 + xt)b+c (1 t)b+c
2 F1
+ k;
k=0
= (1 t)b+c (1 + xt)b+c
Now
2 F1
b + k, c + k;
b, c;
t = (1 t)bck 2 F1
+ k;
t .
+ k;
Therefore
= (1 + xt)b+c
2 F1
b, c;
+ k;
k
k n+k
(1)
(b)
(c)
k
k ( b)n ( c)n x t
= (1 + xt)b+c
k!n!()n+k
n,k=0
n
X X (1)s (b)s (c)s ( b)ns ( c)ns xs tn
= (1 + xt)b+c
s!(n s)!()n
n=0 s=0
n, b, c;
X
(1)n ( b c)n xn tn X
=
3 F2
n!
n=0
n=0
1 + b n, 1 + c n;
k=0
x ,
or
=
n
X
X
n=0 k=0
3 F2
k, b, c;
nk
(1)
x
1 + c b, 1 + c x;
a1 , . . . , ap ;
L tcp Fq
b1 , . . . , bq ;
1 + c, a1 , . . . , ap ;
(1 + c)
zt =
Fq
s1+c p+1
b1 , . . . , b q ;
a1 , . . . , a p ;
L tc p Fq
b1 , . . . , bq ;
z
.
s
(m + 1)
. Then,
sm+1
X
(a1 )n . . . (ap )n z n
zt
=
L {tn+k }
(b
)
.
.
.
(b
)
n!
1 n
q n
n=0
(a1 )n . . . (ap )n z n
(1 + c)n (1 + c)
n+k+1
(b
)
.
.
.
(b
)
n!s
1
1 n
p n
n=0
1 + c, a1 , . . . , ap ;
z
F (1 + c)
=
.
p+1 Fq
5
s1+c
b1 , . . . , b q ;
a1 , . . . , ap ;
1
L1
Fq+1
sp
s + 1, b1 , . . . , bq ;
z =p Fq+1
a1 , . . . , ap ;
z(1 e1 ) .
1, b1 , . . . , bq ;
It follows that
L 1
F
s p q+1
z = p Fq+1
a1 , . . . , ap ;
1 + s, b1 , . . . , bq ;
sk
(s z)k+1
47
a1 , . . . , ap ;
z(1 et ) .
1, b1 , . . . , bq ;
k + 1;
zt
= 1 F1
1;
1
1
1
s
=
= 1 F0
(s z)k+1
s (1 54 )k+1
s
By Exercise 9 with c = 0,
k + 1;
(1)
1
L
1 F0
s
;
= L 1
k + 1;
(1)
z
.
5
1, k + 1;
2 F1
k + 1;
zt
= t0 F
1;
k + 1;
zt .
=F
1;
1;
d
Fq
dz p
p
Y
a1 , . . . , ap ;
z =
b1 , . . . , bq ;
m=1
q
Y
am
p Fq
bj
a1 + 1, . . . , ap + 1;
z .
b1 + 1, . . . , bq + 1;
j=1
Solution 12.
a1 , . . . , ap ;
d
p Fq
dz
b1 , . . . , b q ;
X
(a1 )n . . . (ap )n z n1
(b1 )n . . . (bq )n (n 1)!
n=1
X
(a1 )n+1 . . . (a0 )n+1 z n
(b1 )n+1 . . . (bq )n+1 n!
n=0
a1 + 1, . . . , ap + 1;
a1 . . . ap
=
p Fq
b 1 . . . bq
b1 + 1, . . . , bq + 1;
z .
Problem 13. In Exercise 19 page 71, we found that the complete elliptic integral
of the first kind is given by
1
1 1
2
K(k) = 2 F1
, ; 1; k .
2
2 2
Show that
t
p
K( x(t x))dx = arcsin
1
t .
2
1 1
2
, ; 1; k .
2 2
Now consider
Z
A=
p
K( x(t x))dx.
1 1
, ; 1; x(t x) dx
2 2
0
1 1 1 1
, , , ;
2 2 1 1
= B(1, 1)t21 4 F3
2 3
1, , ;
2 2
(1)(1)
1 1 3 t2
=
t 2 F1
, ; ;
2 (2)
2 2 2 4!
2
1 1 3 t
t
, ; ;
=
2 F1
2
2 2 2 2
t
= arcsin
,
2
A =
2 F1
1 1t2
1
22
8. Chapter 6 Solutions
1
J2n+1 (x) =
2
n=0
J0 (y)dy.
0
Solution 1.
J2k+1 (x)
k=0
X
(1)n ( x2 )2n+2k+1
n!(n + 2k + 1)!
k,n=0
X
n
X
(1)nk ( x )2n+1
2
n=0 k=0
X
n
X
(n k)!(n + k + 1)!
n, 1;
X
(1)n ( x2 )2n+1
1
=
2 F1
n!(n + 1)!
n=0
n + 2;
n x 2n+1
X
(1) ( 2 )
=
(2n + 1)n!n!
n=0
Z xX
(1)n ( y2 )2n
1
dy
=
2 0 n=0
n!n!
Z x
1
J0 (y)dy.
=
2 0
=
Problem 2. Put the equation of Theorem 39, page 113, into the form
X
1
(A)
exp z(t t1 ) = J0 (z) +
Jn (z)[tn + (1)n tn ].
2
n=1
Use equation (A) with t = i to conclude that
cos z = J0 (z) + 2
k=1
sin z = 2
k=0
X
X
X
1
1
n
n
Jn (z)t =
Jn (z)t + J0 (z) +
Jn (z)tn .
exp z(t t ) =
2
n=
n=
n=1
Now Jn (z) = (1)n Jn (z). Hence
X
X
z
1
exp (t
=
Jn (z)E n + J0 (z) +
Jn (z)T tn
2
t
n=1
n=1
X
n
= J0 (z) +
Jn (z)[t + (1)n tn ].
n=1
1
exp
i
= exp(iz) = cos z + i sin z.
2
i
49
Therefore
cos z + i sin z = J0 (z) + 2
in Jn (z).
n=1
(1)n in Jn (z).
n=1
(note: above argument can be done more easily by equating even and odd function
of z) Thus we get
cos z = J0 (z) + 2
n=1
n=1
and
i sin z = 2
n=0
or
sin z = 2
n=0
Problem 3. Use t = e
k=1
sin(z sin()) = 2
k=0
X
J0 (z) +
Jn (z)[tn + 9 1)n tn ] we thus obtain
n=1
n=1
Now equate even function of z on the two sides, then odd functions of z on the two
sides to get
X
cos(z sin ) = J0 (z) + 2
J2k (z) cos(2k)
k=1
and
sin(z sin ) = 2
X
k=0
51
Problem 4. Use Bessels integral, page 114, to obtain for integral n in the relations
Z
2
(B)
[1 + (1)n ]Jn (z) =
cos(n) cos(z sin())d,
0
Z
2
n
(C)
[1 (1) ]Jn (z) =
sin(n) sin(z sin())d.
0
With the aid of (B) and (C) show that for integral k,
Z
1
J2k (z) =
cos(2k) cos(z sin(thetaA))d,
0
Z
1
J2k+1 (z) =
sin(2k + 1) sin(z sin())d,
0
Z
cos(2k + 1) cos(z sin()d = 0,
0
Z
sin(2k) sin(z sin())d = 0.
0
Then
Jn (z) =
2
[1 + (1) ]Jn (z) =
Z
2
sin(n) sin(z sin )d.
0
Use (B) with n = 2k and (C) with n = 2k + 1 to obtain
Z
1
J2k (z) =
cos(2k) cos(z sin )d,
0
Z
1
J2k+1 (z) =
sin(2k + 1) sin(z sin )d.
0
Use (B) with n = 2k + 1 and (C) with n = 2k to obtain
Z
cos((2k + 1)) cos(z sin )d = 0,
(C)
Problem 5. Expand cos(z sin()) and sin(z sin()) in Fourier series over the interval < < . Thus use Exercise 4 to obtain in another way the expansions
in Exercise 3.
X
1
a0 +
(an cos(n) + bn sin(n)),
2
n=1
in which
Z
1
f () cos(n)d,
Z
1
bn =
f () sin(n)d.
Consider first f () = cos(z sin ), an even function of . For this function
Z
2
an =
cos(n) cos(z sin )d, bn = 0.
0
an =
k=1
k=0
1
1
x(t t1 ) by exp x(t t1 ) , obtain
2
2
the coefficient of t0 and thus show that
X
J02 (x) + 2
Jn2 (x) = 1.
n=1
1
For real x conclude that |J0 (x)| 1 and |Jn (x)| 2 2 for n 1.
Solution 6. We know that
Z
1
exp
t
=
Jn (x)tn
2
t
n=
and thus that
Z
1
exp
t
=
(1)k Jk (x)Jn (x)tn+k .
2
t
n=
The coefficient of t0 on the right is (k = n)
n=
n=1
53
Jn2 (x) = 1.
n=1
m
X
dm
J
(z)
=
(1)mk Cm,k Jn+m2k (z),
n
dz m
n=0
d
, from (8), page 100,
dz
m
X
k=0
=
=
=
m
X
k=0
m+1
X
m
X
k=1
k=0
m
X
(1)m+1k [Cm,k + Cm,k1 ]Jn+m+12k (z) + (1)m+1 Jn+m+1 (z) + Jnm1 (z).
k=1
Now Cm,k + Cm,k1 = Cm+1,k (Pascals triangle), so that (using the last two terms
also),
m+1
X
2m+1 D m+1 Jn (z) =
(1)m+1k Cm+1,k Jn+m+12k (z),
k=0
4x .
a, b, a + b 1;
Then
z n z m
Jn (z)Jm (z)
z2
z2
2
2
)0 F1 (; m + 1; )
0 F1 (; n + 1;
(n + 1)(m + 1)
4
4
z n+m
n+m+2 n+m+1
,
;
2
2
2
=
2 .
2 F3
z
(n + 1)(m + 1)
n + 1, m + 1, n + m + 1;
=
Problem 10. Start with the power series for Jn (z) and use the form (2), page 18,
of the Beta function to arrive at the equation
Z 2
2( 21 z)n
Jn (z) =
sin2n cos(z cos )d,
( 12 )(n + 21 ) 0
1
for R(n) > .
2
Solution 10. We know that
1
z 2k+n
(1)
k 2k+n
X
X
2
(1) z
k
Jn (z) =
=
.
22k+1 k!(k + n + 1)
2n (2k)!(k + n + 1)
k
k=0
k=0
Also
1
2 k
(k + n + 1)
1
1
n+
k+
2
2
=
1
1
(k + n + 1) n +
2
2
1
1
B k + ,n +
2
2
=
1
1
n+
2
2
Z 2
2
=
cos2k sin2n d.
1
1
0
n+
2
2
Therefore
z n
2
Jn (z) = 2
1
n+
2
z n
2
= 2
1
n+
2
Z
1
2
1
2
d
1 du
0 F1 (; a; u) =
0 F1 (; a + 1; u). Since
dx
a dx
z n
(1)
Jn (z) =
z2
2
0 F1 ; 1 + n;
(1 + n)
4
we obtain
d n
[z Jn (z)]
dz
1
1 z
z2
F
;
2
+
n;
0
1
2n (1 + n) 1 + n
2
4
z n+1
z2
= z n 2
0 F1 ; 2 + n;
(2 + n)
4
= z n Jn+1 (z),
0 F1
1 + ;
2xt t2
t 2x
t
12
p
X
J+n (t)xn
J ( t2 2xt) =
.
n!
n=0
55
1
Solution 12. Consider 0 F1 ; 1 + ; (2xt t2 ) . We obtain
4
X
tn (2x t)n
2xt t2
F
=
=
0 1
2n
2 (1 + )n n!
4
n=0
1 + ;
X
n
X
(1)k (2k)n tn+2k
=
2n+2k
w
(1 + )n+k k!n!
n=0 k=0
X
X
tn (2x)n
(1)k t2k
=
22k k!(1 + + n)k 22n n!(1 + )n
n=0 k=0
n
t
xn (1 + )
X
t2
2
=
.
0 F1 ; 1 + + n;
4
n!(1 + + n)
n=0
Now
+n
t
t2
2
J+n (t) =
,
0 F1 ; 1 + + n;
( + n + 1)
4
so we obtain
;
0 F1
1 + ;
2xt t
4
= (1 + )
X
Jn+ (t)xn
t
2
n!
n=0
or
t2 2xt
2
!
(1 + )J
or
t 2x
t
2
p
X
Jn+ (t)xn
t
,
t2 2xt = (1 + )
2
n!
n=0
p
X
Jn+ (t)xn
.
J ( t2 2xt) =
n!
n=0
Problem 13. Use the realtions (3) and (6) of Section 60 to prove that: For real
x, between any two consecutive zeros of xn Fn (x), there lies one and only one zero
of xn Fn+1 (x).
Solution 13. We are given that
d n
[x Jn (x)] = xn Jn1 (x),
dx
d n
x Jn (x) = xn Jn+1 (x).
dx
We know that Jn (x) has exactly n zeros at x = 0. Let the others (we have proved
there are any) on the axis of reals be at 1,n , 2,n , . . . . The curve y = xn Jn (x) has
its real zeros only at the s. By Rolles theorem we see that the zeros 1,n1 , 2,n2 , . . .
of
y 0 = xn Jn+1 (x)
are such that an odd number of them lie between each two consecutive s. The
curve
y2 = xn+1 Jn+1 (x)
57
In (z)
z n
2
z2
=
0 F1 ; 1 + n;
(1 +
n)
4
2 k
n
z
X
1
z
2
=
(1 + n)
(1 + n)k k! 4
k=0
X
z 2k+n
.
=
22k+n k!(k + n + 1)
k=0
X
k=0
= zn
=z
z 2k+n2n1
+ k)
22k+n1 k!(n
k=0
X
k=0
z 2k+n1
22k+n1 k!(n + k)
z 2k+n1
+ (n 1) + 1)
22k+n1 k!(k
= z n In1 (z),
or
z n In0 (z) + nz n1 In (z) = z n In1 (z),
which is equivalent to
zIn (z) = zIn1 (z) nIn (z),
which is (1).
Similarly,
d n
[z In (z)]
dz
=
=
d X
z 2k
dz
22k+n k!(k + n + 1)
k=0
X
z 2k1
k=1
22k+n1 (k 1)!(k + n + 1)
z 2k+1
22k+n+1 k!(k + n + 1 + 1)
k=0
n
= z In+1 (z),
=
and
z n In0 (z) nz n1 In (z) = z n In+1 (z),
or
zIn0 (z) = nIn (z) + zIn+1 (z),
which is (2).
Adding (1) and (2):
2zIn0 (z) = zIn1 (z) + zIn+1 (z)
or
2In0 (z) = In1 (z) + In+1 (z),
which is (3).
Equating the right sides of (1) and (2):
zIn1 (z) nIn (z) = zIn+1 (z) + nIn (z),
or
2nIn (z) = z[In1 (z) In+1 (z)],
which is (4).
Problem 15. Show that In (z) is one solution of the equation
z 2 w00 + zw0 (z 2 + n2 )w = 0.
Solution 15. (Solution by Leon Hall) Because In is a 0 F1 function times z n , we
know from Section 46 that u = 0 F1 (; b; y) is a solution of
y
d2 y
du
+b
u = 0,
dy 2
dy
z2
and so 0 F1 ; 1 + n;
is a solution of
4
d2 u
du
zu = 0.
+ (2n + 1)
dz 2
dz
Thus is w = z n u, In (z) will be a solution of
z
59
1
Problem 16. Show that, for Re(n) > ,
2
n
Z 2
1
2 2z
sin2n cosh(z cos())d.
In (z) =
12 n + 12 0
Solution 16. (Solution by Leon Hall) For n not a negative integer,
In (z) = in Jn (iz),
and using the result of Problem 10,
In (z) = in
2( 12 iz)n
1
( 2 )(n + 12 )
1
2
1
as desired.
2
Problem 17. For negative integral n define In (z) = (1)n In (z), thus completing
the definition in Section 65. Show that In (z) = (1)N In (z) and that
"
#
X
1
n
1
exp z(t + t ) =
In (z)t .
2
n=
Solution 17. (Solution by Leon Hall) We have
In (z)tn
n=
1
X
(1)n In (z)tn +
n=
In (z)tn
n=0
(1)
n+1
In+1 (z)t
n1
In (z)tn .
n=0
n=0
X
1
In (z)tn = exp z(t + t1 ) .
2
n=
Problem 18. Use the integral evaluated in Section 56 to show that
Z tp
p
n+ 1
t
n
n
2
1
[ x(t x)] Jn ( x(t x))dx = 2
t
Jn+ 2
.
2
0
Solution 18.
Z
0
p
p
[ x(t x)]n Jn ( x(t x))dx =
1
2n (1
+ n)
xn (tx)n 0 F1
1 + n;
x(z x)
dx.
4
k = 1, s = 1. The result is
,
n + 1, n + 1;
Z tp
t2
p
1
0
2n + 2 2n + 3
1 + n,
,
;
2 2
t2
(1 + n)(1 + n) 2n+1
= n
t
0 F1
16
2 (1 + n)(2 + 2n)
3
n+ ;
2
;
n+ 12
2
3
t
1
t
tn+ 2
(1 + n) n +
2n+1
2
r
2
F
=
0 1
3
4
(2 + 2n) n +
3
2
n+ ;
2
3
(1 + n)
+n
t
2
1
= 2n+1 tn+ 2
Jn+ 12
.
(2 + 2n)
2
1
2z1
2
(z) z +
2
, we get
By Legendres duplication formula, (2z) =
3
1+2n
2
(1 + n)
+n
2
(2 + 2n) =
.
Hence
Z
p
p
[ x(t x)]n Jn ( x(t x))dx
1
2n+1 tn+ 2
t
1
J
n+ 2
1+2n
2
2
n+ 1
t
n
=2
t 2 Jn+ 12
.
2
=
2
c1 12 n
(1 x) x Jn ( x)dx = (c)
Jn+c ().
0
Solution 19. Consider
Z
1 x sin( x)dx.
We know that
3 z2
sin z = z 0 F1 ; ;
.
2
4
61
Hence
1
1 x sin( x)dx =
1
1
3 2 x
x 2 (1 x) 2 0 F1 ; ;
dx
2
4
2
, k = 1, s = 0. We get
4
3
3
3
, = , p = 0, q = 1, b1 = , c =
2
2
2
1 x sin( x)dx
1
(1 x)c1 x 2 n Jn ( x)dx =
0
2
= B
1 F
4
3 3
, ;
2 1
3
3
2
2
2
=
0 F1 ; 3;
(3)
4
3
;
2
( 2 )n
(1 + n)
3 3
,
2 2
1
c1 n
(1 x)
x 0 F1
2 x
; 1 + n;
4
dx
n + 1;
Z 1
n
2
(2)
B(n + 1, c)F
(1 x)t1 xtn Jn ( x)dx =
1
(1
+
n)
4
0
1 + n,c + n + 1;
( 2 )n
(n + 1)(c)
2
F
=
;
c
+
n
+
1;
0 1
(1
+n) (n + c + 1)
4
c
= (c)
Jn+c (),
2
as desired.
Problem 20. Show that
Z t
Z t
exp[2x(t x)]I0 [2x(t x)]dx =
exp( 2 )d.
0
Use a =
1
and z = 2x(t x) to get
2
exp[2x(t x)]I0 [2x(t x)] = 1 F1
1
; 1; 4x(t x) .
2
Then,
t
Z
exp[2x(t x)]I0 [2x(t x)]dx =
1 F1
1
; 1; 4x(t x) dx,
2
1 1 1
, , ;
2 1 1
Z t
t2
exp[2x(t x)]I0 [2x(t x)]dx = B(1, 1)tF
4
4
0
2 3
1, , ;
2 2
1 3
2
= t 1 F1
; ; t
2 2
1
n
t2n+1
(1)
X
2 n
=
3
n=0
n!
2 n
X
(1)n tn+1
=
n!(2n + 1)
n=0
Z tX
(1)n 2n
=
d
n!
0 n=0
Z t
=
exp( 2 )d.
1
, b1 =
2
1
1
, = , p = q = 0, k = 1, s = 1, c = 4 :
2
2
1 1
, ;
2 2
1 1 0
4t2
exp[4x(t x)]dx = B
,
t F
2 2
4
1 2
, ;
2 2
( 12 )( 12 )
1
=
F
; 1; t2
1 1
(1)
2
1 2
t4
= exp
t 0 F1 ; t;
16
22 2
t
t
= exp
I0
.
2
2
Z
0
[x(t x)] 2
63
X
(n + 2k)(n + k 1)!Jn+2k (z)
k!
k=0
, n 1.
X
(n + 2k)(n + k 1)! z n
k!
k=0
Then
d z n
Jn+2k (z)
dz
2
Jn+2k (z).
n z n
0
Jn+2k
(z)
Jn+2k (z)
2 2
z2 n h
i
n
0
Jn+2k
(z) Jn+2k (z)
=
2
z
( z2 )n
n(n + 2k)
0
0
nJn+2k (z) + 2kJn+2k (z)
Jn+2k (z).
=
n + 2k
z
=
z n
Using (8), Section 60, and (1), Section 61, and simplifying gives
( z )n
d z n
Jn+2k (z) = 2
[kJn+2k1 (z) (n + k)Jn+2k+1 (z)] .
dz
2
n + 2k
So
Fn0 (z)
#
"
z n X
X
(n + k)!
k(n + k 1)!
=
Jn+2k1 (z)
Jn+2k+1 (z).
2
k!
k!
k=0
k=0
Note that the k = 0 term in the first series is zero, and so shifting the index in the
first series yields the second series.
Thus, Fn0 (z) = 0, making Fn (z) = constant. From the structure of the Bessel
functions, we see that
z n n! z n
Fn (0) =
=1
2
n! 2
and so Fn (z) = 1, from which Neumanns expansion immediately follows.
Problem 23. (Solution by Leon Hall)
Prove Theorem 39, page
113, by
forming
1 1
1
zt and the series for exp zt
.
the product of the series for exp
2
2
Solution 23. Theorem 39 is: for t 6= 0 and for all finite z,
X
z
1
exp
t
=
Jn (z)tn .
2
t
n=
We know
exp
z X
zn n
t =
t
2
2n n!
n=0
and
0
z
X
X
(1)m z m m
(1)n z n n
exp t1 =
t
=
t .
2
2m m!
2n (n)!
n=
m=0
!
an t
0
X
bn t
, the coefficient of tn , n 0, is
n=
n=0
given by
an+k bk
k=0
ak bn+k .
k=0
z
hz
z
i
t exp t1 , or exp
t 1t , the coefficient of
Thus, in the product exp
2
2
2
tn for n 0 is:
X
k=0
z k+n
2k+n (k + n)!
(1)k z k
2k k!
(1)k z 2k+n
22k+n k!(k + n)!
k=0
= Jn (z),
=
X
X
z
(1)k+n z k+n
(1)k z 2k+n
n
=
(1)
2k k!
2k+n (k + n)!
22k+n k!(k + n)!
k=0
k=0
n
= (1) Jn (z)
= Jn (z).
9. Chapter 7 Solutions
exp(t2 )dt
1 3
2
; ; x .
2 2
Solution 1. Let
2
erf x =
Then,
Z
0
exp(t2 )dt.
erf x
=
=
=
=
t2n dt
(1)
2 X
0
n!
n=0
2 X (1)n x2n+1
n=0 n!(2n + 1)
1
2x X (1)n 2 n x2n
n=0
n! 23 n
2x
1 3
1 F1
; ; x2 .
2 2
(, x) =
0
Show that
(, x) = 1 x 1 F1 (; + 1; x).
Solution 2. Let
Z
(, x) =
0
Then,
Z
(, x)
=
0
xX
(1)n tn+1
n!
n=0
X
(1)n xn+
.
=
n!( + n)
n=0
Now, ( + n) =
( + 1)n
. Hence
()n
(, x) = 1 x
X
(1)n ()n xn
= 1 x 1 F1 (; + 1; x).
n!(
+
1)
n
n=0
Solution 3. Let D =
d
. Consider D[ez 1 F1 (a; b; z)]. We know that
dx
1 F1 (a; b; z)
Hence
= ez 1 F! (b a; b; z).
65
We used D 1 F! (a; b; z) =
= D k 1 F1 (b a; b; z)
(b a)k (1)k
=
1 F! (b 1 + k; b + k; z)
(b)k
(1)k (b a)k z
=
e 1 F! (a; b + k; z).
(b)k
a
1 F! (a + 1; b + 1; z) k times.
b
Then,
1 F1 (a; b; z)
=
=
X
(a)n z n
n!(b)n
n=0
1 X (a + n)z n
z2
(a)z n
F
.
a;
n
+
1;
1 1
2n+1 (n + 1)
4
Solution 5. We obtain
Z
A=
0
Z
0
2
et t2an1 z n tn
z 2 t2
;
1
+
n;
F
dt.
0 1
2n (1 + n)
4
Put t = . Then
Z
zn
1 a1
z2
A = n
e
d
0 F1 ; 1 + n;
2 (1 + n) 2 0
4
n
2
z
(a)
z
= n+1
,
1 F! a; 1 + n;
2
(1 + n) 1
4
as desired.
Problem 6. If k and n are non-negative integers, show that
k, + n;
;k > n
0
(n)k
1 =
F
; 0 k n.
;
()k
k
X
(k)s ( + n)s
s!()s
s=0
k
X
s=0
(1)s k!()n+s
.
s!(k s)!()s ()n
Hence
=
=
X
V (k, n)tk
k=0
X
k
X
k=0 s=0
k!
(1)s ()n+s tk
s!(k s)!()s ()n
X (1)s ( + n)s ts
= et
s!()s
s=0
= et 1 F! ( + n; ; t)
= et et 1 F! (n; ; t)
n
X
(n)k tk
=
.
k!()k
=
k=0
Thus
(n)k
V (k, n) =
()k
;0 k n
; k > n.
n=0
n (xy) =
n
X
y k (1 y)nk k (x)
(n k)!
k=0
X
n
1
k (x)
x =
.
2
(n k)!
k=0
Solution 1. Let
et (xt) =
X
n=0
Then
n (x)tn .
67
e (xyt) =
n (xy)tn
n=0
and
eyt (xyt) =
n (x)y n tn .
n=0
But
et (xyt) = e(1y)t eyt (xyt)
so that
n (xy)t
!
!
X
X
(1 y)n tn
n n
n (x)y t
=
n!
n=0
n=0
X
n
X
(1 y)nk y k k (x) n
t .
=
(n k)!
n=0
n=0
k=0
Hence
n (xy) =
n
X
y k (1 y)nk k (x)
(n k)!
k=0
For y =
1
, we obtain
2
2 n
X
n
k (x)
1
x =
.
2
(n k)!
k=0
n (x)tn .
n=0
(A)
n (x)tn .
n=0
n0 (x)tn .
n=0
Hence
X
n=0
or
n (x)tn+1 =
X
n=0
n0 (x)tn ,
n1 (x)tn =
n=1
69
n0 (x)tn .
n=0
n (x)tn
n=0
and wish to apply Theorem 50, page 239. In the notation of Theorem 50 we have
A(t)et , H(t) = t, (t) =
n tn , 0 6= 0.
n=0
n
X
1
t
,, so an = ; H(t) = t, so h0 = 1, hn = 0 for n 1.
Now A(t) = et =
n!
n
n=0
Then also
X
tet
tA0 (t)
= t =t=
n tn+1 ,
A(t)
e
n=0
so that 0 = 1, n = 0 for n 1.
Furthermore,
X
tH 0 (t)
t1
=
=1=1+
n tn+1
H(t)
t
n=0
et (xt) =
n (x)tn ,
n=0
(1 t)c F
xt
1t
=
(c)n n (x)tn ,
n=0
for a certain function F . By applying Theorem 50, page 141, to (B), conclude
that the n (x) of (A) satisfy the relation
(c)n [xn0 (x) nn (x)] =
n1
X
k=0
for arbitrary c.
(1 t)
(B)
F
xt
1t
=
(c)n n (x)tn ,
n=0
where
F (u) =
(c)n n un
n=0
in terms of (u) =
n un .
n=0
t
.
1t
Then
log A(t) = c log(1 t),
H(t) = 1 +
1
1
; H 0 (t) =
,
1t
(1 t)3
c
A0 (t)
=
,
A(t)
1t
X
tH 0 (t)
t
1t
1
tn .
=
=
=
1
+
H(t)
(1 t)2 t
1t
n=1
Hence
X
tH 0 (t)
=1+
tn+1 , so n = 1 for n 0.
H(t)
n=0
Thus Theorem 50 yields
n1
X
k=0
n1
X
0
1(c)n1k n1k
(x),
k=0
n1
X
k=0
Problem 5. Apply Theorem 50, page 141, to the polynomials yn (x) defined by (1),
page 135. You do not, of course, get Theorem 47, since that theorem depended upon
the specific character of the exponential.
71
xt
1t
=
yn (x)tn .
n=0
1
t
=1
1t
1t
from which
H 0 (t) =
X
1
tH 0 (t)
1t
t
1
,
=
=
1
+
tn .
(1 t)2 H(t)
(1 t)2 t
1t
n=1
so
tA0 (t) X
=
n tn+1
A(t)
n=0
and
X
tH 0 (t)
=1+
tn+1 ,
H(t)
n=0
so that n = 1 for n 0.
Therefore, we may conclude that there exist constants n such that
xyn0 (x) nyn (x) =
n1
X
k yn1k (x) x
k=0
n1
X
0
1 yn1k
(x).
k=0
n1
X
()
k=0
()
xt
1t
=
Ln() (x)tn .
n=0
t
1
=1
,
1t
1t
1
,
(1 t)2
X
tA0 (t)
(1 + )t
=
= (1 + )
tn+1 ,
A(t)
1t
n=0
X
X
1t
t
1
tH 0 (t)
n
=
=
=
1
+
t
=
1
+
tn+1 .
H(t)
(1 t)2 t
1t
n=1
n=0
Hence
n = 1 + , n 0;
n = 1, n 0.
Therefore we obtain
xDLn() (x)
nLn() (x)
= (1 + )
n1
X
()
Ln1k (x)
n1
X
k=0
()
DLn1k (x)
k=0
or
xDLn() (x) nLn() (x) =
(A)
n1
X
()
()
k=0
(B)
n1
X
()
Lk (x).
k=0
hn (x)tn .
n=0
hn (x)tn = 1 F0
n=0
3xt t3 .
xt
1t
F
F
(1 t)tA0 (t)
F
t
= t2
F
x
t
t
A(t)
xt
=
yn (x)tn .
Solution 8. Let F = A(t) exp
1t
n=0
Then
F
t
=
A(t) exp
x
1t
F
x
=
A(t) exp
t
(1 t)2
xt
1t
xt
1t
+ A0 (t) exp
xt
1t
Thus we obtain
x
F
F
A0 (t)
t(1 t)
= t(1 t)
F,
x
t
A(t)
or
x
F
F
F
tA0 (t)
t
= t2
(1 t)
F.
x
t
t
A(t)
Now let
tA0 (t) X
=
n tn+1 .
A(t)
n=0
Then
.
73
n=0
=
=
yn (x)ctn+1
n=1
!
n tn+1
n=0
nyn (x)t
n=0
(n
n=2
n+1
!
yn (x)tn
n=0
X
n
X
X
n
X
n+1
!
n tn+2
n=0
nk yk (x)t
+
nk yk (x)t
n=0 k=0
n=0 k=0
n1
n2
X
X
X
X
1)yn1 (x)tn
n1k yk (x)tn +
n2k yk (x)tn .
n=1 k=0
n=2 k=0
n2
X
k=0
or
F
F
xt F
tA0 (t)
t
=
F,
x
t
1 t x
A(t)
from which it follows that
x
[xyn0 (x)
n=0
nyn (x)]y
= x
!
n+1
n=0
n
X
X
= x
!
yn0 (x)tn
n=0
yk0 (x)tn+1
n
X
X
!
n+1
n t
n=0
nk yk (t)tn=1 .
n=0 k=0
n=0 k=0
Hence, for n 1,
xyn0 (x)
nyn (x) =
n1
X
k=0
(1 t)
A
xt
1t
=
an (x)tn
n=0
obtain what results you can parallel to those of Theorem 48, page 137.
Solution 9. Let an (x) be defined by
(1)
(1 t)
A
xt
1t
=
X
n=0
Let
(2)
yn (x)tn
n=0
n+2
Hence we get y00 (x) = 0, xy10 (x) y1 (x) = 0 y0 (x), and for n 2,
x(1 t)
A(u) =
X
n=0
n un .
an (x)tn .
X
n=0
!
n
yn (x)t
75
Put
F = (1 t)c A
xt
1t
.
Then
F
= t(1 t)c1 A0
x
F
= c(1 t)c1 A x(1 t)c2 A0 .
t
Hence
x
F
F
t(1 t)
= ctF.
x
t
F
F
F
t
= ctF t2
.
x
t
t
= c
n=0
an (x)tn+1
n=0
ngn (x)tn+1
n=0
[c + n 1]gn1 (x)tn .
n=1
Therefore,
a00 (x)
= 0 and for n 2,
xa0n (x) nan (x) = (c + n 1)an1 (x).
F
F
F
t(1 t)
= ctF xt
.
x
t
x
F
F
ct
xt F
t
=
F
x
t
1t
1 t x
[xa0n (x)
n=0
nan (x)]t
= c
= c
=
n=0
X
n
X
n=0 k=0
n1
X
X
!
n
an (x)t
n=0
n+1
ak (x)t
X
n
X
k=0 k=0
n=0 k=0
!
n+1
!
t
n+1
n=0
a0k (x)tn+1
X
n=0
!
a0n (x)tn
n1
X
k=0
v
1
t
. Then t =
and 1 t =
.
In (1), let us put v =
1t
1t
1v
Hence (1) becomes
(1 v)c A(xv) =
X
an (x)(v)n
,
(1 v)n
n=0
or
n xn v n
n=0
=
=
=
X
an (x)(1)n v n
(1 v)n+c
n=0
k!(c)n
(c)n (1)nk ank (x) n
v
k!(c)nk
(1)k (c)n ak (x) n
v ,
(c)k (n k)!
an (x)t
n=0
X
n (x)n tn
=
(1 t)n+c
n=0
X
n (1)n xn (c)n+k tn+k
=
(c)n k!
n,k=0
X
n
nk nk
X
(1)
x
nk (c)n tn
=
k1(c)nk
n=0
k=0
or
an (x)tn =
n=0
X
n
X
(1)k xk k (c)n tn
n=0 k=0
(n k)!(c)k
Therefore,
an (x) =
n
X
(1)k xk k (c)n
k=0
(n k)!(c)k
No problems in Chapter 9.
77
Problem 1. Start with the defining relation for Pn (x) at the beginning of this
chapter. Use the fact that
1
(1 2xt + t2 ) 2 = [1 (x +
x2 1)t] 2 [1 (x
x2 1)t] 2
Pn (x) =
n
X
( 1 )k ( 1 )nk (x +
2
x2 1)nk (x
k!(n k)!
k=0
x2 1)k
(1 2xt + t2 ) 2 =
Pn (x)tn .
n=0
Now
[1 (x +
x2 1)t][1 (x
Hence
X
n=0
Pn (x)tn
1
1
x2 1)t] 2 [1 (x x2 1)t] 2
!
!
X
X ( 1 )n (x x2 1)n tn
( 21 )n (x + x2 1)n tn
2
=
n!
n!
n=0
n=0
n
X
X
( 12 )k ( 21 )nk (x + x2 1)nk (x x2 1)k tn
.
=
k!(n k)!
n=0
= [1 (x +
k=0
Hence we obtain
Pn (x) =
X
( 1 )k ( 1 )nk (x +
2
k=0
x2 1)nk (x
k!(n k)!
x2 1)k
1
n; ;
2
+ x2 1)
Pn (x) =
(x x2 1)2
2 F1
.
n!
1
n;
2
x2 1)n (x x2 1)2k
k!n!(1 12 n)k
k=0
1
n, ;
1
n
2
2
( )n (x + x 1)
= 2
(x x2 1)2
2 F1
n!
1
n;
2
Problem 3. In Section 93, equation (4), page 166, is
1
1
1
n,
n
+
;
2
2
2
( 21 )n (2x)n
1
.
Pn (x) =
2 F1
n!
x2
1
n;
2
We know from Section 34 that the 2 F1 equation has two linearly independent
solution:
Pn (x)
n
X
( 1 )k (n)k ( 1 )n (x +
2
2 F1 (a, b; c; z)
and
z 1c F (a + 1 c, b + 1 c; 2 c; z).
Combine these facts to conclude that the differential equation
(1 t2 )y 00 2ty 0 + n(n + 1)y = 0
has the two linearly independent solutions y1 = Pn (t) and y2 = Qn (t), where
Qn (t) is as given in equation (4) of Section 102.
Solution 3. We know that
n n 1
,
;
2
2
( 1 )n (2x)n
1
,
Pn (x) = 2
2 F1
2
n!
x
1
n;
2
from which w = xn Pn (x) is a solution of the 2 F1 differential equation with
n
n1
1
a = ,b =
, c = n, z = x2 .
2
2
2
Then consider
d2 w
3
dw n(n 1)
1
z(1 z) 2 +
n
n z
w=0
dz
2
2
dz
4
dz
dx
1
and put z = x2 . Then,
= 2x3 ,
= x3 ,
dx
dz
2
dw
dx dw
1 dw
=
= x3
dz
dz dx
2 dx
2
2
d w
1 d w 3
1
dw
1
3
= x6 2 x2 x3
= x6 w00 + x5 w1 .
dz 2
4 dx
2
2
dx
4
4
79
(1x
1 6 00 3 5 2
3
n(n 1)
2
0 1 3 1
2
) x w + x x (1x )w x
n
n x
w = 0,
w0
4
4
2
2
2
4
or
x2 (x2 1)w00 + 3x(x2 1)w0 x[(1 2n)x2 (3 2n)]w0 n(n 1)w = 0,
or
x2 (x2 1)w00 + x[(2 + 2n)x2 2n]w0 n(n 1)w = 0.
Now put
w = xn y
w0 = xn y 0 nxn1 y
w00 = xn y 00 2nxn1 y 0 + n(n + 1)xn2 y
or
xn w = y
xn w0 = y 0 nx1 y
xn w00 = y 00 2nx1 y 1 + n(n + 1)x2 y
and the differential equation becomes
x2 (x2 1)y 00 2nx(x2 1)y 0 +n(n+1)(x2 1)y+x[(2+2n)x2 2n]y 0 n[(2+2n)x2 2n]yn(n1)y = 0,
or
x2 (x2 1)y 00 + x[2x2 ]y 0 + [n(n + 1)x2 (1 2) n2 n + 2n2 n2 + n]y = 0,
or
(1)
(1 x2 )y 00 2xy 0 + n(n + 1)y = 0.
Since the differential equation for w has the two linearly independent solutions
w1 = xn Pn (x) and
n1 1
n 1
+ + n;
+ + n;
2
2
2
2
1+n 2+n
,
;
2
2
12n
1 21 +n
z =x
w2 = z
2 F1
2 F1
3
3
+ n;
+ n;
2
2
Then the differential equation (1) for y has the linearly independent solutions
y1 = Pn (x) and
1+n 2+n
,
;
2
2
1
1n
,
y2 = x
F
2 1
x2
3
+ n;
2
which is a constant multiple of Qn (x) as given on page 314.
x2
.
Note that the whole thing could have been done much more simply by using the
properties of the Riemann P-symbol.
Problem 4. Show that
n=0
and
n
[2]
X
X
n=0 k=0
k=0
(1 2xt + t2 ) 2 =
Pn (x)tn .
n=0
Put
1
F = (1 2xt + t2 ) 2 .
Then
3
F
= t(1 2xt + t2 ) 2
x
and
3
F
= (x t)(1 2xt + t2 ) 2 .
t
Hence
3
F
F
t
= t2 (1 2xt + t2 ) 2 ,
x
t
from which it follows that
(1)
n=0
[2]
X
X
n=0 k=0
n,k=0
2 1
= (1 t )
n=0
81
0
0
(2n + 1)Pn (x) = Pn+1
(x) Pn1
(x), n 1.
Hence
n
[2]
X
X
"
(2n 4k + 1)Pn2k (x)t
0
Pn+1
(x)tn
n=2
Now
= 1 and
P00 (x)
n=0
= 0. Hence we have
[2]
X
X
#
0
Pn1
(x)tn
= (1 t2 )1 1 +
n=1
n=1
"
#
X
X
0
n1
0
n+1
2 1
= (1 t )
1+
PN (x)t
Pn (x)t
.
n=0 k=0
P10 (x)
"
(2n 4k + 1)Pn2k (x)t
2 1
= (1 t )
n=0 k=0
2 1
= (1 t )
Pn0 (x)tn1
n=0
1 F
F
t
t x
x
#
Pn0 (x)tn+1
n=0
1 F
t x
3
= (1 2xt + t2 ) 2 .
=
Now we use (1) to conclude that
[xPn0 (x)
n=0
nPn (x)]t
[2]
X
X
n=0 k=0
n2
[X
2 ]
X
n=2 k=0
Hence, for n 2,
[ n2
2 ]
xPn0 (x)
nPn (x) =
k=0
t(x 1)
t(x + 1)
Pn (x)tn
F
;
1;
F
;
1;
=
.
0 1
0 1
2
2
(n!)2
n=0
Use x = 0, t = 2y to obtain
0 F1 (; 1; y)0 F1 (; 1; y)
X
2n Pn (0)y n
.
(n!)2
n=0
(1)k ( 12 )k
. Therefore
k!
X
22k (1)k ( 21 )k y 2k
F
(;
1;
y)
F
(;
1;
y)
=
0 1
0 1
k![(2k)!]2
k=0
X (1)k 22k ( 1 )k y 2k
2
=
k!22k k!( 12 )k 22k k!( 21 )k
k=0
X
(1)k y 2k
=
22k (k!)3 ( 21 )k
k=0
1 1 2
= 0 F3 ; 1, 1, ; y ,
2 4
as desired.
Problem 6. Use Brafmans generating function, page 168, to conclude that
2 F1
c, 1 c;
1t
1 + t2 2 F1
c, 1 c;
1+t
2
1;
1 1
1 1 1
1
c, c + , c, 1 c;
2 2
2 2 2
2
t2
= 4 F3
1
1, 1, ;
2
Solution 6. Brafmans generating relation, page 287, is
c, 1 c;
2 F1
1;
1tp
2 F1
2
c, 1 c;
1;
2 F1
1 + t2
(n!)2
2
n=0
with p = (1 2xt + t2 ) 2 . We use x = 0. Note that P2n+1 (0) = 0 and P2n (0) =
(1)n ( 12 )n
. Then
n!
c, 1 c;
c, 1 c;
2
(c)2n (1 c)2n (1)n ( 21 )n t2n
1+tp
1t 1+t F
=
2 1
n![(2n)!]2
2
2
n=0
1;
1;
1c
2c
n 2n
X
( 2c )n ( c+1
2 )n ( 2 )n ( 2 )n (1) t
=
n!n!( 12 )n n!
n=0
c c+1 1c 2c
,
,
,
;
2
2
2
2
t2
= 4 F3
,
1
1, 1, ;
2
as desired.
Problem 7. Use equation (5), page 168, to obtain the results
sinn Pn (sin ) =
n
X
k=0
Pn (x) =
n
X
k=0
Pn (1 2x2 ) =
n
X
k=0
Pn (cos ) =
First use =
Thus (1) leads to
sin
sin
n X
n
Cn,k
k=0
sin( )
sin
nk
Pk (cos ).
n X
nk
n
cos
1
Pn (sin ) =
=
Cn,k
Pk (cos ),
sin
cos
k=0
or
(sin )n Pn (sin ) =
(2)
n
X
k=0
Pn (cos )
n
X
k=0
k=0
Pn (x) =
n
X
k=0
2 sin cos
sin
n X
n
k=0
Cn,k
sin
2 sin cos
nk
Pk (cos ),
83
Pn (cos 2)
= (2 cos )n
=
n
X
n
X
k=0
k=0
(4)
n
X
k=0
n=0
yt(x t + )
22
yt(x t )
0 F1
2
2
= 1 0 F1
1;
1;
1
n=0
1 1
1
c, c + ;
2 2
2
2c1 2
2 c
=
( + xyt yt ) 2 F1
.
y 2 t2 (x2 1)
(2 + xyt yt2 )2
3 F2 (n, c, 1
c; 1, 1; y)Pn (x)tn .
n=0
(1)
n1
Pn
xt
p
=
X
(n + k)!Pn+k (x)tk
k=0
k!n!
t(x 1)
; 1;
2
we replace x by
X
t(x + 1)
Pn (x)tn
=
,
0 F1 ; 1;
2
(n!)2
n=0
xt
yt
and t by
to get
p
p
, p = (1 2xt + t2 ) 2 .
0 F1
yt
; 1;
2p
xt
1
p
0 F1
yt
; 1;
2p
xt
+1
p
=
=
=
=
85
n n
X
(1)n pn1 Pn ( xt
p )y t
(n!)2
n=0
n=0 k=0
X
n
X
n=0 k=0
k![(n k)!]3
(1)k n!y k
Pn (x)tn ,
(k!)3 (n k)!
or
X
n;
yt(x
t
+
p)
yt(x
p)
F
F
;
1;
=
p1 0 F1 ; 1;
0
1
1
2
2p2
2p2
n=0
1, 1;
y Pn (x)tn .
Next let us apply the same process to the known generating relation
c c+1
,
;
2
2
(1 xt)c 2 F1
X
(c)n Pn (x)tn
t2 (x2 1)
.
=
n!
(1 xt)2 n=0
1;
xt
yt
We replace x by
, t by
and note that (1 xt) becomes
p
p
1+
yt(x t)
= p2 [p2 + xyt yt2 ]
p2
"
xt
p
2
#
1 =
y 2 t2 2
y 2 t2 (x2 1)
2
2
[x
2xt
+
t
t
+
2xt
t
]
=
.
p4
p4
We thus obtain
c c+1
,
;
2
p2c1 [p2 + xyt yt2 ]c 2 F1 2
2 2
y t (x 1)
+ xyt yt2 )2
(p2
n n
X
(1)n (c)n pn1 Pn ( xt
p )y t
=
n!
n=0
=
=
=
X
(1)n (c)n (n + k)!Pn+k (x)y n tn+k
k!n!n!
n,k=0
n
nk
X X (1)
(c)nk n!Pn (x)y nk tn
n=0 k=0
X
n
X
n=0 k=0
k![(n k)!]2
(1)k (c)k n!y k Pn (x)tn
.
(k!)2 (n k)!
c c+1
,
;
2
p2c1 [p2 +xytyt2 ]c 2 F1 2
n, c;
X
=
2 F1
n=0
1;
y 2 t2 (x2 1)
2
(p + xyt yt2 )2
y Pn (x)tn .
3 F2 (n, c, 1
c; 1, 1; y)Pn (x)t
n=0
=
=
X
n
X
(1)k n!(c)k (1 c)k Pn (x)y k tn
(k!)3 (n k)!
n=0 k=0
k!n!
k=0 n=0
3 F2 (n, c, 1
n=0
k=0
= p1
k
k
pk1 Pk ( xt
p )(1) (c)k (1 c)k (yt)
(k!)2
yt k
X
(c)k (1 c)k Pk ( xt
p )( p )
k=0
=p
c; 1, 1; y)Pn (x)tn
(k!)2
c, 1 c;
r
2 2
2y(xt)t
1+ yt
+ yp2t
1+
p
p2
2 F1
1
c, 1 c;
p+yt
= p1 2 F1
"
2 F1
12xt(1y)+(1y)2 t2
2p
c, 1 c;
1 yt
p p
pyt
1;
1
Pn
Solution 9. Consider
1 xt
=
n
X
k=0
p2 +2yt(xt)+y 2 t2
2
c, 1 c;
2 F1
1;
12xt(1y)+t2 (1y)2
2
pn Pn
n=0
1 xt
p
yn
87
2 2 2
= [1 2 1xt
p yp + y p ]
1
= [1 2y(1 xt) + y 2 p2 ] 2
1
= [1 2y + 2xyt + y 2 2xy 2 t + y 2 t2 ] 2
1
= [(1 y)2 + 2xyt(1 y) + y 2 t2 ] 2
1
yt
yt
= (1 y)1 [1 2x( 1y ) + ( 1y )2 ] 2
X
(1)k Pk (x)y k tk
=
(1 y)k+1
k=0
n
k
X X (1) Pk (x)n!tk y n
=
k!(n k)!
n=0 k=0
n
X
X
(1)k Cn,k Pk (x)tk y n .
=
n=0 k=0
It follows that
pn Pn
1 xt
p
=
n
X
k=0
as desired.
Problem 10. With the aid of the result in Example 7 pg. 31, show that
n, n + 1, ;
1 .
1, + ;
(1+x)1 (1x)1
n
X
(n)k (n + 1)k (1 x)k
k=0
2k (k!)2
Hence
2++k1 ()( + k)
.
( + + k)
dx.
n
X
(n)k (n + 1)k 2++k1 ()( + k)
k=0
+1
2k (k!)2 P ( + + k)
n
()() X (n)k (n + 1)k ()k
( + )
k=0
(k!)2 ( + )k
Therefore
Z
(A)
n, n + 1, ;
1 .
1, + ;
n, n + 1, ;
2
(1)()
F
(1 x)1 Pn (x)dx =
( + 1)
1
1, 1 + ;
1 .
3 F2
n, 1 + n, ;
1 + , 1;
(1 )n
1 =
.
(1 + )n
Thus we get
Z
(1 x)1 Pn (x)dx =
(1)
1
2 (1 )n
2 (1 )n
.
(1 + )n
()n+1
(1 + x)1 Pn (x)dx
n, n + 1, 1;
()(1)
F
= 2
( + 1)
1, +1;
n, n + 1;
1
=
2 F1
+ 1;
2
( + 1)()
=
( + 1 + n)( n)
2 (1)n (1 )n
=
(1 + )n
(1)n 2 (1 )n
=
.
()n+1
1
1
(1+x)
1
Hence
(1x)
n, n + 1, n + 1 ;
+ 1 )
F
(n + 1)
1, n + 1;
n ()(n
Pn (x)dx = 2
1 .
n, n + 1 ;
2n ()(1 )(1 )n
1
=
2 F1
n!
1;
2n ()(1 )(1 )n (1)n (1 + 1 1)n
=
,
n!
(1)n
(1)n 2n (1 )n (1 )n
.
sin()(n!)2
Problem 11. Obtain from equation (5), page 168, the result
!
r
n
X
n
1+x
n
2
2
(1 + x) Pn
Cn,k Pk (x)
=2
2
k=0
n
2
(1 + x) Pn
1
1+x
2
!
Pm (x)dx.
Pn (cos ) =
sin
sin
n X
n
Cn,k
k=0
sin( )
sin
nk
Pk (cos )
to get
Pn
cos
2
1
2 cos( 2 )
!n
n
X
Cn, Pk (cos ),
k=0
or
r
n
2
(2)
[1 + x] Pn
1+x
2
!
n
= 2 2
n
X
Cn,k Pk (x).
k=0
Then
Z
B=
n
2
(1 + x) Pn
1
1+x
2
!
n
Pm (x)dx = 2 2
n
X
k=0
= 21 2
Cn,k
89
n
X
(1)k (n)k ( 1 )k
2
k!( 32 )k
k=0
1
n, ;
n
1
= 21 2 2 F1
.
3
;
2
Pk (x)Pm (x)dx.
1
Z
1 1 n
x Pn2k (x)dx,
2 1
0
and check your result by means of Theorem 65, page 181. Thus show that
Z
xn Pn2k (x)dx =
xn Pn2k (x)dx =
n!
2n k!( 32 )nk
xn+2k Pn (x)dx =
(n + 2k)!
2n+2k k!( 32 )n+k
1
x Pn2k (x)dx =
2
n
A=
0
xn Pn2k (x)dx.
We have
Z
A=
0
n + 2k, n 2k + 1;
1x
dx
2
x n 2 F1
1;
and we apply Theorem 37 with = n + 1, = 1, t = 1, p = 2, q = 1, a1 =
1
n + 2k, a2 = n 2k + 1, b1 = 1, c = , k = 0s = 1. The result is
2
n + 2k, n 2k + 1, 1;
Z 1
1
a, 1 a;
2 F1
21c (c)( 12 )
1
=
ca+1 ,
( c+a
)
2
2 )(
2
c;
which we use with a = n + 2k, 1 a = 1 + n 2k, c = n + 2. Then
(n + 1)(1) 2n1 (n + 2)( 12 )
2n2k+3
(n + 2) ( 2k+2
)
0
2 )(
2
3
( 2 )
n!
= n
2 k!(n k + 32 )
n!
= n 3
.
2 k!( 2 )nk
Recall that we found in Theorem 65 that
Z
xn Pn2k (x)dx
[2]
(2n + 4s + 1)Pn2s (x)
n! X
.
x = n
2 s=0
s!( 32 )ns
n
91
1
=
2
x Pn2k (x)dx
0
xn Pn2k (x)dx
n
[2]
(2n 4s + 1)
n! X
2n+1
R1
1
s!( 32 )ns
s=0
.
.
xn Pn2k (x)dx =
n! 2n 4k + 1
2
n!
.
= n 3
2n+1 k!( 32 )nk 2n 4k + 1
2 k!( 2 )nk
xn+2k Pn (x)dx =
(n + 2k)!
.
2n+2k k!( 32 )n+k
Problem 13. Use the formula (5), page 104, to obtain the result
n
t
1
xn (1 x)n dx
=
Q
,
n
2 )n+1
(1
x
2
t
0
where Qn (t) is the function given in (4), page 182.
Z
B1 =
n + 1,
B1 = B(n + 1, n + 1)t2n+1 F
n+1 n+2
,
;
2
2
2n + 2 2n + 3
,
;
2
2
t2
,
or
Z
0
n+1 n+2
,
;
2
2
(n + 1)(n + 1) 2n+1
F
=
t
(2n + 2)
3
n+ ;
2
(n!)2 t2n+1 2n ( 1t )n+1 ( 32 )n
1
=
Qn
,
(2n + 1)!
n!
t
xn (t x)n
dx
(1 x2 )n+1
t2
in terms of the !n (t) of page 182. We thus obtain, since n!22n ( 32 )n = (2n + 1)!,
Z
0
xn (t x)n
dx =
(1 x2 )n+1
n
t
1
Qn
.
2
t
Pn (x) =
2n ( 12 )n (x
1)
n!
2n;
.
1x
=
=
n
X
(1)nk (2n k)!( 1x )nk
2
k![(n k)!]2
k=0
n X
n
2 k
(2n)!(n)k (n)k ( 1x
)
x1
(2n)k k!(n!)2
n, n;
2n ( 21 )n (x 1)n
2
=
.
2 F1
n!
1x
2n;
2
k=0
Pn (x) =
2n ( 21 )n (x
n!
+ 1)
F
2n;
.
1+x
Solution 15. Since Pn (x) = (1)n Pn (x), it follows from the result in Exercise 14
that
n, n;
2n ( 21 )n (x + 1)n
2
Pn (x) =
.
2 F1
n!
1+x
2n;
Problem 16. Show that for |t| sufficiently small
n=0
(1 2xt + t2 ) 2 =
Pn (x)tn
n=0
we obtain
1
t(1 2xt2 + t4 ) 2 =
Pn (x)t2n+1 .
n=0
93
n=0
Now replace t by t2 :
n=0
( 21 )n
to arrive at
n!
(1 x)n = 2n (n!)2
n
X
(1)k (2k + 1)Pk (x)
k=0
(n k)!(n + k + 1)!
1
(1 x) and
2
Pn (x)tn
= (1 2xt + t2 ) 2
n=0
1
(1 t) 1 F0
X
4vt
=
Pn (1 2v)tn .
2
(1 t)
n=0
( 12 )n
and c = 1. This yields
n!
vn =
or
n
(1 x) = 2 (n!)
n
X
(1)k (2k + 1)Pk (x)
k=0
as desired.
(n k)!(n + k + 1)!
0
= nPn1 (x) (1 x)Pn1
(x)
n1
n1
X
X
=
Pk (x) 2(1 x)
Pk0 (x)
k=0
n1
X
k=0
(1)
nk+1
(1 + 2k)Pk (x).
k=0
d
Pn (1 2v) nPn (1 2v)
dv
d
Pn1 (1 2v)
dv
n1
X
d
=
[Pk (1 2v) + 2v Pk (1 2v)]
dv
k=0
n1
X
(1)nk (2k + 1)Pk (1 2v).
=
= nPn1 (1 2v) v
k=0
dy
dy
Since x = 1 2v, we know that v
= (1 x) .
dv
dx
Hence the above relations become
0
(1 x)Pn0 (x) + nPn (x) = nPn1 (x) (1 x)Pn1
(x),
n1
X
k=0
n1
X
k=0
Problem 19. Use Rodrigues formula, page 162, and successive integrations by
parts to derive the orthogonality property for Pn (x) and to show that
Z
Pn2 (x)dx =
2
.
2n + 1
1
Solution 19. We know Pn (x) = n D n (x2 1)n .
2 n!
Then
Z
Pn (x)Pm (x)dx
1
Z 1
1
= n
[D n (x2 1)n ]Pm (x)dx
2 n! 1
Z 1
1
1 n1 2
1
= n
{D
(x 1)n }Pm (x) 1 n
[D n1 (x2 1)n ][DPm (x)]dx
2 n!
2 n! 1
= ...
Z
(1)n 1 0 2
= n
[D (x 1)n ][D n Pm (x)]dx.
2 n! 1
95
Pn (x)Pm (x)dx = 0, m 6= n.
1
Pn2 (x)dx =
Now Pn (x) =
2n ( 21 )n xn
n!
(1)n
2n n!
1
.
2 n
+ n1 , so D n Pn (x) = 2n
Therefore
Z
Pn2 (x)dx =
( 21 )n
n!
(1 x2 )n dx =
( 12 )n
n!
Pn2 (x)dx =
22n+1 ( 21 )n (n + 1)(n + 1)
( 12 )n n+1+n+11
2
B(n + 1, n + 1) =
.
n!
n!(2n + 2)
Hence
Z
Pn2 (x)dx
22n+1 ( 21 )n n!
(2n + 1)!
22n+1 ( 21 )n n!
= 2n 3
2 n!( 2 )n
2 21
=
n + 12
2
.
=
2n + 1
=
1 2 2
e 0 F1 ; t; x t =
yn (x)tn
4
n=0
t
and that Theorem 45, page 133, is applicable to this yn (x). Translate the result
into a property of Pn (x), obtaining equation (7), page 159.
Solution 20. Consider the polynomial
n
(1 x2 ) 2
yn (x) =
Pn
n!
We find that
1 x2
.
yn (x)tn
n=0
P
X
n
(t 1 x2 )n
1
1x2
n!
n=0
1
t2 (1 x2 )
1
1 x2
= et 0 F1
1;
;
t 2 x2
.
4
= et 0 F1
1;
wn (x)tn ,
n=0
with
(u) =
n un .
n=0
Show that
n=0
(c)2kk
k=0
t2 (x2 1)
(1 xt)2
k
and thus obtain a result parallel to that in Theorem 46, page 134. Apply your
new theorem to Legendre polynomials to derive equation (2), page 164.
Solution 21. (Solution by Leon Hall) Multiplication of series gives
!
!
X
X
X
xn t n
n
2
n 2n
wn (x)t
=
n (x 1) t
n!
n=0
n=0
n=0
n
X
X
(2k + 1 + tx)
nk x2k (x2 1)nk t2n ,
=
(2k
+
1)!
n=0
k=0
and so
n
X
nk (x2 1)nk x2k
w2n (x) =
k=0
and
w2n+1 (x) =
(2k)!
n
X
nk (x2 1)nk x2k+1
k=0
(2k + 1)!
n
X
k (x2 1)k x2n2k
k=0
and
w2n+1 (x) =
(2n 2k)!
X
k (x2 1)k x2n2k+1
k=0
(2n 2k + 1)!
97
Thus
(c)n wn (x)tn
n=0
X
n
X
(c)2n k (x2 1)k
n=0 k=0
(2n 2k)!
x2n2k t2n +
X
n
X
(c)2n+1 k (x2 1)k
n=0 k=0
(2n 2k + 1)!
x2n2k+1 t2n+1 .
X
(c)2n x2n2k t2n X (c)2n+1 x2n2k+1 t2n+1
+
.
(2n 2k)!
(2n 2k + 1)!
n=k
n=k
Noting that (c)2n+2k = (c + 2k)2n (c)2k and (c)2n+2k+1 = (c + 2k)2n+1 (c)2k we have
#
"
X
X
X (c + 2k)2n (xt)2n X
(c + 2k)2n+1 (xt)2n+1
n
2
k 2k
(c)n wn (x)t =
+
(c)2k k (x 1) t
(2n)!
(2n + 1)!
n=0
n=0
n=0
k=0
X
1
(c + 2k)n (xt)n
=
.
c+2k
n!
(1
xt)
n=0
1
) Now we have
(1 xt)c+2k
2 2
k
X
X
t (x 1)
n
c
(c)n wn (x)t = (1 xt)
(c)2k k
,
(1 xt)2
n=0
k=0
which is similar to the result of Theorem 46, pg. 134. For the second part, note
that the formula (4) on pg.165 is
X
Pn (x)tn
1 2 2
xt
= e 0 F1 ; 1, t (x 1) ,
n!
4
n=0
where Pn is the Legendre polynomial of degree n. If we now let
Pn (x)
n!
wn (x) =
and
(u) =
X
k=0
uk
,
22k (k!)2
1
we get k = 2k
, and our result gives
2 (k!)2
2 2
n
X
X
(c)n Pn (x)tn
(c)2n
t (x 1)
.
= (1 xt)c
n!
22n (n!)2 (1 xt)2
n=0
n=0
c c 1
(c)2n
Note that 2n =
+
and that
2
2 n 2 2 n
2 F1
(1 , 2 ; 1; z) = 1 +
X
(1 )n (2 )n z n
.
n!
n!
n=1
c
c 1
t2 (x2 1)
we have
, 2 = + , and z =
2
2 2
(1 xt)2
X
(c)n Pn (x)tn
c c 1
t2 (x2 1)
= (1 xt)c 2 F1
, + ; 1;
n!
2 2 2
(1 xt)2
n=0
Thus, with 1 =
Hn (x) =
[2]
X
n!Hn2k (1)xn2k (x2 + 1)k
k!(n 2k)!
k=0
Solution 1. From
exp(2xt t2 ) = exp(2xt x2 t2 )exp[t2 (x2 + 1)]
we obtain
n=0
X
Hn (1)(xt)n
n!
n=0
X
(x2 + 1)n t2n
n!
n=0
[2]
X
X
Hn2k (1)(x)n2k (x2 + 1)k tn
k!(n 2k)!
n=0 k=0
It follows that
n
Hn (x) =
[2]
X
n!Hn2k (1)xn2k (x2 + 1)k
k!(n 2k)!
k=0
2k
[2]
X
n!Hn2s (x)
x =
.
2n s!(n 2s)!
s=0
n
Then
n!
.
k!
[2]
X
s=0
n!
n s!(n 2k)!
99
The integrals involved on the right are zero except for s = k. Hence
Z
2
n!
2
ex Hn2k
(x)dx
n
2 k!(n 2k)!
n!2n2k (n 2k)!
=
n
2 k!(n 2k)!
n!
= 2k .
2 k!
Problem 3. Use the integral evaluation in equation (4), page 192, to obtain the
result
Z
(1)k+s 22k+2s ( 12 )k ( 32 )s
.
2s + 1 2k
ex Hn (x)Hm (x)dx = 0, m 6= n.
If m and n are both odd or both even, the integrand above is an even function of
x and we obtain
ex Hn (x)Hm (x)dx = 0
2
0
for m n mod 2.
Now consider
Z
2(2n 2s 1)
0
h
i
2
0
(x) 202n (x)H2s+1 (x)
= ex H2n (x)H2s+1
0
0
0
(0) + H2n
(0)H2s+1 (0)
= H2n (0)H2s+1
0
= H2n (0)H2s+1 (0).
0
(0) from page 323, we get
Using the values of H2n (0) and H2s+1
Z
0
Problem 4. By evaluating the integral on the right, using equation (2), page 187,
ans term-by-term integration, show that
(A)
Pn (x) =
n!
n!
2
=
et tn Hn (xt)dt
et tn
[2]
X
(1)k (2xt)n2k dt
k=0
[n
2]
X (1)k (2k)n2k
k!(n 2k)!
k=0
[n
2]
k!(n 2k)!
Z
1
1
e nk 2 d
0
X (1)k (2x)n2k (n k + 1 )
2
k!(n 2k)!
( 12 )
k=0
[n
2]
k!(n 2k)!
k=0
Hence
2
n!
et tn Hn (xt)dt = Pn (x).
Problem 5. Let vn (x) denote the right member of equation (A) of Exercise 4.
Prove (A) by showing that
vn (x)y n = (1 2xy + y 2 ) 2 .
n=0
n!
Then
vn (x)y
n=0
et tn Hn (xt)dt.
Z
n
n
2 X t Hn (xt)t
2
=
et
dt
n!
0
n=0
Z
2
2 2
2
=
et e2(xt)yty t dt
Z0
2
2
2
=
et [12xy+y ] dt.
0
p
Use t 1 2xy + y 2 = to get
2
vn (x)y n = p
1 2xy + y 2
n=0
Hence vn (x) = Pn (x).
Z
0
e d = (1 2xy + y 2 ) 2 =
X
n=0
Pn (x)y n .
101
(2t) Pn
x
t
[2]
X
n!(x2 t2 )k (2x)n2k
(k!)2 (n 2k)!
k=0
derived from equation (1), page 164, and term-by-term integration to prove the
validity of (B), which is equation (5), page 191.
Solution 6. We know
n
Pn (x) =
[2]
X
n!(x2 1)k xn2k
k=0
(2t) Pn
x
t
[2]
X
n!(x2 t2 )k (2x)n2k
(k!)2 (n 2k)!
k=0
n+1 x2
t2 n+1
Pn
x
t
dt =
[2]
X
k=0
n!ex
2
(k!) (n 2k)!
Put t x2 = . Then
n+1 x2
t2 n+1
Pn
[2]
Z
X
n!(2x)n2k
e (1)k k d
dt =
t
(k!)2 (n 2k)! 0
x
k=0
[n
2]
X n!(1)k (2x)n2k k!
k=0
[n
2]
(k!)2 (n 2k)!
X n!(1)k (2x)n2k
k=0
k!(n 2k)!
Hence
2
n+1 x2
et tn+1 Pn
x
t
dt = Hn (x).
d
dx
; m 6= n
;m = n
[D n ex ]Hm (x)dx
Z
hn
o
i
2
= (1)n D n1 ex Hm (x)
+ (1)n+1
ex Hn (x)Hm (x)dx
= (1)n
= ...
= (1)2n
ex [D n Hm (x)]dx.
Hence
Z
ex Hn (x)Hm (x)dx = 0, m 6= n
2
n
= 2 n!
ex dx
= 2n n! .
14. Chapter 12 Solutions
[2]
X
(1)k n!(2x)n2k
k=0
n
X
=
=
k=0
n
X
k=0
n
X
k!(n 2k)!
we get
(1)k (2n)!(2x)2n2k
k!(2n 2k)!
(1)nk (2n)!(2x)2k
(2k)!(n k)!
Hence
[D n1 ex ][DHm (x)]dx
( 12 )
103
(x2 ).
H2n+1 (x)
=
=
n
X
(1)k (2n + 1)!(2x)2n+12k
k=0
n
X
k=0
n
X
k!(2n + 1 2k)!
nk
(1)
(2n + 1)!(2x)2k+1
(n k)!(2k + 1)!
Hence
(1)
1
1
(n k), (n k 1);
2
2
Pn() (x)
n
X
=
2 F3
3
k=0
1
1
+ k, (1 + + k), (2 + + k);
2
2
2
[n]
2
X
(2x)n
(2n 4k + 1)Pn2k (x)
=
.
n!
k!( 32 )nk
k=0
Then
1
4
.
2k (n k)!( 3 ) (1 + )
k
2 k
Ln() (x)tn
n=0
X
n
X
(1)s (1 + )n xs tn
n=0 s=0
s!(n s)!(1 + )s
[2]
X
X
(1)s (1 + )n+s (2s 4k + 1)Ps2k (x)tn+s
=
2s n!(1 + )s k!( 23 )sk
n,s=0 k=0
X
(1)s (1 + )n+s+2k (2s + 1)Ps (x)tn+s+2k
=
2s+2k k!n!(1 + )s+2k ( 23 )s+k
n,k,s=0
k
n+k+2s
change letters X (1) (1 + )n+k+2s (2k + 1)Pk (x)t
=
3
2k+2s s!n!(1 + )k+2s ( 2 )k+s
n,k,s=0
n
[2]
X
X
(1)k (1 + )n+k (2k + 1)Pk (x)tn+k
=
s!(n 2s)!(1 + )k+2s ( 32 )k+s 2k+2s
n,k=0 s=0
n
[2]
X
X
(n)2s 22s
(1)k (1 + )n+k (2k + 1)Pk (x)tn+k
s!(1 + + k)2s ( 23 + k)s
(1 + )k ( 32 )k 2k n!
n,k=0 s=0
n n1
,
;
2
2
k
n+k
X
1
(1) (1 + )n+k (2k + 1)Pk (x)t
=
.
2 F3
3
4
2k (1 + )k ( 2 )k n!
3
n,k=0
1++k 2++k
+ k,
,
;
2
2
2
=
We thus get
Ln() (x)tn =
n=0
X
n
X
nk nk1
,
;
2
2
F
2 3
3
n=0 k=0
1++k 2++k
+ k,
,
;
2
2
2
1
4
2k (1 + )k ( 32 )k (n k)!
Hence
nk nk1
,
;
2
2
1
1
(n k), (n k 1);
2
2
n
X
L()
2 F2
n (x) =
1
k=0
1
(1 + + k), (2 + + k);
2
2
n
X
Ln() (x)
1
4
(1)k (1 + )n Hk (x)
k!(n k)!2k (1 + )k .
105
Solution 3. From
[n]
2
X
(2x)n
Hn2s (x)
=
n!
s!(n 2s)!
s=0
we obtain
Ln() (x)tn
n=0
X
n
X
(1)k (1 + )n xk tn
n=0 k=0
X
n,k=0
k!(n k)!(1 + )k
[2]
X
X
(1)k (1 + )n+k Hk2s (x)tn+k
=
n!(1 + )k 2k s!(k 2s)!
n,k=0 s=0
X
(1)k+2s (1 + )n+k+2s Hk (x)tn+k+2s
=
n!(1 + )k+2s 2k+2s s!k!
n,k,s=0
n
[2]
n,k=0
Thus we obtain
X
n=0
Ln() (x)tn
[2]
X
X
n n1
,
;
2
2
k
n+k
X
1
(1) (1 + )n+k Hk (x)t
=
F
2 2
k
k!n!2 (1 + )k
4
1++k 2++k
n,k=0
,
;
2
2
nk nk1
2
2
n
X
k
n
X
(1) (1 + )n Hk (x)t .
=
2 F2
k
4
1++k 2++k
2 k!(n k)!(1 + )k
n=0 k=0
,
;
2
2
=
Problem 4. Use the results in Section 56, page 102, to show that
Z
Ln [x(t x)]dx =
0
(1)n H2n+1 ( 2t )
.
22n ( 32 )n
Ln [x(t x)]dx =
t
1 F1 (n; 1; x(t
x))dx.
In Theorem 37 we use = 1, = 1, p = 1, q = 1, a1 = n, b1 = 1, k = 1, s =
1, c = 1 to get
Ln [x(t x)]dx
= B(1, 1)t3 F3
2 3
1, , ;
2 2
3 t2
= t1 F1 n; ;
2 4
1 2
n!
( 12 )
= 3 Ln
t .
4
( 2 )n
n, 1, 1;
t2
4
Ln 2
1 2
t
4
=
(1)n H2n+1 ( 12 t)
.
22n tn!
Hence
Z
Ln [x(t x)]dx =
(1)n
H2n+1
22n ( 32 )n
1
t .
2
Problem 5. Use the results in Section 56, page 102, to show that
Z
0
p
H2n ( x(t x))dx
1
1 2
n
2n
p
= (1) 2
Ln
t .
2 n
4
x(t x)
Solution 5. Since
H2n (x) = (1)n 22n
1 2
1
F
;
x
,
n;
1 1
2 n
2
Z t
p
p
1
n 2n 1
21
21
H2n ( x(t x)) x(t x)dx = (1) 2 ( )n
x (tx) 1 F1 n; ; x(t x) dx.
2
2
0
1
1
, = , p = 1, q = 1, a1 = n, b1 =
2
2
1 1
n, , ;
2 2
1 1 0
n 2n 1
= (1) 2 ( 2 )n B
,
t 3 F3
2 2
1 1 2
, , ;
2 2 2
n;
( 12 )( 12 )
t2
n 2n 1
= (1) 2 ( 2 )n
1 F1
(1)
4
1;
1
= (1)n 22n ( 12 )n Ln
t2 .
4
Z
0
p
H2n ( x(t x))
p
dx
x(t x)
t2
4
107
L()
n (x) =
(1)
()
(2c2)
n
(x)
(1 + )n X (1 + c)k Lk (x)Lnk
.
(c)n
(1 + )k
k=0
1
1
1
1
In (1) choose c = 1 + + m. Then 2c 2 = m and 1 + c = m.
2
2
2
2
Hence
Ln() (x) =
()
(m
n
X
( m
(1 + )n
2 )k Lk (x)Lnk (x)
.
(1 + )k
(1 + 12 + 21 m)n k=0
()
(+2m)
Ln (x)
=
=
()
(+2m)
n
(x)
(1 + )n X (m)k Lk (x)Lnk
(1 + + m)n
(1 + )k
k=0
()
(+2m)
n
X
(m)k Lk (x)Lnk
(x)
(1 + )n (1 + )m
(1 + )m+n
(1 + )k
k=0
Problem 8. Use integration by parts and equation (2), page 202, to show that
Z
()
ey Ln() (y)dy = ex [Ln() (x) Ln1 (x)].
x
An =
and integrate by parts with u = e
()
()
v = Ln (x) Ln+1 (x) to get
h
An = e
x
y
ey Ln() (y)dy
()
n
oi Z
()
()
Ln (y) Ln+1 (y)
+
x
h
i
()
ey Ln() (y) Ln+1 (y) dy.
h
i
()
An = ex Ln+1 (x) Ln() (x) + An An+1 .
Hence
h
i
()
An+1 = ex Ln+1 (x) Ln() (x) ,
so that, by a shift of index, we get
Z
An =
h
i
()
ey Ln() (y)dy = ex Ln() (x) Ln1 (x) .
(1 + )() (1 + )n t+ (+)
L
(t).
(1 + + ) (1 + + )n n
x (t x)
(1 + )n
=
n!
Ln() (x)dx
In Theorem 37 use = + 1, = , p = 1, q = 1, a1 = n, b1 = 1 + , c =
1, k = 1, s = 0 to arrive at
(1 + )n
B( + 1, )t+ 2 F2
n!
n,
+1
;
1
1 + , + + 1;
(1 + )n (1 + |alpha)() +
=
t
1 F1 (n; + + 1; t)
n!
( + + 1)
(1 + + n) (1 + )()t+
n!
(+)
(t)
=
Ln
n!(1 + )
(1 + + ) (1 + + )n
(1 + )n ((1 + )() + (+)
=
(t).
t
Ln
(1 + + )n (1 + + )
st
e
0
1
Ln (t)dt =
s
n
1
1
.
2
Solution 10.
Z
0
est Ln (t)dt
=
0
n
X
est
n
X
(1)k n!tk dt
(k!)2 (n k)!
k=0
(1)k n!
=
k!(n k)!sk+1
k=0
n
1
1
=
1
.
s
s
109
Problem 11. Show by the convolution theorem for Laplace transforms, or otherwise, that
t
Ln (t x)Lm (x)dx =
0
n
1
1
Solution 11. We know that L
1
= Ln (t). (where L denotes
s
s
Laplace transform and Ln denotes the Laguerre polynomial).
Then by the convolution theorem,
n
m
Z t
1
1
1
1
Ln (t x)Lm (x)dx =
1
1
L
s"
s
s
s
0
n+m #
1
1 1
.
1
=
s s
s
1
Hence
Z
( "
n+m #)
1
1
1
1
.
Ln (t x)Lm (x)dx = L 1
s s
s
But
L
(
n+m )
1
1
1
= Ln+m (t).
s
s
Hence
Z
Ln (t x)Lm (x)dx =
Ln+m ()d.
Ln (t x)Lm (x)dx
of (7), page 206, by the following method. From (4), Section113, show that
Z
X
n=0
x e
2
2n
[L()
n (x)] dxt
x(1 + t)
dx
= (1 t)
x exp
1t
0
= (1 t2 )1 (1 + )
X
(1 + + n)t2n
=
.
n!
n=0
22
xt
1t
=
Ln() (x)tn .
n=0
Then
(1 t)22 exp
2xt
1t
=
X
n
X
()
()
Lk (x)Lnk (x)tn .
n=0 k=0
()
x e
0
()
()
Lk (x)Lnk (x)tn dx
= (1t)
22
Z
0
n=0 k=0
2xt
dt,
x exp x
1t
or
Z
X
Z
h
i2
x ex Ln() (x) dxt2n = (1 t)22
n=0
x exp
x(1 + t)
dx.
1t
1+t
= and thus obtain
1t
+1 Z
+1
x(1 + t)
1t
1t
x exp
dx =
e d = (1 + )
.
1t
1+t
1+t
0
Therefore we have
Z
X
n=0
h
i2
x ex Ln() (x) dxt2n
= (1 t)1 (1 + t)1 (1 + )
= (1 + )(1 t2 )1
X
(1 + )n (1 + )t2n
=
n!
n=0
X
(1 + + n)t2n
=
.
n!
n=0
Hence
Z
0
h
i2
(1 + + n)
x ex Ln() (x) dx =
.
n!
15. Chapter 13 Solutions
111
d
.
dx
A(t)(xH(t))
=
=
n (x)tn
n=0
n+m (x)tn+m .
n=m
Now the n (x) form a simple set. Hence D m k (x) = 0 for k < m. Therefore,
using the operator D m on (1) we obtain
A(t)[H(t)]m (m) (xH(t)) =
D m n+m (x)tn+m .
n=0
(2)
H(t)
A(t)
t
(m)
(xH(t)) =
D m n+m (x)tn ,
n=0
from which it follows at once (Theorem 72) that D m n+m (x) are polynomials
of Sheffer A-type zero an that they belong to the same operator as do the n (x),
because the H(t) in (1) and (2) are the same function.
Problem 2. Prove Theorem B: If n (x) is of Sheffer A-type zero,
n (x) = n (x)
"m
Y
#1
(1 + i )n
i=1
is of Sheffer A-type m.
Solution 2. We are given that n (x) is of Sheffer A-type zero and wish to consider
n (x) =
n (x)
m
Y
(1 + i )n
i=1
Let n (x) belong to the operator J(D). By Theorem 74, there exists k , vk such
that
n1
X
k=0
k=0
such that
Bn (x) = nn (x).
Then
(B = i )n (x) = (n + i )N (x),
so that
(
J(D)
m
Y
)
[B(x, D) + i ] n (x) = J(D)
i=1
n (x)
m
Y
(1 + i )n1
i=1
n1 (x)
m
Y
= n1 (x).
(1 + i )n1
i=1
Since n (x) belongs to one operator whose coefficients have maximum degree m,
n (x) is of Sheffer A-type m and n (x) belongs to the operator
J1 (x, D) = J(D)
m
Y
[i + B(x, D)].
i=1
k D k+1 + x
vk D k+1 .
k=0
k=0
= 1 +
m D m+1 + x
m=0
= 1 2 + 1
#"
vm D m+1
2 +
m=0
k D
k+1
+ 1 x
k=0
k=0
k D k+1 + x
k=0
vk D
k+1
+ 2
X
k=0
m D
m+1
m=0
+ x2
#
vk D k+1
"
vm D
m=0
n
X
k=0
merely because the n (x) form a simple set. Then, because of (1),
m+1
X
m=0
(m + xvm )D m
(2)
n1
X
113
k=0
X
Hn (x)tn
n!
n=0
= exp(2xt t2 )
= exp(t2 ) exp(2xt).
Thus by Theorem 72, n (x) is a Sheffer A-type zero with A(t) = exp(t2 ) and
t
H(t) = 2t. Because J(x) is the inverse of H(t) we get J(t) = .
2
Hn (x)
The operator J may be found directly as follows. Denote
by n (x). Then
n!
we need (see p.189 for the first few Hermite polynomials):
T0 (x)D1 (x) = 0 (x)
T0 (x)D(2x) = 1.
So,
T0 (x) =
1
.
2
Continuing,
T1 (x)D2 2 (x) = 1 (x) T0 (x)D2 (x)
1
4T1 (x) = 2x 4x = 0.
2
So, T1 (x) = 0. In general we need
Tn (x)D
n+1
n1
X
k=1
Because
Hn0 (x)
2nHn1 (x)
= 2n1 (x)
n!
= n (x) n (x)
n1
X
k=1
n1
X
k=1
Because we have T1 (x) = 0, this makes T2 (x) = 0, which means T3 (x) = 0, and so
on. Therefore, from Theorem 70,
X
1
J(x, D) =
Tk (x)Dk+1 = D.
2
k=0
From the formulas for A(t) and H(t) it is easily seen that, in Theorem 73,
0
; k 6= 1
ak =
2 ; k = 1
and
0 ; k = 1, 2, . . .
2 ; k = 0.
This makes equation (11) from Theorem 73, for n 2,
1
(xD D2 )n (x) = nn (x)
2
or
00n (x) 2x0n (x) + 2nn (x) = 0,
which is Hermites differential equation. Hermites differential equation is also a
1
result of Theorem 74 using k = 0 except 1 = and v0 = 1, vk = 0; k 1.
2
Equation (17) in Theorem 75, using the k and k values we found before, becomes
k =
Hn (x)
Hn1 (x) 2Hn2 (x)
=
(n 1)!
(n 2)!
(n 1)!
or
1
[2xHn1 (x) 2(n 1)Hn2 (x) = Hn (x)] ,
(n 1)!
where the equation in brackets is the pure recurrence relation for Hermite polynomials.
()
Problem 4. Show that Ln (x) is of Sheffer A-type zero, and proceed as in Exercise 3.
Solution 4. (Solution by Leon Hall) A generating function for the Laguerre polynomials (see (4), p.202) is
X
1
xt
exp
=
Ln() (x)tn ,
(1 t)1+
1t
n=0
()
115
and
t
.
1t
An easy inverse calculation for H(t) yields
H(t) =
J(t) =
X
t
=
tn+1 .
1t
n=0
From
1+ X
A0 (t)
=
=
k tk
A(t)
1t
k=0
(k + 1)tk
k=0
()
(1 + x(k + 1))J k+1 (D)L()
n (x) = nLn (x).
()
k=0
By definition
()
J(D)Ln (x)
n1
X
()
= Ln1 (x) so
J k+1 (D)L()
n (x) =
n1
X
()
Ln1k (x) =
()
Lk (x),
k=0
k=0
k=0
n1
X
()
Lk (x) = DL()
n (x)
k=0
= (1 + x)
k=0
n1
X
J k+1 (D)L()
n (x) x
k=0
= (1 +
n1
X
k=0
()
x)DLn (x)
()
= (1 + x)DLn (x) x
n1
X
k=0
n1
X
kJ
k+1
()
k=0
()
()
k=0
()
()
= D2 Ln (x).
Thus, () becomes, finally
2 ()
()
(1 + x)DL()
n (x) xD Ln (x) = nLn (x)
(D)Ln() (x)
kLn1k (x).
k=1
k=1
kJ k+1 (D)L()
n (x)
or
()
()
xD2 L()
n (x) + (1 + x)DLn (x) + nLn (x) = 0,
()
k=1
t
, we get
1t
uA0 (u)
= (1 + )t
A(u)
and
uH 0 (u) = t t2 ,
making 0 = (1 + ), v0 = 1, v1 = 1, and the rest of the k and vk zero.
Then, using these values in equation (14) of Theorem 74 we again get Laguerres
differential equation.
Problem 5. Show that the Newtonian polynomials
(1)n (x)n
n!
are of Sheffer A-type zero, and proceed as in Exercise 3.
Nn (x) =
X
X
(1)n (x)n n
Nn (x)tn =
t = (1 + t)x = ex log(1+t) ,
n!
n=0
n=0
so in Theorem 72 we have A(t) = 1 and
H(t) = log(1 + t) =
X
(1)k
k=0
k+1
tk+1 .
X
k=0
tn+1
.
(n + 1)!
A(t) constant means all the k = 0 in Theorems 73 and 75 and that all the k = 0
in Theorem 74. Because
H 0 (t) =
X
1
=
(1)k tk ,
1+t
k=0
X
(1)k k+1
t
(k + 1)!
k=0
we get vk =
117
(1)k
in Theorem 74. Note that also if Nn (x) Nn (x+1)Nn (x),
(k + 1)!
then
(1)n
(1)n
((x + 1))n
(x)n
n! n1
n!
x+1 x+1n
(1)
(x)n1
=
(n 1)!
n
n
= Nn1 (x).
Nn (x)
k=0
or using Theorem 75
x
n1
X
k=0
n1
X
k=0
Theorem 74 yields
x
n1
X
k=0
(1)k k+1
D
Nn (x) = nNn (x).
(k + 1)!
(1)k
, we have
k+1
(1)k
Nn1k (x) = DNn (x)
k+1
or
1
1
(1)n1
Nn0 (x) = Nn1 (x) Nn2 (x) + Nn3 (x) . . . +
N0 (x).
2
3
n
With some work, this last result could also be obtained by logarithmic differentiation.
Problem 6. Show that
()
Ln (x)
(1 + )n
is of Sheffer A-type unity but that with chosen to be
n (x) =
= D( + )
the polynomials n (x) are of -type zero.
Solution 6. Consider
()
n (x) =
()
Ln (x)
.
(1 + )n
Now
()
X
Ln (x)tn
= et 0 F1 (; 1 + ; xt),
(1 + ) n
n=0
Ln (x)
. Since Ln (x) is of Sheffer A-type zero (by
(n!)2
Exercise 4), we may use Exercise 2 to conclude that n (x) is of Sheffer A-type 2.
We now wish to find the operator to which n (x) belongs.
Let
Solution 7. Consdier n (x) =
J(x, D) =
Tk (x)D k+1 .
k=0
1x
1
=
; T0 (x)(1) = 1, T0 (x) = 1.
(1!)2
(0!)2
Then also
[T0 (x)D + T1 (x)D 2 ]
1
1
L2 (x) =
L1 (x),
(2!)2
(1!)2
or
1 2
[D + T1 (x)D ] 1 2x + x = 4(1 x)
2
2
119
3
1
1
[D + (2 3x)D 2 + T2 (x)D 3 ] 1 3x + x2 x3 = 9 1 2x + x2 .
2
6
2
or
1
3 + 3x x2 + (2 3x)(3 x) + T2 (x)(1) = 9 18x + 9x2 ,
2
or
9
1
3 3x + x2 + 6 11x + 3x2 T2 (x) = 9 18x + x2 ,
2
2
T2 (x) = 4x + x2 ; T2 (x) = 4x x2
Then
1
3
1
2
[D+(23x)D 2 +(4xx2 )D 3 +T4 (x)D 4 ] 1 4x + 3x2 x3 + x4 = 16 1 3x + x2 x3 ,
3
24
2
6
or
1
1
8
4 + 6x 2x2 + x3 +(23x) 6 4x + x2 +(4xx2 )(4+x)+T4 (x) = 1648x+24x2 x3 .
6
2
3
Thus we get
x3
(1 + 9 + 6 16) = x2 ; T4 (x) = x2 .
6
Let us consider the effect of the operator
T4 (x) = x2 +
n (x)tn
n=0
or the simple set of polynomials n (x) belonging to a Sheffer operator J(x, D),
no matter what the A-type of n (x), we can obtain a generating fuction (sum the
series)
n (x)tn
n=0
for any other polynomial n (x) belonging to the same operator J(x, D).
Solution 8. We are given that n (x) is a simple set of polynomials and
(1)
y(xt) =
X
n=0
n (x)tn .
Let n (x) belong to the operator J(x, D) in the Sheffer sense. Let n (x) belong
to the same operator J(x, D). Then, by Theorem 71, there exist constants bk such
that
(2)
n (x) =
n
X
bk nk (x).
k=0
n (x)t
n=0
X
n
X
bk nk (x)tn
!
!
X
n
n
bn t
n (x)t
n=0 k=0
X
n=0
n=0
= B(t)y(x, t).
Note that the bk s can be computed identically.
Problem 9. Obtain a theorem analogous to that of Exercise 8 but with Sheffer
A-type replaced by -type.
Solution 9. Follow the proof in Exercise 8 except for two substitutions:
(1): Replace J(x, D) by J(x, );
(2): Replace Theorem 71 by Theorem 78.
Problem 10. Show that if Pn (x) is the Legendre polynomial,
1
x
(1 + x2 ) 2 n
Pn
n!
1 + x2
is a simple set of polynomials of Sheffer A-type zero.
n (x) =
(1 + x2 ) 2
n (x) =
Pn
n!
1 + x2
.
We know that
X
Pn (u)v n
v 2 (u2 1)
uv
= e 0 F1 ; 1;
.
n!
4
n=0
Therefore
X
n=0
n (x)t
P
X
n
x
1+x
(1 + x2 ) 2 tn
n!
n=0
=e
t 1+x2
x
1+x2
xt
= e 0 F1
1;
0 F1
;
t2 (1 + x2 )
1;
t2
.
4
4
x2
1+x2
1
121
t2
Hence n (x) is of Sheffer A-type zero with H(t) = t, A(t) = 0 F1 ; 1;
.
4
Problem 11. Let Pn (x) be the Legendre polynomial. Choose = D and show
that the polynomials
(x)n
Pn
n (x) =
(n!)2
are of -type zero for that .
x+1
x1
0 F1
; 1;
v(u 1)
2
0 F1
; 1;
x+1
x1
,
v(u + 1)
2
=
X
Pn (u)v n
.
(n!)2
n=0
x+1
1
u+1
x
u1
=
,
=
, and
. Then
x1
2
x1
2
x1
(1) yields
(2)
0 F1 (; 1; t)0 F1 (; 1; xt) =
n (x)tn .
n=0
Using Theorem 79 we may conclude from (2) that n (x) is of -type zero with
= D( + 1 1) = D and
A(t) = 0 F1 (; 1; t),
H(t) = t,
(t) = 0 F1 (; 1; t).
Problem 12. Prove that if the operator J(x, D) is such that a simple set of polynomials n (x) belongs to it in the Sheffer sense, then J(x, D) transforms every
polynomial of degree precisely n into a polynomial of degree precisely (n 1).
Solution 12. This result was proved as part of our solution to Problem 2 of this
chapter. See the theorem at the end of that solution.
Theorem : If the operator J(x, D) is such that a simple set of polynomials n (x)
belongs to it in the Sheffer sense,
(1)
gn (x) =
n
X
k=0
merely because the n (x) form a simple set. Then, because of (1),
J(x, D)gn (x) =
(2)
n1
X
k=0
Problem 1. Show that the Batemans polynomial (see Section 146, page 285)
Zn (t) = 2 F2 (n, n + 1; 1, 1; t)
has the recurrence relation
n
X
(1)k (n + k)!tk
(k!)3 (n k)!
k=0
or
Zn (t) =
(k, n).
k=0
Then
Zn1 (t) =
X
(1)k (n 1 + k)!tk
k=0
Zn2 (t) =
k=0
(k!)3 (n 1 k)!
X
nk
k=0
n+k
(k, n),
(n k)(n k 1)(n k 2)
(k, n).
(n + k)(n + k 1)(n + k 2)
Also
tZn1 (t) =
X
(1)k (n 1 + k)!xk+1
k=0
(k!)3 (n 1 k)!
X
(1)k1 (n 2 + k)!xk
k=1
or
tZn1 (t) =
X
k=0
tZn2 (t) =
k 3
(k, n)
(n + k)(n + k 1)
X
(1)k (n 2 + k)!xk+1
k=0
(k!)3 (n 2 k)!
X
(1)k1 (n 3 + k)!xk
k=1
tZn2 (t) =
X
k=0
123
k 3 (n k)
(k, n).
(n + k)(n + k 1)(n + k 2)
k=n:
coef f k 4 :
B + D = 0 :
(n 2)2
D
2(2n 3)
(2n 1)(n 2)2
=
n2 (2n 3)
E=
k = 1 n : B(1 n)3 (1) + C(2n 1)(2n 2)(1) D(1 n)3 (2n 1) + E(2n 1)(2n 2)(2n 3) = 0,
or
B(n 1)3 2C(n 1)(2n 1) + D(n 1)3 (2n 1) + E 2(n 1)(2n 1)(2n 3) = 0
Thus we have
2C(2n 1)
.
2
2
n (2n 3)
n
n2
THen
n2 C
2(2n 1)
n2
2(2n 1)
D=
n2
B=
Hence
C=
3n2 6n + 2
.
n2
(2n 1)(2n 2)(2n 3) + A(1)(2n 2)(2n 3) B(n 1)3 (2n 3) D(n 1)3 (1) = 0
or
2(2n 1)(2n 3) + 2A(2n 3) B(n 1)2 (2n 3) D(n 1)2 = 0,
or
2(2n 1)(n 1)2 (2n 3) 2(2n 1)(n 1)2
+
.
2A(2n 3) = 2(2n 1)(2n 3) +
n2
n2
Hence we get
k =n1:
n2 (2n 3)A
We now have
(2n 1)(3n2 6n + 2)
2(2n 1)
3n2 6n + 2
2(2n 1
(2n 1)(n 2)2
,B =
, C+
,D =
,E =
.
2
2
2
2
n (2n 3)
n
n
n
n2 (2n 3)
The desired recurrence relation is therefore
A=
n2 (2n 3)Zn (t) (2n 1) 3n2 6n + 2 2(2n 3)t Zn1 (t)
+(2n 3)[3n2 6n + 2 + 2(2n 1)t]Zn2 (t) (n 2)2 (2n 1)Zn3 (t).
Note the one sign change from that in Sister Celines paper. The sign was correct
in her thesis.
Problem 2. Show that Sister Celines polynomial fn (a; ; x), or
1
Fn (x) = 3 F2 n, n + 1, a; 1, ; x
2
has the recurrence relation
nfn (x)[3n24(n1+a)x]fn1 (x)+[3n44(n1a)x]fn2 (x)(n2)fn3 (x) = 0.
Solution 2. Consider
1
fn (x) = 3 F2 n, n + 1, a; 1, ; x ,
2
for which we seek a pure recurrence relation.
Now
fn (x) =
n
X
(1)k (n + k)!(a)k xk
k=0
k!(n k)!k!(k)k
n
X
(1)k (n + k)!(a)k (4x)k
k=0
k!(2k)!(n k)!
Put
(k, n) =
Then
fn (x) =
125
(k, n),
k=0
fn1 (x) =
X
(1)k (n 1 + k)!(a)k (4x)k
k!(2k)!(n 1 k)!
k=0
fn2 (x) =
X
nk
k=0
X
(n k)(n k 1)
k=0
fn3 (x) =
(n + k)(n + k 1)
n+k
(k, n),
X
(n k)(n k 1)(n k 2)
k=0
(k, n),
(n + k)(n + k 1)(n + k 2)
(k, n).
Furthermore,
(4x)fn1 (x)
=
=
=
X
(1)k (n 1 + k)!(a)k (4x)k+1
k=0
X
k=1
X
k=0
k!(2k)!(n 1 k)!
k1
(1)
(n 2 + k)!(a)k1 (4x)k
(k 1)!(2k 2)!(n k)!
k 2k(2k 1)
(k, n),
(n + k)(n + k 1)(a + k 1)
and
(4x)fn2 (x) =
X
(1)k (n 2 + k)!(a)k (4x)k+1
k=0
k!(2k)!(n 2 k)!
X
(1)k1 (n 3 + k)!(a)n1 (4x)k
k=1
or
(4x)fn2 (x) =
X
k=0
k(2k)(2k 1)(n k)
(k, n).
(n + k)(n + k 1)(n + k 2)(a + k 1)
1+A
nk
2k 2 (2k 1)
(n k)(n k 1)
B
+C
n+k
(n + k)(n + k 1)(a + k 1)
(n + k)(n + k 1)
2k 2 (2k 1)(n k)
(n k)(n k 1)(n k 2)
+E
0,
(n + k)(n + k 1)(n + k 2)(a + k 1)
(n + k)(n + k 1)(n + k 2)
or
(2)
(n+k)(n+k1)(n+k2)(a+k1)+A(nk)(n+k1)(n+k2)(a+k1)
2Bk 2 (2k 1)(n + k 2) + C(n k)(n k 1)(n + k 2)(a + k 1)
k =1a:
B=
a+n1
.
n
B(n 1 a) + D(n 1 + a) = 0
D(n 2) + E(a n + 1) = 0;
k=0:
and
coeff k 4 :
k=0:
Hence
2 + 2C 4B + 4D = 0.
Therefore we have
C
= 1 + 2B 2D
2E + 2n 2 2n 2 2a
= 1 +
+
n
n
3n 4
.
n
Finall,
A = 1 C E = 1
so
A=
Thus we obtain
3n 4 n 2
+
n
n
3n 2
.
n
127
= 3 F2 (n, n + 1, ; 1, p; v)
n
X
(1)k (n + k)!()k v k
=
(k!)2 (n k)!(p)k
k=0
X
(k, n).
=
k=0
Then
Hn1 =
X
(1)k ()k (n 1 + k)!v k
k=0
Hn2 =
X
nk
k=0
X
(n k)(n k 1)
k=0
Hn3 =
(n + k)(n + k 1)
n+k
(k, n),
(k, n),
X
(n k)(n k 1)(n k 2)
(k, n),
((n + k)(n + k 1)(n + k 2)
k=0
vHn1 =
X
k=0
vHn1 =
X
k=0
k 2 (p + k 1)
(k, n),
(n + k)(n + k 1)( + k 1)
and
vHn2 =
X
k=0
k 2 (p + k 1)(n k)
(k, n).
(n + k)(n + k 1)(n + k 2)( + k 1)
k =1 :
k=n:
B(n 1)
.
n+ 1
2(2n 1)(n + 1)
B=
n(n + p 1)
2(2n 1)(n 1)
D=
n(n + p 1)
(n 2)(p + 1 n)
E=
D
2(2n 3)( + 1 n)
(n 2)(2n 1)(p + 1 n)
=
n(2n 3)(n + p 1)
D=
or
n( n)(n + p 1)C
Hence
n(n + p 1)C
so
C=
Finally,
coeff k 4 : 1 A B + C + D E = 0.
Then
A=1B+C +DE
so
129
n
X
k=0
(1)k nxk
.
k!(1 + )k (1 + )k (n k)!
fn (x)
Put gn (x) =
. Then
n!
n
X
X
(1)k xk
gn (x) =
=
(k, n).
k!(1 + )k (1 + )k (n k)!
k=0
k=0
Now
gn1 (x) =
X
k=0
X
(1)k xk
=
(n k)(k, n)
k!(1 + )k (1 + )k (n 1 k)!
k=0
and
gn2 (x) =
(n k)(n k 1)(k, n)
k=0
gn3 (x) =
k=0
Also
xgn1 (x) =
X
k=0
X
(1)k xk+1
(1)k1 xk
=
k!(1 + ) k(1 + )k (n 1 k)!
(k 1)!(1 + )k1 (1 + )k1 (n k)!
k=1
or
xgn1 (x) =
k=0
k=n:
B=
Then
n(n 1)( + n)( + n)C
So
C=
3n 3 + +
.
n( + n)( + n)
Now gn (x) =
131
fn (x)
. Hence we get
n!
( + n)( + n)
fn1 (x) 3n 3 + +
fn3 (x)
fn (x)[3n2 3n+1+n1)(+)+x]
+
fn2 (x)
= 0,
(n 1)!
(n 1)!
(n 2)!
(n 3)!
or
( + n)( + n)fn (x) [3n2 3n + 1 + (2n 1)( + ) + x]fn1 (x)
+(n 1)[3n 3 + + ]fn2 (x) (n 1)(n 2)fn3 (x) = 0.
Problem 5. Define the polynomial wn (x) by
wn (x) =
n
X
(1)k n!Lk (x)
k=0
(k!)2 (n k)!
in terms of the Laguerre polynomial Lk (x). Show that wn (x) possesses the pure
recurrence relation
n2 wn (x)[(n1)(4n3)+x]wn1 (x)+(6n2 19n+16+x)wn2 (x)(n2)(4n9)wn3 (x)+(n2)(n3)wn4 (x) = 0.
Solution 5. Let
wn (x) =
n
X
(1)k n!Lk (x)
k=0
(k!)2 (n k)!
X
X
gk (x)
gk (x)
=
.
(n k)!
(n k)!
k=0
k=0
Then
n1 (x) =
n3 (x) =
X
k=0
X
k=0
(n k)
gk (x)
,
(n k)!
gk (x)
(n k)(n k 1)(n k 2)
,
(n k)!
n2 (x) =
n4 (x) =
(n k(N k 1)
k=0
gk (x)
,
(n k)!
k=0
gk (x)
.
(n k)!
X
k 3 (k 1)gk (x)
k=0
(n k)!
k=n:
coeff k 4 :
coeff k 3 :
X
k 3 (k 1)gk (x)
.
(4n 7)n3 (x) + n4 (x) =
(n k)!
k=0
From
n1 (x) =
X
k=0
gk (x)
,
(n 1 k)!
X
gk1 (x)
(n k)!
k=0
X
(n k)gk1 (x)
k=0
n3 (x) =
(n k)!
(n k)(n k 1)
k=0
gk1 (x)
, etc
(n k)!
(k 1)(2k 1)
k=0
gk1 (x)
.
(n k)!
133
X
(k 1)(2k 1)gk1 (x)
(n k)!
k=0
X
(k 1)gk1 (x)
(n k)!
k=0
we determine A1 and B1 by
A1 + B1 (n k) (k 1).
Then A1 = (n 1), B1 = 1. Hence
(n 1)n1 (x) + n2 (x) =
(3)
X
(k 1)gk1 (x)
(n k)!
k=0
Also
(4)
n2 (x) =
X
k=0
X gk2 (x)
gk (x)
=
.
(n 2 k)!
(n k)!
k=0
Since
X
k 3 (k 1)gk (x) + (k 1)(2k 1 x)gk1 (x) + gk2 (x)
k=0
(n k)!
= 0,
n
n + k;
X
(n)k (x)k
1
wn (x) =
.
1 F1
(k!)3
1 + k;
k=0
Solution 6. Consider
wn (x) =
n
X
(1)k n!Lk (x)
(k!)2 (n k)!
k=0
Let us form
=
=
=
=
=
X
wn (x)tn
n!
n=0
X
n
X
(1)k Lk (x)tn
(k!)2 (n k)!
n=0 k=0
(1)k Lk (x)tn+k
(k!)2 n!
n,k=0
X
X
(1)k+s xs tn+k
n,k=0 s=0
X
k,s,n=0
k!n!(s!)2 (k s)!
(1)k xs tn+k
(s!)2 k!k!(k + s)!
Then
n
X
X
(1)k xs tn+s
(s!)2 k!(n k)!(k + s)!
n,s=0 k=0
n;
X
xs tn+s
1
=
1 F1
n!(s!)3
n,s=0
1 + s;
X
n
n + s;
X
xs t n
1
=
.
1 F1
3
(s!) (n s)!
n=0 s=0
1 + s;
=
Hence
wn (x)
n
X
1 F1
1 F1
1 + s;
n + s;
s=0
s=0
n
X
n + s;
1 + s;
n!xs
s)!
(s!)3 (n
(n)s (x)
1
,
(s!)3
as desired.
Problem 7. Define the polynomial vn (x) by (see Section 131, page 251)
vn (x) =
n
X
(1)k n!Pk (x)
k=0
(k!)2 (n k)!
135
in terms of the Legendre polynomial Pk (x). Show that vn (x) satisfies the recurrence relation
n2 vn (x)[4n2 5n+2(2n1)x]vn1 (x)+[6n2 15n+11(4n5)x]vn2 (x)(n2)(4n72x)vn3 (x)
+(n 2)(n 3)vn4 (x) = 0.
Solution 7. Consider
vn (x) =
n
X
(1)k n!Pk (x)
k=0
(k!)2 (n k)!
vn (x)
(1)k Pk (x)
= gk (x).
= n (x) and
n!
(k!)2
We know that Pk (x) satisfies the relation
Put
Also
(2)
n (x) =
X
gk (x)
.
(n k)!
k=0
Because of the identity of form of equation (2) with the corresponding relations
of Exercise 5, we may write immediately
(3) n3 (n 1)n (x) (n 1)(4n2 5n + 2)n1 (x) + (6n2 15n + 10)n2 (x)
(4n 7)n3 (x) + n4 (x) =
X
k 3 (k 1)
k=0
(n k)!
gk (x).
(4)
X
(k 1)(2k 1)gk1 (x)
(n k)!
k=0
and
(5)
n2 (x) =
X
gk2 (x)
.
(n k)!
k=0
Since
X
k 3 (k 1)gk (x) + (k 1)(2k 1)xgk1 (x) + gk2 9x)
k=0
(n k)!
= 0,
Solution 8.
Problem 9. Let
(1)k ( 21 )nk (2x)n2k
,
k!(n 2k)!
so that the Legendre polynomial of Chapter 10 may be written
(k, n) =
Pn (x) =
(k, n).
k=0
Show that
xPn1 (x) =
X
(n 2k)(k, n)
k=0
xPn0 (x) =
0
Pn1
(x)
2n 2k 1
(n 2k)(k, n),
k=0
X
2k(k, n)
,
2n 2k 1
k=0
X
0
Pn+1
(x) =
(1 + 2n 2k)(k, n),
Pn2 (x) =
k=0
2k(k, n),
etc.
k=0
Use Sister Celines method to discover the various differential recurrence relations and the pure recurrence relation for Pn (x).
Solution 9. Let (x, n) =
Then
Pn (x) =
X
k=0
(k, n)
xPn1 (x) =
X
(1)k ( 1 )n1k (2x)n2k
2
k=0
2 k!(n 1 2k)!
137
n 2k
(k, n)
2(n k 12
k=0
or
xPn1 (x) =
X
k=0
n 2k
(k, n).
2n 2k 1
Next
Pn2 (x)
=
=
X
(1)k ( 1 )n2k (2x)n22k
2
k=0
X
k=1
k!(n 2 2k)!
(1)k1 ( 21 )n1k (2x)n2k
,
(k 1)!(n 2k)!
so that
Pn2 (x) =
X
k
2k
(k,
n)
=
(k, n).
1
2n
2k 1
n
2
k=0
k=0
Now
xPn0 (x) =
X
(1)k ( 1 )nk (2x)n2k
2
k!(n 1 2k)!
k=0
(n 2k)(k, n),
k=0
and
0
Pn+1
(x) =
X
(1)k ( 1 )n+1k 2(2x)n2k
2
k=0
k!(n 2k)!
k=0
Finally,
0
Pn1
(x)
=
=
=
X
9 1)k ( 1 )n1k 2(2x)n22k
2
k=0
X
k=1
k!(n 2 2k)!
(1)k1 ( 21 )N k 2(2x)n2k
(k 1)!(n 2k)!
2k(k, n).
k=0
Problem 10. Apply sister Celines method to discover relations satisfied by the
Hermite polynomials of Chapter 11.
Solution 10.
Problem 11. Find the various relations of Section 114 on Laguerre polynomials
by using Sister Celines technique.
Solution 11.
Problem 12. Consider the pseudo-Laguerre polynomials (Boas and Buck [2;16])
fn (x) defined for nonintegral by
n
fn (x) =
X ()nk xk
()n
.
1 F1 (n; 1 + n; x) =
n!
k!(n k)!
k=0
Show that the polynomials fn (x) are not orthogonal with respect to any weight
function over any interval because no relation of the form
fn (x) = (An + Bn x)fn1 (x) + Cn fn2 (x)
is possible. Obtain the pure recurrence relation
nfn (x) = (x + n 1 )fn1 (x) xfn2 (x).
Solution 12. Consider fn (x) defined by
n
X ()nk xk
()n
.
fn (x) =
1 F1 (n; 1 + n; x) =
n!
k!(n k)!
k=0
Put
()nk xk
.
k!(n k)!
(k, n) =
Then
fn (x) =
(k, n)
k=0
fn1 (x) =
fn2 (x) =
X
X
()n1k xk
nk
=
(k, n)
k!(n 1 k)!
+n1k
k=0
k=0
k=0
(n k)(n k 1)
(k, n)
( + n 1 k)( + n 2 k)
and
xfn1 (x) =
X
()n1k xk+1
k=0
xfn2 (x) =
X
k=0
k!(n 1 k)!
()n2k xk+1
=
k!(n 2 k)!
X
k=0
X
k=1
X
()nk xk
=
k(k, n)
(k 1)!(n k)!
()n1k xk
=
(k 1)!(n 1 k)!
k=0
X
k=0
k(n k)
(k, n).
+ n 1 k
Now consider
Tn fn (x) An fn1 9x) Bn xfN 1 (x) Cn fn2 (x).
At once
X
Tn
1 An
k=0
nk
(n k)(n k 1)
Bn k Cn
(k, n).
+ n 1 k
( + n 1 k)( + n 2 k)
(1)
139
nk
k(n k)
Fn
= 0,
+ n 1 k
+ n 1 k
or
(2)
+ n 1 k Dn k( + n 1 k) En (n k) Fn k(n k) 0.
1
1 Dn n( 1) = 0; Dn = or = 1
n
1
coeff k 2 :
Dn + Fn = 0; Fn =
n
+ n 1
k = + n 1 : ( 1)En + ( 1)Fn ( + n 1) = 0; En =
.
n
Hence, either = 1 or
k=n:
In Exercises 1-8, Hn (x), Pn (x), Ln (x) denote the Hermite, Legendre, and simple
Laguerre polynomials, respectively. Derive each of the stated symbolic relations.
Problem 1.
Hn (x + y) + [H(x) + 2y]n .
Solution 1. Consider
X
Hn (x + y)tn
n!
n=0
= exp[2(x + y)t t2 ]
2
= exp(2yt) exp(2xt
! t )
!
n
n
X Hn (x)tn
X (2y) t
=
n!
n!
n=0
n=0
n
nk
X
X
(2y)
Hk (x) n
t .
=
)
n=0
k!(n k)!
k=0
It follows that
Hn (x + y) =
n
X
n!(2y)nk Hk (x)
k!(n k)!
k=0
or its equivalent
Hn (x + y) + {H(x) + 2y}n .
Problem 2.
n
X
[H(x) + H(y)]n tn
n!
n=0
n
X
X
Hk (x)Hnk (y)tn
k!(n k)!
!
!
X Hn (y)tn
Hn (x)tn
=
n!
n!
n=0
n=0
2
= exp(23xt t ) exp(2yt t2 )
2
= exp[2(x
+ y)t
2t ]
x+y
= exp 2
( 2t) ( 2t)2 .
2
n=0 k=0
It follows that
n n
[H(x) + H(y)] t
+
n!
n=0
n=0
Hn
x+y
( 2t)n
,
or
n
{H(x) + H(y)} + 2
1
2n
Hn
x+y
.
Problem 3.
n
1x
1+x
Pn (x) + n! L
L
2
2
Solution 3. Consider
X
t(x 1)
t(x + 1)
Pn (x)tn
=
F
;
1;
F
;
1;
.
0 1
0 1
(n!)2
2
2
n=0
Now
X
Ln (y)tn
e 0 F1 (; 1; yt) =
.
n!
n=0
t
Hence
X
Pn (x)tn
(n!)2
n=0
t(x 1)
t(x + 1)
t
= e 0 F1 ; 1;
e 0 F1 ; 1;
2
2
!
!
1+x
1x
X (1)n Ln
X Ln
tn
tn
2
2
=
n!
n!
n=0
n=0
X
n
1+x
1x
k
X
(1) Lk ( 2 )Lnk 2
=
tn .
k!(n
k)!
n=0
t
k=0
Hence
Pn (x) = n!
n
X
n!(1)k L ( 1+x )Lnk ( 1x )
2
k!(n k)!
k=)
or
Pn (x) + n! L
Problem 4.
Hn
1x
2
1+x
2
n
.
1
P (x) + [H(x) P (x)]n .
2
Solution 4. Consider
X
Hn
n=0
1
2 P (x)
n!
tn
[2]
X
X
(1)k Pn2k (x)tn
k!(n 2k)!
!
!
X Pn (x)tn
(1)n t2n
=
n!
n!
n=0
n=0
2 2
t
(x
1)
2
= et ext 0 F1 ; 1;
! 4
!
X
X (1)n Pn (x)tn
Hn (x)tn
=
n!
n!
n=0
n=0
X
n
X
(1)k Pk (x)Hnk (x)tn
.
=
k!(n k)!
n=0
n=0 k=0
k=0
It follows that
141
Hn
1
P (x)
2
n
X
(1)k n!Pk (x)Hnk (x)
k!(n k)!
n
+ {H(x) P (x)} .
k=0
Problem 5.
[H(x) 2P (x)]2n+1 + 0,
2n
[H(x) 2P (x)]
n 2n
+ (1) 2
1
Ln (x2 1).
2 n
Solution 5. Consider
=
+
X
[1 + (x) 2P (x)]n tn
n!
n=0
n
X
k
X
(2) Pk (x)Hnk (x)tn
k!(n k)!
!
!
X Pn (x)(2t)n
Hn (x)t
=
n!
n!
n=0
n=0
4t2 (x2 1)
2
= exp(2xt t ) exp(2xt)0 F1 ; 1;
4
= exp(t2 )0 F1 (; 1; t2 (x2 1))
X
(1)n Ln (x2 1)t2n
=
.
n!
n=0
n=0 k=0
It follows that
[H(x) 2P (x)]2n+1 + 0
and
[H(x) 2P (x)]2n +
(1)n (2n)!
Ln (x2 1) = (1)n 22n
n!
1
Ln (x2 1).
2 n
Problem 6.
[H(x) P (2x)]2n+1 + 0,
2n
[H(x) P (2x)]
Solution 6. Consider
n 2n
+ (1) 2
1
1
2
Ln x
.
2 n
4
X
[H(x) P (2x)]n tn
n!
n=0
143
X
n
X
(1)k Pk (2x)Hnk (x)tn
k!(n k)!
!
!
X Pn (2x)(t)n
Hn (x)t
=
n!
n!
n=0
n=0
t2 (4x2 1)
2
= exp(2xt t ) exp(2xt)0 F1 ; 1;
4
1
= exp(t2 )0 F1 ; 1; (t2 )(x2 )
4
X
(1)n Ln (x2 41 )t2n
=
.
n!
n=0
n=0 k=0
[H(x) P (2x)]2n +
1
1
L n x2
.
2 n
4
Problem 7.
Hn
1
H(x)
2
+ 2 2 Hn (2 2 x).
Solution 7. From
X
Hn ( 12 H(x))tn
n!
n=0
[2]
X
X
(1)k Hn2k (x)tn
k!(n 2k)!
!
!
X Hn (x)tn
(1)n t2n
=
n!
n!
n=0
n=0
= exp(t2 ) exp(2xt t2 )
2
= exp(2xt
2t)
2
x
= exp 2
2t ( 2t)
2
x
H (
X
)( 2t)n
n
2
=
n!
n=0
n=0 k=0
n
x
1
H(x) + 2 2 Hn
.
2
2
Problem 8.
Hn (xy) + [H(x) + 2x(y 1)]n .
Solution 8. From
X
Hn (xy)tn
n!
n=0
= exp(2xyt t2 )
= exp(2xt t2 ) exp[2x(y
! 1)t]
!
X [2x(y 1)]tn
X
Hn (x)tn
=
n!
n!
n=0
n=0
X
n
k
X
[2x(y 1)] Hnk (x)tn
=
,
k!(n k)!
n=0
k=0
it follows that
Hn (xy) + [H9x) + 2x(y 1)]n .
Problem 9. Use Laplaces first integral for Pn (x) to derive equation (20), preceding
these exercises.
Solution 9. We know that
Pn (x) =
[x +
x2 1 cos ]n d.
n
X
(1)k n!Pk (x)
k=0
(k!)n (n k)!
Hence
vn (x)
Z
n
1 X (1)k n![x + x2 1 cos
d
0
(k!)2 (n k)!
k=0
Z
p
1
=
Ln x + x2 1 cos d.
0
=
Problem 10. For the vn (x) of equation (15), page 251, evaluate
Z
vm (x)Pk (x)dx,
1
and use your result to establish equation (21), preceding these exercises.
Z 1
Solution 10. For the vn (x) of Exercise 9 we wish to evaluate
vm (x)Pk (x)dx.
1
Now
Z
vm (x)Pk (x)dx =
1
Pk (x)
1
m
X
(1)s m!Ps (x)dx
s=0
(s!)2 (m s)!
If m k, we get
m
X
Z 1
(1)s m!
=
Pk (x)Ps (x)dx
(s!)2 (m s)! 1
s=0
Z
1
(1)k m!
=
P 2 (x)dx
(k!)2 (m k)! 1 k
(1)k 2m!
=
.
(k!)2 (m k)!(2k + 1)
vm (x)Pk (x)dx
1
145
Now consider
vN (x)vm (x)dx.
1
Let m n. Then
1
vn (x)vm (x)dx =
1
vn (x)
1
m
X
(1)k m!Pk (x)
k=0
(k!)2 (m k)!
dx.
For k > n the integrals involved in the sum are zero. We may therefore write
Z
vn (x)vm (x)dx
n
X
k=0
n
X
k=0
n
X
(1)k m!
(k!)2 (m k)!
vn (x)Pk (x)dx
1
(1)k m!
(1)k 2 n!
(k!)2 (m k)! (k!)2 (n k)!(2k + 1)
(n)k (m)k
(k!)4 (k + 21 )
k=0
n
X (n)k (m)k ( 1 )k
2
=2
.
4( 3 )
(k!)
2 k
k=0
=
1
vn (x)vm (x)dx = 23 F4
1
3
1, 1, 1, ;
2
Problem 11. Show that
yx
n
(y + x)n Pn
+ n! {L(x) L(y)} .
y+x
n
X
(y + x)n Pn ( yx
y+x )t
n=0
(n!)2
0 F1 ; 1;
2
= 0 F1 (; 1; xt)0 F1 (; 1; yt)
t
= et 0 F1 (; 1; xt)e
0 F1 (; 1; y(t))!
!
X Ln (x)t4
X (1)n Ln (y)tn
=
n!
n!
n=0
n=0
X
n
k
X
(1) Lk (y)Lnk (x)tn
=
.
k!(n k)!
n=0
= 0 F1
; 1;
t(y + x)[ yx
y+x 1]
k=0
t(y + x)[ yx
y+x + 1]
2
(y + x) Pn
yx
y+x
= n!
n
X
(1)k n!Lk (y)Lnk (x)
k!(n k)!
k=0
n
+ n! {L (x) L (y)} ,
as desired.
Problem 12. Define polynomials n (x, y) by
n (x, y) + Hn (xL(y)) .
Show that
X
n (x, y)tn
= exp(2xt t2 )0 F1 (; 1; 2xyt).
n!
n=0
Then
X
n (x, y)tn
n!
n=0
X
(1)k (2x)n Ln (y)tn+2k
k!n!
n,k=0
X Ln (y)(2xt)n
= exp(t2 )
n!
n=0
= exp(t2 ) exp(2xt)0 F1 (; 1; 2xgt)
= exp(2xt t2 )0 F1 (; 1; 2xyt),
as desired.
18. Chapter 16 Solutions
Problem 1. Let
1x
1+x
gn (x) =
.
(1 + )n (1 + )n
Use Batemans generating function, page 256, to see that
(,)
(1 + x)n Pn
gn (x)tn = 0 F1 (; 1 + ; xt)0 F1 (; 1 + ; t)
n=0
and thus show that, in the sense of Section 126, gn (x) is of -type zero with
= D( + ). Show also that gn (x) is of Sheffer A-type unity.
147
Solution 1. Let
gn (x) =
(,) 1x
( 1+x )
(1 + x)n Pn
(1 + )n (1 + )n
Then
X
n=0
gn (x)tn
(,) 1x
X
Pn
( 1+x )[t(1 + x)]n
(1 + )n (1 + )n
h
i
1x
t(1
+
x)
+
1
1+x
t(1 + x) 1 x
; 1 + ;
= 0 F1 ; 1 + ;
1
0 F1
2
1+x
2
n=0
= 0 F1 (; t + ; xt)0 F1 (; 1 + ; t).
We now see that with = D(+), gn (x) is of -type zero with A(t) = 0 F1 (; 1+
; t), H(t) = t, J(t) = t.
Since gn (x) = gn1 (x), it follows that in the Sheffer classification notation
J(x, D)
= D( + )
= D(xD + )
= (1 + )D + xD 2 .
Thus T0 (x) = 1 + , T1 (x) = x, Tk (x) 0 for k 2. So gn (x) is of Sheffer
A-type unity.
(,)
(x) because of the fact that
It might be of interest to see what we get on Pn
gn (x) is of -type zero.
Problem 2. Show that
(,)
(,)
d
,
dx
(,)
(1)
(2)
(x+1)(++n)DPn(,) (x)(x+1)(+n)DPn1
(,beta)
(,)
(,)
(,)
or
0
nPn (x) = xPn0 (x) Pn1
(x)
(,)
(,)
(,)
(,)
(,)
149
Pn(,) (x) =
n
(,0)
(x)
(1 + )n X (1)nk ()nk (1 + + )n+k (1 + + 2k)Pk
.
(1 + + )n
(n k)!(1 + )n+k+1
k=0
(,)
(,)
1x
s
s
= (1 + )s
(x).
s
(,0)
X
(1 + + 2k)(1 + )k P
(x) (1)k s!
k
(s k)!
(1 + )s+1+k (1 + )k
k=0
(1 + + )n Pn
(1 + )n
(2)
(x)
n
X
(1)s (1 + + )n+s ( 1x )s
s
s=0
s!(n s)!(1 + )s
(,)
X
(x)tn
(1 + + )n Pn
(1 + )n
n=0
s n+s
X
(1)s (1 + + )n+2s ( 1x
2 ) t
=
s!n!(1 + )s
n,s=0
(x, t)
=
=
=
X
s
(,0)
X
(1)k s!(1 + + 2k)(1)s (1 + + )n+2s P
(x)tn+s
k
n,s=0 k=0
(,0)
n,k=0 s=0
Then
(x, t)
=
=
=
=
X
n,k=0
X
n,k=0
2 F1
n, 1 + + + n + 2k;
(,0)
(1 + + )n+2k (1 + + 2k)Pk
1
n!(1 + )2k+1
2 + + 2k;
(,0)
(2 + + 2k)(1 )(1 + + )n+2k (1 + + 2k)Pk
(x)tn+k
(2 + + 2k + n)(1 n)n!(1 )2k+1
(,0)
n,k=0
X
n
X
n=0 k=0
(,0)
(x)tn
(x)tn+k
(x)tn+k
Pn(,) (x) =
n
(,0)
(x)
(1 + )n X (1)nk ()nk (1 + + )n+k (1 + + 2k)Pk
.
(1 + + )n
(n k)!(1 + )n+k+1
k=0
(,)
(1 x2 ) Pn(,) (x)Pk
(x)dx.
Pn(,) (x) =
n
(,)
(x)
(1 + )n X (1)nk ( )nk (1 + |alpha + )n+k (1 + 2)k (1 + 2 + 2k)Pk
.
(1 + + )n
(n k)!(1 + 2)n+k+1 (1 + )k
k=0
(,)
(x) we get
1
(,)
(1 x2) Pn(,) (x)Pm
(x)dx = 0, m 6= n
(1 x2 ) Pn(,) (x)]2 dx =
21+2 (1 + + n)(1 + + n)
.
n!(1 + 2 + 2n)(1 + 2 + n)
(,)
=
1
(1 x2 ) Pn(,) (x)Pk
(1 + )n
(1 + + )n
n
X
s=0
(x)dx
(,)
(1 x2 ) Ps(,) (x)Pk
(x)dx.
A(k, n) =
or
A(k, n)
for 0 k n.
151
(,0)
(1 x) Pn(,) (x)Pk
Z
(x)dx.
(,0)
(1 x) Pn(,) (x)Pk
(x)dx.
We know that
= 0; k 6= s
21+ (1 + + k)(1 + k)
21+
=
=
;s = k
k!(1 + + 2k)(1 + + k)
1 + + 2k
(,0)
(1x) Pk
(x)Ps(,0) (x)dx
(1x)
1
(,0)
Pn(,) (x)Pk
(x)dx
Z
n
(1 + )n X (1)ns ()ns (1 + + )n+s (1 + + 2s) 1
(,0)
(1x) Ps(,0) (x)Pk
(x)dx.
=
(1 + + )n s=0
(n s)!(1 + )n+s+1
1
(,0)
(1 x) Pn(,) (x)Pk
(x)dx = 0, k > n.
If 0 k n, each integral in the sum is zero except for the one in which s = k.
Hence
Z
(,0)
(x)dx =
2 F1
a, b;
a, b;
v 2 F1
1 + a + b c;
c;
v
to conclude that
1v
1
1
a, b, (a + b), (a + b + 1);
2
2
v = 4 F3
4v(1 v) .
a + b, c, 1 + a + b c;
Solution 8. Theorem 84 is
a, b;
2 F1
a, b;
2 F1
1x
c;
Now put x = y =
y
y
x
,
.
= F4 a, b; c, 1 c + a + b;
1y
(1 x)(1 y) (1 x)(1 y)
1 c + a + b;
x
. Then the above becomes
1x
2 F1
a, b;
a, b;
v 2 F1
1 c + a + b;
c;
=
=
X
(a)n+k (b)n+k v n+k (1 v)n+k
k!n!(c)k (1 c + a + b)n
n,k=0
n
XX
(a)n (b)n v n (1 v)n
n, c a b n;
(a)n (b)n v n (1 v)n
1
=
2 F1
n!(1 c + a + b)n
n=0
c;
X
(a + b)2n (a)n (b)n v n (1 v)n
=
(c) (a + b)n n!(1 c + a + b)n
n=0 n
a+b a+b+1
,
;
a, b,
2
2
= 4 F3
4v(1 v) ,
n=0 k=0
c, a + b, 1 c + a + b;
as desired.
Problem 9. Use the result in Exercise 8 above, and Theorem 25, page 67, to show
that
2a, 2b, a + b;
y
y .
= 3 F2
2 F1
1
1
a+b+ ;
2a + 2b, a + b + ;
2
2
Solution 9. In Exercise 8 above replace a by (2a), b by (2b), and then put c =
1
1
1
a + b + to get (since 1 + 2a + 2b a b = a + b + )
2
2
2
1
2a, 2b;
2a, 2b, a + b, a + b + ;
v
4v(1 v)
= 4 F3
2 F1
1
1
a+b+ ;
2a + 2b, a + b + , a + b + ;
2
2
2
2a, 2b, a + b;
4v(1 v) .
= 3 F2
1
2a + 2b, a + b + ;
2
By Theorem 25, page 114, we may now write
2 F1
or
a, b;
a, b;
1
a+b+ ;
2
2a, 2b, a + b;
4v(1 v) = F
3 2
1
2a + 2b, a + b + ;
2
4v(1 v) ,
F1
2a, 2b, a + b;
y = F
3 2
1
2a + 2b, a + b + ;
2
a, b;
1
a+b+ ;
2
153
y .
Problem 1. Show that the Gegenbauer polynomial Cnv (x) and the Hermite polynomial Hn (x) are related by
n
Cnv (x)
[2]
X
2 F0 (k, v
+ n k; ; 1)
k=0
Solution 1. From
n
Cnv (x)
[2]
X
(1)s (v)ns (2x)n2s
s!(n 2s)!
s=0
we get
Cnv (x)tn =
n=0
X
(1)s (v)n+s (2x)n tn+2s
.
s!n!
n,s=0
We know that
[n]
2
X
(2x)n
Hn2k (x)
=
.
n!
k!(n 2k)!
k=0
Then
Cnv (x)tn
n=0
=
=
[2]
X
X
(1)s (v)n+s Hn2k (x)tn+2s
n,s=0 k=0
s!k!(n 2k)!
n,k,s=0
X
k
X
n,k=0 s=0
X
k
X
k, v + n + k;
X
(1)k (v)n+k Hn (x)tn+2k
1
=
2 F0
k!n!
;
n,k=0
n
[2]
X
X
(1)k (v)nk Hn2k (x)tn
=
.
2 F0 (k, v + n k; ; 1)
k!(n 2k)!
n=0
=
k=0
Hence
Cnv (x)
[2]
X
2 F0 (k, v
+ n k; ; 1)
k=0
2
v
(v + n 2k)Cn2k
(x)
Hn (x) X
=
.
(1)k 1 F1 (k; 1 + v + n 2k; 1)
n!
k!(v)n+12k
k=0
Solution 2. Consider
X
X
(1)s (2x)n tn+2s
Hn (x)tn
=
.
n!
s!n!
n,s=0
n=0
2
X
(2x)n
(v + n 2k)Cn2k (x)
=
.
n!
v k!(v)n+1k
k=0
Hence
n
X
Hn (x)tn
n!
n=0
[2]
X
v
X
(1)s (v + n 2k)Cn2k
(x)tn+2s
=
=
=
=
=
n,s=0 k=0
s!k!(v)n+1k
n,k,s=0
X
k
X
n,k=0 s=0
X
k
X
n,k=0 s=0
1 F1 (k; 1
n,k=0
+ v + n; 1)
Then
[n]
2
v
(1 )k (v + n 2k)Cn2k
(x)
Hn (x) X
=
.
1 F1 (k; 1 + v + n 2k; 1)
n!
k!(v)n+12k
k=0
Problem 3. Show, using the modified Bessel function of Section 65, that
e
xt
=
v
X
1
t
(v)
(v + n)Iv+n (t)Cnv (x).
2
n=0
( 2t )v+n
t2
0 F1 ; 1 + v + n;
(1 + v + n)
4
X
( 2t )v+n+2k
=
k!(1 + v + n + k)
Iv+n (t)
k=0
Now
[n]
2
v
X
(v + n 2k)Cn2k
(x)
xn
=
.
n
n!
2 k!(v)n+1k
k=0
Hence
C xt
X
xn t n
n!
n=0
n
=
=
[2]
X
v
X
(v + n 2k)Cn2k
(x)tn
2n k!(v)n+1k
n=0 k=0
(v + n)Cnv (x)tn+2k
2n+2k k!(v)n+1+k
n,k=0
XX
( t )n+2k (v)
2
(v + n)Cnv (x)
k!(v
+
n
+
1
+
k)
n=0 k=0
v X
( 2t )n+v+2k
t
(v + n)Cnv (x).
= (v)
2
k!(v
+
n
+
1
+
k)
n=0
Therefore
e
xt
v X
t
= |Gamma(v)
(v + n)Iv+n (t)Cnv (x).
2
n=0
Cnv (x)
[2]
X
(v 12 )k (v)nk (1 + 2n 4k)Pn2k (x)
=
.
k!( 23 )nk
k=0
Solution 4. From
n
Cnv (x)
[2]
X
(1)k (v)nk (2k)n2k
k=0
k!(n 2k)!
and
[n]
2
X
(2x)n
(2n 4x + 1)Pn2k (x)
=
n!
k!( 32 )nk
k=0
we obtain
155
Cnv (x)tn
n=0
[2]
X
X
(1)s (v)ns (2x)n2s tn
s!(n 2s)!
n=0 s=0
=
=
=
[2]
X
X
(1)s (v)n+s (2n + 1)Pn (x)tn+2k+2s
s!k!( 32 )n+k
n,s=0 k=0
n,k=0 s=0
k
X
X
k, v + n + k;
k
n+2k
X
2 F1
3
k!( 32 )n
n,k=0
+ n;
2
Therefore we get, using Example 5, page 69,
=
Cnv (x)tn
n=0
X
(1)k ( 23 )n (v 12 )k (1)k (v)n+k (2n + 1)Pn (x)tn+2k
( 32 )n+k
k!( 32 )n
n,k=0
X
(v 1 )k (v)n+k (2n + 1)Pn (x)tn+2k
2
k!( 32 )n+k
n,k=0
n
[2]
Hence
n
Cnv (x)
[2]
X
(v 12 )k (v)nk (2n 4k + 1)Pn2k (x)
=
.
k!( 23 )nk
k=0
x =
n
X
k=0
n!Bk (x)
.
k!(n k + 1)!
X
text
Bn (x)tn
=
.
et 1 n=0
n!
157
Then
e
xt
X
1
Bn (x)tn
= (et 1)
=
t
n!
n=0
n1
X
t
n!
n=1
X
Bn (x)tn
n!
n=0
!
,
from which
X
xn tn
n!
n=0
!
!
X Bn (x)tn
tn
=
(n + 1)!
n!
n=0
n=0
X
n
n
X
Bk (x)t
=
k!(n
+ 1 k)!
n=0
k=0
Hence
n
x =
n
X
k=0
n!Bk (x)
.
k!(n + 1 k)!
Problem 2. Let Bn (x) and Bn = Bn (0) denote the Bernoulli polynomials and
numbers as t reated in Section 153. Define the differential operator A(c, D) by
A(x, D) =
X
1
B2k D2k
x
D
.
2
(2k)!
k=1
X
text
Bn (x)tn
=
t
e 1 n=0
n!
Bn (x)
is of Sheffer A-type zero. We have, in the Shefer notation,
n!
t
H(t) = t, J(t) = t, A(t) = t
.
e 1
We wish to apply Theorem 74, page 391. Now
we see that
X
tA0 (t)
t
Bn t n
=1t t
=1t
.
A(t)
e 1
n!
n=0
1
We know that B0 = 1, B1 = , B2n+1 = 0 for n 1. Hence
2
tA0 (t)
t X B2k t2k
=
.
A(t)
2
(2k)!
k=1
X
k=0
t X B2k t2k
k tk=1 =
,
2
(2k)!
k=1
vk tk+1 = tH 0 (t) = t.
k=0
1
Therefore v0 = 1, vk = 0 for k 1, and 0 = , 2k = 0 for k 1, 2k1 =
2
B2k
for k 1.
(2k)!
The identity
k=0
of Theorem 74 becomes
#
"
X
nBn (x)
1
B2k D 2k Bn (x)
=
.
x
D
2
(2k)!
n!
n!
k=1
Hence, with
X
1
B2k D 2k
A(x, D) = x
D
,
2
(2k)!
k=1
(k + xvk )D k+1
k=0
may be replaced by
(k + xvk )D k+1 .
k=0
n (c, x, y) =
(1)n ( 21 + 21 x)n
3 F2
(c)n
n,
1 1
x, 1 c n;
2 2
c,
1 1
x n;
2 2
y
.
Show that
1 F1
X
1
1
n (c, x, y)tn
1 1
x; c; yt 1 F1
= x; c; t =
,
2 2
2
2
n!
n=0
1 1 1 1
x, + x;
2 2 2 2
12 + 12 x
12 21 x
(1yt)
(t+t)
2 F1
159
X
(c)n n (c, x, y)tn
yt2
,
=
n!
(1 yt)(1 + t) n=0
c;
and that n (1, x, 1) = Fn (x), where Fn is Batemans polynomial of Section 148.
Solution 3. We define n (c, x, y) by
n (c, x, y) =
(1)n ( 1+x
2 )n
3 F2
(c)n
n,
1x
, 1 c n;
2
c,
y
.
1x
n;
2
Then
n (c, x, y) =
n
X
(1)n+k n!( 1+x )n ( 1x )(c)n ( 1+x )nk y k
2
k=0
or
n (c, x, y) =
n
X
(1)n+k n!( 1x )k ( 1+x )nk y k
2
k=0
X
n (c, x, y)tn
n!
n=0
X
n
X
(1)nk ( 1x )k ( 1+x )nk y k tn
2
1x
; 1; t
2
1 F1
1+x
; 1; t
2
=
X
n (1, x, 1)tn
.
n!
n=0
Brafman had
1 F1
X
1x
1+x
Fn (x)tn
; 1; t 1 F1
; 1; t =
.
2
2
n!
n=0
Hence
n (1, x, 1) = Fn (x).
X
n
1+x
k n
X
X
(1)n+k (c)n ( 1x
(c)n n (c, x, y)tn
2 )k ( 2 )nk y t
=
n!
k!(n k)!(c)k (c)nk
n=0
n=0 k=0
n
X (1) (c)n+k ( 1x )k ( 1+x )n y k tn+k
2
2
=
k!n!(c)k (c)n
n,k=0
X
k
n
n 1+x
X
(c + n)k ( 1x
2 )k (yt) (1) ( 2 )n t
=
k!(c)k
n!
n=0 k=0
1x
c + n,
;
n 1+x
n
X
(1) ( 2 )n t
2
=
F
2 1
yt
n!
n=0
c;
1x
n,
;
(1)n ( 1+x ) tn
X
2
1x
2 n
yt
(1 yt) 2 2 F1
=
n!
1 yt
n=0
c;
n
X
k k k
n 1+x
n
X
(1)k n!( 1x
1x
2 )k (1) y t (1) ( 2 )n t
= (1 yt) 2
k!(n k)!(c)k (1 yt)k
n!
n=0 k=0
1x
1+x
n+k
k
n+2k
X (
y t
1x
2 )k ( 2 )n+k (1)
= (1 yt) 2
k
k!n!(c)k (1 yt)
n,k=0
X
1+x
1+x
k 2k
X
( 2 + k)n (t)n (1)K ( 1x
1x
2 )k ( 2 )k y t
= (1 yt) 2
n!
k!(c)k (1 yt)k
k=0 n=0
1x
1+x
k k 2k
X(
1+x
1x
2 )k ( 2 )k (1) y t
= (1 yt) 2 (1 + t) 2
k!(c)k (1 + t)k (1 yt)k
k=0
We thus arrive at
1x 1+x
,
;
2
2
X
1+x
(c)n n (c, x, y)tn
1x
= (1yt) 2 (1+t) 2 2 F1
n!
n=0
yt2
.
(1 yt)(1 + t)
c;
Now put c = 1, y = 1 in the above. We get
1x 1+x
,
;
2
2
1+x
1x
t2
= (1 t) 2 (1 + t) 2 1 F1
1 t2
1;
1+x 1+x
,
;
1+x
1+x
2
2
2
1x
= (1 t) 2 (1 + t) 2 (1 t ) 2 2 F1
X
n=0
n (1, x, 1)tn
1;
1+x 1+x
,
;
2
2
x
= (1 t) 2 F1
1;
t2 .
t2
1+x
1
, b = to get
2
2
1+x 1
, ;
2
2
1x
F
2 = (1 t)
2
1
(t)
1;
1+x 1+x
,
;
2
2
F
1;
161
4t
.
(1 t)2
2
2
X
n
1
n (1, x, 1)t = (1 t) 2 F1
n=0
1;
X
4t
=
Fn (x)tn ,
2
(1 t)
n=0
n (x)tn ,
n=0
(1 xt)c 2 F0 c, x; ;
t
1 xt
(c)n n (x)tn .
n=0
Solution 4. Consider
n (x) =
xn
1
F
.
n,
x;
;
2 0
n!
x
THen
n (x) =
n
X
(1)k n!xn (x)k (1)k xk
k=0
k!(n k)!n!
n
X
(x)k xnk
=
.
k!(n k)!
k=0
Hence
X
n=0
Next consider
n (x)tn
X
n
X
(c)k xnk tn
k!(n k)!
!
!
X xn t n
(x)n tn
=
n!
n!
n=0
n=0
x xt
= (1 t) e .
n=0 k=0
(c)n n (x)tn
X
n
X
(c)n (x)k xnk tn
n=0
k!(n k)!
n=0 k=0
X
(c + k)n (xt)n (c)k (x)k tk
=
=
n!
k=0 n=0
k=0
k!
(c)k (x)k t
.
k!(1 xt)c+k
Therefore
(c)n n (x)tn = (1 xt)c 2 F0 c, x; ;
n=0
t
1 xt
.
Problem 5. For Sylvesters polynomials of Exercise 4 find what properties you can
from the fact that n (x) is of Sheffer A-type zero.
Solution 5. For the n (x) of Exercise 4 recall that
(1 t)x ext =
n (x)tn .
n=0
n=0
Therefore n (x) are of Sheffer A-type zero with A(t) = 1, H(t) = t log(1 t).
Problem 6. Show that Batemans Zn (x), the Legendre plynomial Pn (x), and the
Laguerre polynomial Ln (x) are related symbolically by
Zn (x) + Pn (2L(x) 1).
Solution 6. We know that Batemans Zn (x) has the generating relation
(1)
Zn (x)t = (1 t)
n=0
1 F1
1
4xt
; 1;
2
(1 t)2
(c)n Ln (x)tn
xt
(1 t)c 1 F1 c; 1;
=
,
1t
n!
n=0
including the special case
X
( 21 )n Ln (x)v n
1
xv
; 1;
=
.
2
1v
n!
n=0
2
4t
1t
v
4t
In (2) put v =
. Then 1 v =
(1 + t)2
1+t
1 v (1 t)2
(2)
(1 v) 2 1 F1
163
(1 t)
(3)
(1 + t)1 F1
1
4xt
; 1;
2
(1 t)2
=
X
( 1 )k Lk (x)22k tk
2
k=0
k!(1 + t)2k
or
1
(1 t)
1 F1
1
4xt
; 1;
2
(1 t)2
=
X
( 1 )k 22k Lk (x)tk
2
k!(1 + t)1+2k
k=0
Therefore we have
Zn (x)tn
n=0
=
=
X
(1)n 22k ( 21 )k (1 + 2k)n Lk (x)tn+k
k!n!
n.k=0
n
X
(1) (n + 2k)!Lk (x)tn+k
n,k=0
n
X
X
n=0 k=0
(k!)2 n!
(1)n+k (n + k)!Lk (x) n
t .
(k!)2 (n k)!
n.n + 1;
1+x
.
2
Pn (x) = (1)n 2 F1
1;
Hence
Pn (2y 1)
n.n + 1;
= (1)n 2 F1
=
1;
n
X
(1)n+k (n + k)!y k
k=0
(k!)2 (n k)!
Zn (x) =
n
X
(1)n+k (n + k)!Lk (x)
k=0
(k!)2 (n k)!
= (1)n
n
X
(n)k (n + 1)k Lk (x)
k=0
k!k!
1
n, n + 1; 1, ; x .
2
Then
fn (x)tn =
n=0
X
n
X
(1)s (n + s)!xs tn
n=0 s=0
s!s!( 21 )s (n s)!
X (1)s Lk (x)
xs
=
.
2
(s!)
k!(s k)!
k=0
Hence,
fn (x)tn
n=0
=
=
s
n X
X
X
(1)k+s (n + s)!Lk (x)tn
n=0 s=0 k=0
X
s
X
n,k,s=0
X
n
X
s!k!n!( 12 )s+k
n, 1 + n + 2k;
n+k
X
1 (n + 2k)!Lk (x)t
.
=
F
1
k!( 12 )k n!
n,k=0
+ k;
2
By Example 5, page 119, we get
=
2 F1
n, 1 + n + 2k;
1
+ k;
2
n
1 = (1) (1 + 1 + 2k
( 12 + k)n
1
2
k)n
(1)n ( 23 + k)n
(1)n ( 23 )n+k ( 12 )k
=
.
( 12 + k)n
( 32 )k ( 12 )n+k
Then
fn (x)tn
n=0
X
(1)n ( 23 )n+k ( 12 )k (n + 2k)!Lk (x)tn+k
( 23 )k ( 12 )n+k k!( 12 )k n!
n,k=0
X
n
X
(1)nk ( 32 )n (n + k)!Lk (x)tn
=
.
( 23 )k ( 12 )n k!(n k)!
n=0 k=0
Therefore, since
(1)
( 32 )n
(n + 21 )
=
= 2n + 1,
( 12 )n
( 12 )
n
X
(n)k (n + 1)k Lk (x)
fn (x) = (1)n (2n + 1)
.
k!( 23 )k
k=0
165
Equation (1) will also be of use to us in later work. We now note that
Z
Z
n
X
(n)k (n + 1)k
k=0
k!( 32 )k
ex Lk (x)dx.
ex L0 (x)dx = 1.
Hence
n, n + 1, m, m + 1;
Z
0
n
(1) (2n + 1)(n)k (n + 1)k
x
e Lk (x)fn (x)dx =
k!( 32 )k
Z
0
3 3
1, , ;
2 2
1 ,
; 0 k n,
; k > n,
n, n + 1, k, k + 1;
3
1, 1, ;
2
1 .
1
Solution 8. Again we use fn (x) = 2 F2 n, n + 1; 1, ; x as in Exercise 7.
2
Recall that
fn (x) = (1)n (2n + 1)
(1)
n
X
(n)k (n + 1)k Lk (x)
k=0
k!( 32 )k
Zn (x) = (1)n
n
X
(n)k (n + 1)k Lk (x)
k=0
k!k!
At once
Z
0
Z
n X
m
X
(n)k (n + 1)k (m)s (m + 1)s
k=0 s=0
k!( 32 )k s!( 32 )s
Now, by the orthogonality property of Laguerre polynomials, and the fact that
Z
ex Ln2 (x)dx = 1, we get
0
ex Lk (x)Ls (x)dx.
min(n,m)
n, n + 1, m, m + 1;
1 .
= (1)n+m (2n + 1)(2m + 1)4 F3
3 3
1, , ;
2 2
ex fn (x)fm (x)dx
= (1)n+m (m + 1)(2m + 1)
Next consider
Z
ex Lk (x)fn (x)dx.
At once
ex Lk (x)fn (x)dx
= (1)n (2n + 1)
Z
n
X
(n)s (n + 1)s
s!( 32 )s
s=0
n
(1) (2n + 1)(n)k (n + 1)k
=
.
k!( 32 )k
ex Lk (x)Ls (x)dx
ex fN (x)Zk (x)dx
= (1)n+k (2n + 1)
Z
k
n X
X
(n)s (n + 1)s (k)i (k + 1)i
s!( 23 )s i!i!
s=0 i=0
min(n,k)
n, n + 1, k, k + 1;
1 .
= (1)n+k (2n + 1)4 F3
3
1, 1, ;
2
= (1)n+k (2n + 1)
Z
0
n, n + 1, k, k + 1;
1 .
1, 1, 1, ;
ex Ls (x)Li (x)dx
Zn (x)Zk (x)dx
= (1)
n+k
167
Z
n X
k
X
(n)s (n + 1)s (k)i (k + 1)i
s=0 i=0
min(n,k)
s!s!i!i!
ex Ls (x)Li (x)dx
n, n + 1, k, k + 1;
1 .
= (1)n+k 4 F3
1, 1, 1;
= (1)n+k
X
n (x; )tn
= et 1 F1 (1 + x; 1; t(1 e )),
n!
n=0
n (x, + ) =
n
(e 1)n X n!(1 e+ )k k (x; )
,
en (1 e )n
k!(n k)!(e 1)k
k=0
(s n)(s + x + 1)
,
(s + 1)(s + x n + 1)
x 2 (t x) 2 fn (; ; x)dx = Zn (t),
0
t
x 2 (t x) 2 fn (; ; x(t x))dx = Zn
0
1 2
t ,
4
and
Z
0
1
x (t x) fn ; ; x(t x) dx = Ln (t)Ln (t).
2
1
2
1
2
Solution 11. We shall use Theorem 37, page (??), to evaluate the integrals in
question.
1
Recall that fn (; ; x) = 2 F2 n, n + 1; 1, ; x .
2
1
1
1
Then use = , = , p = 2, q = 2, a1 = n, a2 = n + 1, b1 = 1, b2 = , c =
2
2
2
1, k = 1, s = 0 in Theorem 37 to get
1
n, n + 1, ;
2
1 1
=B
,
F
2 2
1
1, , 1;
2
n,
n
+
1;
1
1
( 2 )( 2 )
=
F
2 2
(1)
1, 1;
= Zn (t).
x 2 (t x) 2 fn (; ; x)dx
2
2
= F
1 1 2
1, , , ;
2 2 2
n, n + 1;
t2
= 2 F2
4
1, 1;
1 2
= Zn
t .
4
x 2 (t x) 2 fn (; ; x(t x))dx
t2
4
Finally, consider
1
fn ; ; z
2
n, n + 1;
= 2 F3
1
In Theorem 37 we now use = , =
2
1
1, b2 = 1, b 3 = , c = 1, k = 1, s = 1, to
2
Z t
1
1
1
x 2 (t x) 2 fn ; ; x(t x) dx
2
0
1 1
1, , ;
2 2
z .
1
, p = 2, q = 2, a1 = n, a2 = n + 1, b1 =
2
get
( 21 )( 12 )
F
(1)
n, n + 1;
= Ln (t)Ln (t),
1
1, 1, ;
2
t2
4
(c)n
2 F2 (n, c + n; 1, 1; x).
n!
Show that
n (x) =
n
X
(1)nk (c 1)nk (c)n+k (2k + 1)Zk (x)
(n k)!(n + k + 1)!
k=0
(c)n
2 F2 (n, c + n; 1, 1; x).
n!
Then
n (x) =
n
X
(1)k (c)n+k xk
k=0
k!(n k)!(k!)2
so
n (x)t
n=0
X
n
X
(1)s (c)n+s xs tn
n=0 s=0
(s!)3 (n s)!
s
X
(s)k (2k + 1)Zk (x)
k=0
(s + k + 1)!
Then
n (x)t
n=0
X
s
X
(1)s+k (c)n+2s (2k + 1)Zk (x)tn+s
=
n!(s k)!(s + k + 1)!
n,s=0 k=0
We now have
169
n (x)tn
n=0
=
=
=
X
n,k=0
X
n,k=0
2 F1
n, c + n + 2k;
n+k
2 + 2k;
(2 + 2k)(2 c)
(c)n+2k (2k + 1)Zk 9x)tn+k
(2 + 2k + n)(2 c n)
n!(2k + 1)!
n
nk
X X (1)
(c)n+k (c 1)nk (2k + 1)Zk (x)
(n k)!(n + k + 1)!
n=0 k=0
tn .
Hence
n (x) =
n
X
(1)nk (c 1)nk (c)n+k (2k + 1)Zk (x)
k=0
(n k)!(n + k + 1)!
Problem 13. Show that the F (t) of equation (10), page 286, can be put in the
form
v;
4zt
.
(1 z)2
F (t) = tb et (1 z)2v 1 F1
b + 1;
Solution 13. From equation (10), page 500, we get
n, 2v + n;
n
X
(2v)n z
t
F (t) = tb et
2 F2
1
n!
n=0
v + , b + 1;
2
n
X
X
(1)k n!(2v)n (2v + n)k tk z n
b t
=t e
k!(n k)!n!(v + 12 )k (b + 1)k
n=0 k=0
X
(2v + 2k)n z n (1)k (2v)2k tk z k
= tb et
n!
k!(v + 12 )k (b + 1)k
k=0 n=0
X
(1)k tk z k 22k (v)k (v + 21 )k
= tb et
k!(v + 12 )k (b + 1)k (1 z)2v+2k
k=0
X
(1)k 22k (v)k tk z k
= tb (1 z)2v et
k!(b + 1)k (1 z)2k
k=0
4tz
b
2v t
= t (1 z) e 1 F1 v; b + 1;
.
(1 z)2
21. Chapter 19 Solutions
171
1
4
10 = 2p 4 G
(1 q 2n )2
n=1
and that
G=
(1 q 2n ).
n=1
Hence
1
10 = 2q 4 G3 .
Problem 2. The following formulas are drawn from Section 168. Derive (9) and
(10):
42 12 (z) + 32 22 (z) = 22 32 (z),
32 12 (z) + 42 22 (z) = 22 42 (z),
22 12 (z) + 42 32 (z) = 32 42 (z),
22 22 (z) + 42 42 (z) = 32 32 (z).
(8)
(9)
(10)
(11)
Solution 2. We are given
Problem 3. Use (8) and (10) of Exercise 2 and the equation 24 + 44 = 34 to show
that
14 (z) + 34 (z) = 24 (z) + 44 (z).
173
= 24 34 42 (x)32 (y) + 44 32 (x)42 (y) 32 44 32 (x)32 (y) + 42 (x)42 (y)
22 34 42 (y)32 (x) + 44 32 (y)42 (x) 22 42 32 (y)32 (x) + 42 (y)42 (x)
= 24 (34 44 ){42 (x)32 (y) 32 (x)42 (y)} .
(A)
We shall now transform (A) by means of the basic table. Note that (A) is the
first of the required results
5.
inExercise
In (A) replace x by x +
to get
2
22 2 (x + y)2 (x y) = 32 (x)32 (y) 42 (x)42 (y).
In (B) replace x by x +
to obtain
2
(B)
(D)
Problem 6. Use equation (10) of Exercise 2 and the first relation of Exercise 5 to
obtain the first relation below; then use the basic table to get the remaining three
results.
32 1 (x + y)1 (x y) = 12 (x)32 (y) 32 (x)12 (y),
32 2 (x + y)2 (x y)22 (x)32 (y) 42 (x)12 (y),
32 3 (x + y)3 (x y) = 32 (x)32 (y) + 12 (x)12 (y),
32 4 (x + y)4 (x y) = 42 (x)32 (y) + 22 (x)12 (y).
Solution 6. From Exercise 5 we use
22 1 (x + y)1 (x y) = 42 (x)32 (y) 32 (x)42 (y).
From Exercise 2 we use equation (10) in the form
(10)
Thus we obtain
22 1 (x+y)1 (xy) = 32 [22 12 (x)32 (y)+42 32 (x)32 (y)22 32 (x)12 (y)42 32 (x)32 (x)].
Two terms drop out and we are left with
(A)
32 1 (x + y)1 (x y) = 12 (x)32 (y) 32 (x)12 (y),
which was desired.
In (A) replace x by x +
to get
2
32 2 (x + y)2 (x y) = 22 (x)32 (y) 42 (x)12 (y).
In (B) replace x by x +
to get
2
(B)
In (C) replace x by x +
to get
2
(D)
32 4 (x + y)4 (x y) = 42 (x)32 (y) + 22 (x)12 (y).
Equations (A), (B), (C), (D) are the required ones.
Problem 7. Use the first relation of Exercise 6 and the identities from Exercise 2
to obtain the frist relation below. Derive the other three relations from the first one.
32 1 (x + y)1 (x y) = 42 (x)22 (y) 22 (x)42 (y),
32 2 (x + y)2 (x y) = 32 (x)22 (y) 12 (x)42 (y),
32 3 (x + y)3 (x y) = 22 (x)22 (y) + 42 (x)42 (y),
175
= 34 [24 42 (x)22 (y) 44 22 (x)42 (y) + 22 42 {42 (x)42 (y) 22 (x)22 (y)}]
34 [24 42 (y)22 (x) 44 22 (y)42 (x) + 22 42 {42 (y)42 (x) 22 (y)22 (x)}].
We thus obtain
32 1 (x + y)1 (x y) = 34 [(24 + 44 ){42 (x)22 (y) 22 (x)42 (x)42 (y)}].
Since 24 + 44 = 34 , we arrive at the desired result
32 1 (x + y)1 (x y) = 42 (x)22 (y) 22 (x)42 (y).
to obtain
In (A) replace x by x +
2
(A)
(C)
Problem 8. Use equation (10) of Exercise 2 and the first relation of Exercise 6
to obtain the first relation below. Derive the others with the aid of the basic table,
page 319.
42 1 (x + y)1 (x y) = 12 (x)42 (y) 42 (x)12 (y),
42 2 (x + y)2 (x y) = 22 (x)42 (y) 32 (x)12 (y),
42 3 (x + y)3 (x y) = 32 (x)42 (y) 22 (x)12 (y),
42 4 (x + y)4 (x y) = 42 (x)42 (y) 12 (x)12 (y).
(D)
Problem 9. The first relation below was derived in Section 169. Obtain the other
three.
42 1 (x + y)1 (x y) = 32 (x)22 (y) 22 (x)32 (y),
42 2 (x + y)2 (x y) = 42 (x)22 (y) 12 (x)32 (y),
42 3 (x + y)3 (x y) = 42 (x)32 (y) 12 (x)22 (y),
42 4 (x + y)4 (x y) = 32 (x)32 (y) 22 (x)22 (y).
Solution 9. From Section 169, equation (7), page 371, we obtain
42 1 (x + y)1 (x y) = 32 (x)22 (y) 22 (x)32 (y).
In (A) replace x by x +
to get
2
(A)
(D)
177
Problem 10. Use the method of Section 169 together with the basic table, page
319, to derive the following results, one of which was obtain in Section 169.
3 4 1 (x + y)2 (x y) = 1 (x)2 (x)3 (y)4 (y) + 1 (y)2 (y)3 (x)4 (x),
2 4 1 (x + y)3 (x y) = 1 (x)2 (y)3 (x)4 (y) + 1 (y)2 (x)3 (y)4 (x),
2 3 3 (x + y)4 (x y) = 1 (x)2 (y)3 (y)4 (x) + 1 (y)2 (x)3 (x)4 (y),
2 3 2 (x + y)3 (x y) = 2 (x)3 (x)2 (y)3 (y) 1 (x)4 (x)1 (y)4 (y),
2 4 2 (x + y)4 (x y) = 2 (x)4 (x)2 (y)4 (x) 1 (x)3 (x)1 (y)3 (y),
3 4 3 (x + y)4 (x y) = 3 (x)4 (x)3 (y)4 (y) 1 (x)2 (x)1 (y)2 (y).
Solution 10. For brevity, since we know how to discover such formulas, let us
consider the function
1 (x)2 (x)3 (y)4 (y) + 1 (y)2 (y)3 (x)4 (y)
.
1 (x + y)2 (x y)
we shall show that 1 (x) is an elliptic function of order less than two, and then
obtain its constant value.
Let N1 (x) = the numerator of 1 (t) and D1 (x) be the denominator of 1 (x).
Then at once the basic table yields
1 (x) =
Hence the only poles of 1 (x) in a cell are simple ones at x = y = 0 and x y = .
2
But x = y is also a zero of the numerator because
N1 (y) = 1 (y)2 (x)3 (y)4 (y) + 1 (y)2 (y)3 (y)4 (y) = 0.
Hence 1 (x) is an elliptic function order less than two and is constant. Now
0 + 1 (y)2 (y)3 4
= 3 4 ,
1 (y)2 (y)
since 2 (y) is an even function. Thus we obtain the first of the derived relations,
1 (0) =
(A)
(B)
179
which is the
equation, slightly rewritten, of the desired results. In (B)
fourth
replace x by x +
to find that
2
2 4 2 (x + y)4 (x y) = 2 (x)2 (y)4 (x)4 (y) 1 (y)1 (x)3 (y)3 (x),
(E)
which
is a rewritten form of a desired result. Finally, turn to (A) and replace x
by x +
to get
2
1
3 4 iq 2 ei(x+y) ei(xy) 4 (x+y)3 (xy) = iq 2 e2ix 4 (x)3 (x)3 (y)4 (y)+iq 2 e2ix 1 (y)2 (y)2 (x)1 (x),
or
3 4 4 (x + y)3 (x y) = 3 (x)4 (x)3 (y)4 (y) + 1 (x)2 (x)1 (y)2 (y).
Now change y to (y) to arrive at the desired result
(F )
Problem 11. Use the results in Exercises 4-10 above to show that
23 2 (2x) = 24 (x) 14 (x) = 34 (x) 44 (x),
33 3 (2x) = 14 (x) + 34 (x) = 24 (x) + 44 (x),
43 4 (2x) = 44 (x) 14 (x) = 34 (x) 24 (x),
32 2 2 (2x) = 22 (x)32 (x) 12 (x)42 (x),
42 2 2 (2x) = 22 (x)42 (x) 12 (x)32 (x),
22 3 3 (2x) = 22 (x)32 (x) + 12 (x)42 (x),
42 3 3 (2x) = 32 (x)42 (x) 12 (x)22 (x),
22 4 4 (2x) = 12 (x)32 (x) + 22 (x)42 (x).
32 4 4 (2x) = 12 (x)22 (x) + 32 (x)42 (x).
Solution 11. We shall use y = x in various equations of Exercises 4-10. From
Exercise 4, second equation, we get
22 2 (2x) = 24 (x) 14 (x).
From Exercise 4 third equation, we get
22 3 3 (2x) = 22 (x)32 (x) + 12 (x)42 (x).
From Exercise 4, fourth equation, we get
22 4 4 (2x) = 22 (x)42 (x) + 12 (x)32 (x).
181
3 (z| ) = (i ) 2 exp
z2
i
3
z t
.
From the above identity, obtain corresponding ones for the other theta functions
with the aid of the basic table, page 319.
Solution 12. Consider the function
(z) =
exp
3 (z| )
.
3 ( z | 1
)
z2
|
First
the
theta functions are analytic for all finite z since Im( ) > 0 implies
1
Im
> 0.
To get the zeros of the denominator, recall that 3 (y|t) has its zeros all simple
ones located at
t
+
+ n + mt
2
2
z 1
for integral n and m. Then 3
m
=
+ n
2
2
or at
y=
z=
+ n m,
2
2
which is equivalent to
z=
+
+ n1 + m1 ,
2
2
z 1
for integral n1 and m1 . Therefore the zeros of 3 (z| ) and those of 3
coincide. Hence the function (z0 has no singular points in the finite plane.
To show that (z) has the desired two periods, consider (z + ) and (z + ).
(z + ) =
exp
3 (z + | )
.
1
3 z+
|
(z+)2
i
(z + )
3 (z| )
exp
exp
exp i
3 z | 1
3 (z| )
=
z2
i
i
exp i
exp 2iz
exp
exp exp
3 (z| )
=
z2
exp i 3 z | 1
= (z).
=
z2
i
2z
i
2iz
3 z |
(z + ) =
3 (z + | )
=
(z+ )2
z+
1
exp
exp
3
i
z2
i
ei e2iz 3 (z| )
.
3 z + | 1
exp
i
We thus obtain
|psi(z + ) =
exp
3 (z| )
= (z).
3 z | 1
z2
i
3 (0| )
3 (0| )
=
.
e0 3 (0| 1 )
3 (0| 1 )
3 (z| ) = c exp
z2
i
z 1
3
,
In (A) replace z by z +
to get
2
2
(z +
z
1
2 )
3
+
.
3 z +
= c exp
2
i
We conclude that
(B)
2 (z| ) = c exp
z2
i
4
z 1
.
Return to (A) and replace z by z +
to get
2
(z + 2 )2
1
z
3
+
3 z + = c exp
2
i
or
183
z
z
z2
1 1
exp
3 +
exp
i
4i
2
i
2
i
z
z
iz
z 1
exp
exp
= c exp
exp
exp
2
4
i
i
4i
2
z
z 1
= c exp
2
.
i
4 (z| )
= c exp
Hence we arrive at
(C)
z2
i
z 1
2
.
4 (z| ) = c exp
Now in (C) replace z by z +
. We thus obtain
2
2
(z +
1
z
2 )
4 z +
2
+
,
= c exp
2
i
or
i
i exp
4
exp(iz)1 (z| ) = c exp
z2
i
exp
z
c
exp
4i
z 1
.
(1)1
We wish to put z = 0 in (E) and use the known result that 10 = 2 3 4 . From
(E) we thus get
10 (z| )
ic
2
0|
3
0|
4
or
2 (0| )
3 (0| )
4 (0| )
ic
1
1
1 = .
4 (0| ) 3 (0| ) 2 (0| )
With the aid of (A), (B), (C), we may conclude that
ccc =
ic
,
or
c2 =
1
Hence c = (i ) 2 or c = (i ) 2 .
Now 3 (z) was defined by
1
.
i
0|
,
3 (z| ) = 1 + 2
ein
cos(2nz).
n=1
en
(i )
>0
n=0
2
1
z
z 1
4
,
2 (z| ) = (i ) 2 exp
i
2
1
z
z 1
4 (z| ) = i(i ) 2 exp
2
,
i
2
1
z
z 1
1 (z| ) = i(i ) 2 exp
1
.
i
With regard to the usefulness of these formulas, note that q = exp(i ) so that
|q| = exp[Im( )]. For Im( ) > 0, |q| < 1 and we have convergence of the series
definitions of the theta functions but that convergence is slow if Im( ) is small so
that |q| is near unity.
Since
1
Im( )
Im
=
,
| |2
the parameter ( 1 ) will be useful in computations with Im(t) small and positive.
23. Chapter 21 Solutions
Problem 1. Derive the preceding addition theorem (7) by the method of Section
179. You may use results from the exercises at the end of Chapter 20.
Solution 1. We seek an addition theorem for cn(u). Now
cn(u) =
4 2 (u32 )
.
2 4 (u32 )
185
2 (x + y)
4 (x + y)
2
2
4 cn(u) 4 cn(v)
or
cn(u + v) =
2
3
2
3
3 sn(u) 4 dn(u) 3 sn(v) 4 dn(v)
2
2
2
cn(u)cn(v) sn(u)dn(u)sn(v)dn(v)
,
1 k 2 sn2 (u)sn2 (v)
as desired.
Problem 2. Derive preceding (8) by the method of Section 179, with the aid of
results from the exercises at the end of Chapter 20.
Solution 2. We know that
dn(u) =
4 3 (u32 )
2
,k = 2.
3 4 (u32 )
3
3
3
4 dn(u) 4 dn(v)
2
2
2
2
3 sn(u)) 4 cn(u) 3 sn(v) 4 sn(v)
.
2
2
2
dn(u)dn(v) k 2 sn(u)cn(u)sn(v)cn(v)
.
1 k 2 sn2 (u)sn2 (v)
1
cn3 (x)dn(x)dx = sn(x) sn3 (x) + c.
3
cn3 (x)dn(x)dx.
d
dn(x) = cn(x)dn(x)
dx
and that
cn2 (x) = 1 sn2 (x).
Hence
Z
cn3 (x)dn(x)dx
[1 sn2 (x)]cn(x)dn(x)dx
1
= sn(x) sn2 (x) + c.
3
Problem 4. Show that if g(x) and h(x) are any two different ones of the three
functions sn(x), cn(x), dn(x), and if m is a non-negative integer, you can perform
the integration
Z
g 2m+1 (x)h(x)dx.
Solution 4. Consider
Z
A=
g 2m+1 (x)h(x)dx.
We know that
d
sn(x) = cn(x)dn(x);
dx
d
cn(x) = sn(x)dn(x);
dx
d
dn(x) = k 2 sn(x)cn(x).
dx
Now let g(x) and h(x) be any two different ones of sn(x), cn(x), dn(x). Let (x) be
the other of the three functions. Then
g(x)h(x)dx = c1 d[(x)],
in which c1 = 1, 1 or K 2 according to whether (x) is sn(x), cn(x), or dn(x).
Now we also know that
sn2 (x) + cn2 (x) = 1,
k 2 sn2 (x) + dn2 (x) = 1,
dn2 (x) k 2 cn2 (x) = x2 .
Therefore the square of any one of sn(x), cn(x), dn(x) is a linear function of the
square of any other one of them. Hence we may write
g 2 (x) = a1 + b1 2 (x).
We know have
Z
g 2m+1 (x)h(x)dx
[g 2 (x)]m g(x)h(x)dx
Z
= c1 [a1 + b1 2 (x)]m d(x),
Z
Solution 5. Consider
sn(x)dx.
Z
Z
sn(x)dx =
Z
=
Z
=
187
We may write
sn(x)[dn(x) kcn(x)]
dx
dn(x) kcn(x)
sn(x)dn(x) ksn(x)cn(x)
dx
dn(x) kcn(x)
1
d[dn(x)] + k d[dn(x)]
Z dn(x) kcn(x)
1
d[dn(x)] k[cn(x)]
=
k
dn(x) kcn(x)
1
= log[dn(x) kcn(x)] + c.
k
dn(2x) =
and
dn(2x) =
cn(2x) =
Then
dn(2x) kcn(2x)
1 k 2 sn2 (x) k 2 sn2 (x) + k 2 sn4 (x) k + ksn2 (x) + ksn2 (k) k 3 sn4 (x)
1 k 2 sn4 (x)
1 + ksn2 (x) + ksn2 (x)[1 + ksn2 (x)] k[1 + ksn2 (x)] k 2 sn2 (x)[1 + ksn2 (x)]
=
[1 ksn2 (x)][1 + ksn2 (x)]
1 + ksn2 (x) k k 2 sn2 (x)
=
1 ksn2 (x)
1 k + k(1 k)sn2 (x)
=
.
1 ksn2 (x)
=
Hence
dn(2x) kcn(2x) =