Professional Documents
Culture Documents
Bessel Function PDF
Bessel Function PDF
Project for the Penn State - Göttingen Summer School on Number Theory
Martin Kreh
Content i
Content
References 21
In the first chapter, we introduce the (modified) Bessel differential equation and deduce
from it the (modified) Bessel functions of first and second kind. This will be done via a
power series approach. We decided to start from a differential equation, since this seems
kind of naturally. We will point out later that we could have defined the Bessel function
in an other, equivalent, way.
x2 y 00 + xy 0 + (x2 − n2 )y = 0
for n ∈ N, which can for example be obtained by the separation of the wave equation in
cylindric or polar coordinates. Since this is a second order differential equation, we will
have two linearly independent solutions. We will now start to construct a first solution.
We take as approach for a solution y a power series. It will be convenient to write
∞
X
y(x) = xn bk xk .
k=0
Then we have
∞
X ∞
X
0 n k n
xy (x) = nx bk x + x kbk xk
k=0 k=0
and
∞
X ∞
X ∞
X
x2 y 00 (x) = n(n − 1)xn bk xk + 2nxn kbk xk + xn k(k − 1)xk .
k=0 k=0 k=0
We further have
∞
X
2 n
x y(x) = x bk−2 xk
k=2
∞
X
= xn (n(n − 1)bk + 2nkbk + k(k − 1)bk + nbk + kbk + bk−2 − n2 bk )xk
k=0
X∞
= xn (2nkbk + k 2 bk + bk−2 )xk .
k=0
b−1
Since b−1 = 0 we get b1 = − 1+2n = 0 and recursively b2k−1 = 0 for all k as long as
2n is not a negative whole number. But in this case the recursion will also be fullfilled if
b2k−1 = 0 for all k. So it remains to consider even k. Here we have no condition on b0 .
We can choose it freely and will do so later. Now if −n ∈
/ N we get
b2k−2
b2k = −
4k(n + k)
b0
b2k = (−1)k .
4k k!(n + k)(n + k − 1) · · · (n + 1)
1
It is convenient to choose b0 = 2n n! . So together we get
∞
x n X (−1)k x 2k
y(x) = .
2 k!(n + k)! 2
k=0
This is convergent due to the quotient criterion, therefore we have indeed constructed a
solution for the Bessel equation.
We also want to construct a solution for complex order ν. Here we need to exchane
the factorial with the Gamma function. Then we get
∞
x ν X (−1)k x 2k
Jν (x) =
2 Γ(k + 1)Γ(ν + k + 1) 2
k=0
= (−1)n Jn (x)
This means that Jν and J−ν are two linearly independent solutions if ν ∈
/ Z. If ν ∈ Z,
they are linearly dependent. Since (−1)n = cos nπ, the function Jν (z) cos νπ − J−ν (z) is
a solution of the Bessel equation, which vanishes if n ∈ N0 . Therefore, we define
for ν ∈
/ Z. Then for n ∈ Z the limit
exists, as can be shown by L’Hospitals rule, and Jν and Yν are two linearly independent
solutions of the Bessel equation for all ν ∈ C. This can e.g. be shown be computing the
Wronskian determinant.
x2 y 00 + xy 0 − (x2 + ν 2 )y = 0.
Analog to the previous section we can compute a solution of this differential equation
using the power series approach. This gives
∞
x ν X 1 x 2k
Iν (x) =
2 Γ(k + 1)Γ(ν + 1 + k) 2
k=0
As before, we search for a second, linearly independent solution. This will, for ν ∈
/ Z, be
given by
π I−ν (x) − Iν (x)
Kν (x) =
2 sin(νπ)
and its limit for ν → n ∈ Z exists.
In this chapter we will prove some properties of Bessel functions. There are of course
more interesting facts, in particular, there are connections between Bessel functions and,
e.g. Legendre polynomials, hypergeometric functions, the usual trigonometric functions
and much more. These can be found in [Wat22] and [GR00]
Many facts about Bessel functions can be proved by using its generating function. Here
we want to determine the generating function.
Theorem 2.1 We have
∞
x −1 )
X
e 2 (z−z = Jn (x)z n (2.1)
n=−∞
x −1 )
i.e. e 2 (z−z is the generating function of Jn (x).
Proof: We have
∞ ∞
x x 1 X ( x2 )m m X (−1)k ( x2 )k −k
e 2 z e− 2 z = z z
m! k!
m=0 k=0
∞ k x m+k
X X (−1) ( 2 ) n
= z
m!k!
n=−∞ m−k=n
m,k≥0
∞ ∞
!
X X (−1)k x 2k x n
= zn
n=−∞
(n + k)!k! 2 2
k=0
X∞
= Jn (x)z n .
n=−∞
q.e.d.
∞
X
1 = J0 (x) + 2 J2n (x).
n=1
∞
X ∞
X
ix sin φ inφ
cos(x sin φ)+i sin(x sin φ) = e = Jn (x)e = Jn (x)(cos(nφ)+i sin(nφ))
n=−∞ n=−∞
so
∞
X
cos(x sin φ) = J0 (x) + 2 J2n (x) cos(2nφ)
n=1
and
∞
X
sin(x sin φ) = 2 J2n+1 (x) sin((2n + 1)φ).
n=0
π
This gives the results with φ = 2 and φ = 0, respectively. q.e.d.
for all n ∈ Z.
Proof: To show the first equality, we make the change of variables x → −x, z → z −1 in
the generating function. Then the function does not change and we get
∞
X ∞
X ∞
X
Jn (−x)z −n = Jn (x)z n = J−n (x)z −n
n=−∞ n=−∞ n=−∞
and comparing coefficients gives the result. The second equality follows analog with the
change of variable z → −z −1 . q.e.d.
Proof: The first statement follows, when we differentiate equation (2.1) with respect to
x:
∞ ∞ ∞
X 1 1 x 1 1 X 1 X
Jn (x)z n = z− e 2 (z− z ) = z Jn (x)z n − Jn (x)z n
n=−∞
2 z 2 n=−∞ 2z n=−∞
∞
X 1
= (Jn−1 (x) − Jn+1 (x))z n .
n=−∞
2
With the last two lemmata we can describe some integrals with Bessel functions via a
different Bessel function.
Lemma 2.5 For any n ∈ Z we have
Z Z
x n+1
Jn (x) dx = x n+1
Jn+1 (x) and x−n+1 Jn (x) dx = −x−n+1 Jn+1 (x).
Proof: The first equation follows directly from (2.2). For the second one, we use (2.2)
and lemma 2.3 to get
Z Z
x−n+1 Jn (x) dx = x−n+1 (−1)n J−n (x) dx = (−1)n x−n+1 J−n+1 (x) = −x−n+1 Jn+1 .
q.e.d.
∞ ∞
!
X X X
= Jn2 (x) + Jn+m (x)Jn (x) z n .
n=−∞ m∈Z n=−∞
m6=0
and
∞
X
0= Jn+m (x)Jn (x)
n=−∞
for all n ∈ Z.
Most of the results hold (in a similar form) also for the other three Bessel functions.
Since the proofs are anlog, we will only state the results.
Theorem 2.9 (Results for Jn (x), Yn (x) and Kn (x))
x −1 )
• The generating function for the In (x) is e 2 (z+z , i.e.
∞
x −1 )
X
e 2 (z+z = In (x)z n .
n=−∞
2n
Yn−1 (x) + Yn+1 (x) = Yn (x), Yn−1 (x) − Yn+1 (x) = 2Yn0 (x),
x
2n
In−1 (x) − In+1 (x) = In (x), In−1 (x) + In+1 (x) = 2In0 (x),
x
2n
Kn+1 (x) − Kn−1 (x) = Kn (x), Kn−1 (x) + Kn+1 (x) = −2Kn0 (x).
x
Indeed this and the next two sets of formulae, as well as those for Jn (x), hold also
for complex orders, as can be shown through the series representation.
d n
(x Yn (x)) = xn Yn−1 (x),
dx
d n
(x In (x)) = xn In−1 (x),
dx
d n
(x Kn (x)) = −xn Kn−1 (x).
dx
• We have
∞
X
Ik (x) = ex .
n=−∞
• It holds
X
In (x + y) = Ik (x)In−k (x).
k∈Z
In this section we want to examine how the Bessel functions look like when we plug in
some special values.
1
Lemma 2.10 If ν ∈ 2 + Z, the Bessel functions are elementary functions.
1. We have
r
2
J 1 (x) = Y− 1 (x) = sin(x),
2 2 πx
r
2
J− 1 (x) = −Y 1 (x) = cos(x),
2 2 πx
√
I 1 (x) = 2πx sinh(x),
2
r
2
I− 1 (x) = cosh(x),
2 πx
r
π −x
K 1 (x) = K− 1 (x) = e .
2 2 2x
Proof: The first formulae follow directly from the series representations of Jν and Iν ,
the definitions of Yν and Kν and the properties of the Γ-function. For example, we have
∞
(−1)k
r
xX x 2k
J 1 (x) =
2 2
k=0
Γ(k + 1)Γ(k + 32 ) 2
r ∞
x X (−1)k x 2k
= √
(2k+1)! π
2 2
k=0 k! k!22k+1
r ∞
2 X (−1)k 2k+1
= x
πx (2k + 1)!
k=0
r
2
= sin x
πx
and
cos(− π2 )J− 1 (x) − J 1 (x)
2 2
Y− 1 (x) = = J 1 (x).
2 sin(− π2 ) 2
The rest follows from the recurrence formulas of the last section. q.e.d.
π n−1
In (x) = i−n Jn (ix) and Kn (x) = i (Jn (ix) + iYn (ix)).
2
In this section we will give an integral representation for each of the four Bessel functions
that we have introduced. There are much more representations as can be shown here, see
e.g. [Wat22] and [GR00].
Zπ Z∞
1 sin(πν)
Jν (x) = cos(x sin t − νt) dt − e−x sinh(t)−νt dt,
π π
0 0
Zπ Z∞
1 1
Yν (x) = sin(x sin t − νt) dt − e−x sinh(t) (eνt + cos(πν)e−νt ) dt,
π π
0 0
Zπ Z∞
1 sin(πν)
Iν (x) = ex cos t
cos(νt) dt − e−x cosh(t)−νt dt,
π π
0 0
Z∞
Kν (x) = e−x cosh(t) cosh(νt) dt.
0
1 1
t−z et dt where γ 0 is some contour from
R
Proof: We use the representation Γ(z) = 2πi γ0
−∞, turning around 0 in positive direction and going back to −∞. Then we get
∞
(x/2)ν (−1)k (x/2)2k t−ν−k−1 t (x/2)ν
Z X Z
2
Jν (x) = e dt = t−ν−1 et−x /(4t) dt.
2πi k! 2πi
γ 0 k=0 γ0
for some contour γ̃ of the same kind. Now let u = ew . Then the contour γ, given as
rectangular with vertices ∞ − iπ, −iπ, iπ, ∞ + iπ without the right edge is the image of
a suitable γ̃ (see figure 2.1). So we get
Z
1 1
Jν (x) = e−νw ex sinh(w) dw = (I1 + I2 − I3 )
2πi 2πi
γ
with
Zπ
I1 = e−iνt ex sinh(it) i dt,
−π
Z∞
I2 = e−ν(iπ+t) ex sinh(iπ+t) dt,
0
Z∞
I3 = e−ν(−iπ+t) ex sinh(−iπ+t) dt.
0
We have
Zπ
1 1
I1 = e−iνt ex sinh(it) i dt
2πi 2πi
−π
π
Z0
Z
1
= ei(x sin t−νt) dt + ei(x sin t−νt) dt
2π
0 −π
Zπ
1
= ei(x sin t−νt) + e−i(x sin t−νt) dt
2π
0
Zπ
1
= cos(x sin t − νt) dt
π
0
Z∞ Z∞
1 1
(I2 − I3 ) = e−ν(iπ+t) ex sinh(iπ+t) dt − e−ν(−iπ+t) ex sinh(−iπ+t) dt
2πi 2πi
0 0
Z∞
1
= e−x sinh t−νt (e−νiπ − eνiπ ) dt
2πi
0
Z∞
sin(πν)
=− e−x sinh t−νt dt
π
0
which gives the second part. This proves the first result.
I1 sin(νπ)
sin(νπ)Yν (x) = cos(νπ)Jν (x) − J−ν (x) = − I2
π π
with
Zπ Zπ
I1 = cos(νπ) cos(x sin(t) − ν(t)) dt − cos(x sin(t) + νt) dt,
0 0
Z∞
I2 = e−x sinh(t) (cos(νπ)e−νt + eνt ) dt.
0
Since
cos(νπ) cos(x sin(t) − νt) = cos(x sin(t) + ν(π − t)) + sin(νπ) sin(x sin(t) − νt)
and
Zπ Zπ
cos(x sin(t) + ν(π − t)) dt = cos(x sin(t) + νt) dt,
0 0
Rπ
we have I1 = sin(νπ) 0 sin(x sin(t) − νt) dt and this proves the formula for ν ∈
/ Z.
Because of the continuity, we therefore get it for all ν.
The formula for Iν is proven completely analog to that for Jν and that for Kν follows
then from the definition, similarly to those for Yν . q.e.d.
With this theorem we can again follow that Yn and Kn exist for n ∈ Z.
With the generating function, these formulae can be proven even simplier, at least for
n ∈ Z. We will show this as an example for the functions Jn .
We use
Zπ
eint dt = 2πδn0
−π
to get
Zπ X Zπ
eiz sin θ e−inθ dθ = Jk (x) eikθ e−inθ dθ = 2πJn (x).
−π k∈Z −π
So
Zπ Zπ
1 i(x sin θ−nθ) 1
Jn (x) = e = cos(x sin θ − nθ) dθ.
2π π
−π 0
Then we have
∞ Zπ
x n X (−1)k x 2k 1
yn (x) = and yn (x) = cos(x sin φ − nφ) dφ.
2 (n + k)!k! 2 π
k=0 0
Z 4u2 −x2
1 e 4u 2
yn (x) = 2 n+1 n+1 x du
2πi (x) u
γ
Z
1 x n x2
= eu e− 4u u−n−1 du
2πi 2
γ
∞ ∞
(−1)k x 2k 1
x n XX Z
= um−k−n−1 du
2 m!k! 2 2πi
m=0 k=0 γ
∞
x n X (−1)k x 2k
= .
2 (n + k)!k! 2
l=0
Z2π x iφ −iφ )
1 e 2 (e −e
yn = dφ
2π eniφ
0
Z2π
1
= exp(xi sin φ − niφ) dφ
2π
0
Zπ
1
= cos(x sin φ − nφ) dφ.
π
0
q.e.d.
From this integral representation we can then easily, using partial integration, see that
this function solves the Bessel differential equation.
2.5 Asymptotics
Now we want to use the proven integral representations to get asymptotic formulae.
Theorem 2.13 For x ∈ R, x → ∞ we have
r
2 π νπ
Jν (x) ∼ cos(x − − ),
πx 4 2
r
2 π νπ
Yν (x) ∼ sin(x − − ),
πx 4 2
r
1 x
Iν (x) ∼ e ,
2πx
r
π −x
Kν (x) ∼ e .
2x
Proof: • Jν and Yν :
The second integrals in both of the representations go to 0 exponentially, therefore
we have
Zπ
1
Jν (x) + iYν (x) = ei(x sin t−νt) dt + O(x−A )
π
0
2e−iπν/2
Jν (x) + iYν (x) = (I1 (x) + I2 (x)) + O(x−A )
π
with
Zπ/3 Zπ/2
I1 (x) = eix cos t cosh(νt) dt, I2 (x) = eix cos t cosh(νt) dt.
0 π/3
Z1/2
I2 (x) = eixu φ(u) du
0
1/2 Z1/2
eixu 1
eixu φ0 (u) du = O(x−1 ),
I2 (x) = φ(u) −
ix 0 ix
0
and so
√ Z∞
ix 2
I1 (x) ∼ 2xe eit dt.
0
R∞ √
eit2 dt = π −iπ/4
Using 0 2 e we have
r
2 i(x−pi/4−πν/2)
Jν (x) + iYν (x) ∼ e ,
πx
• Iν :
R∞
Again we have 0 exp(−x cosh t − νt) dt → 0 so we only have to consider the other
integral. We split this in two integrals, on from 0 to π/2 and the other from π/2 to
π. Since cos(t) ≤ 0 in the second interval, the second integral is bounded. For the
√
first one we set u = 2 x sin(t/2) to get
√
Z 2x
ex
− u2
2
−1 u du
Iν (x) = √ e cos 2ν sin √ q + O(1).
π x 2 x 1− u2
0 4x
R∞ u2
e−
pπ
If now x → ∞, the integral goes to 0
2 = 2.
• Kν :
√
Let u = 2 x sinh(t/2). Then
Z∞
e−x
2
− u2 −1 u du
Kν (x) = √ e cosh 2ν sinh √ q .
x 2 x 1+ u2
0 4x
R∞ u2
e−
pπ
Again, for x → ∞ the integral goes to 0
2 = 2.
q.e.d.
In the last section, we want to compute three of the most used integral transforms. Here
we will restrict to the Bessel function J0 (x).
Theorem 2.14 (Laplace Transform) We have
1
L[J0 ](s) = √ .
1 + s2
d
0=− L[y 00 + y] + L[y 0 ]
ds
d
= − (s2 L[y](s) + L[y](s) − sy(0) − y 0 (0)) + sL[y](s) − y(0)
ds
= −(1 + s2 )L[y]0 (s) − sL[y](s),
!
1 1
= lim p +p
ε→0 1 + (ε − is)2 1 + (ε + is)2
2
√
1−s2
, |s| < 1
= ,
0, |s| ≥ 1
2s−1 πs s 2
M[J0 ](s) = sin Γ .
π 2 2
Proof: We use the integral representation for J0 . Since 0 < <(s) < 1, we can interchange
the two occuring integrals. The substitution y = x sin(t) then yields
Zπ Z∞
1
M[J0 ](s) = xs−1 cos(x sin(t)) dx dt
π
0 0
Zπ Z∞
1 −s
= sin(t) dt y s−1 cos(y) dy.
π
0 0
Using
Z∞ Zπ πs Γ( s )
πs √
s−1
t cos(t) dt = cos Γ(s) and sin(t)−s dt = π tan 2
2 2 Γ( s+12 )
0 0
1√ πs πs s Γ(s)
M[J0 ](s) = π tan cos Γ
π 2 2 2 Γ( s+1
2 )
2s−1 πs s 2
= sin Γ .
π 2 2
q.e.d.
References
[Coh07] Cohen, Henri: Number Theory Volume II: Analytic and Modern Tools. Springer,
2007
[Dav01] Davies, Brian: Integral Transforms and Their Applications. Third Edition.
Springer, 2001
[GR00] Gradshteyn, I.S. ; Ryzhik, I.M.: Table of Integrals, Series and Products.
Sixth Edition. Academic Press, 2000