Sie sind auf Seite 1von 13

Poularikas A. D. “Legendre Polynomials” The Handbook of Formulas and Tables for Signal Processing. Ed. Alexander D. Poularikas Boca Raton: CRC Press LLC,1999

© 1999 by CRC Press LLC

21

Legendre Polynomials

21.1 Legendre Polynomials

21.2 Legendre Functions of the Second Kind (Second Solution)

21.3 Associated Legendre Polynomials

21.4 Bounds for Legendre Polynomials

21.5 Table of Legendre and Associate Legendre Functions

References

21.1 Legendre Polynomials

21.1.1 Definition

P

n

( ) =

t

[/] =

n

2

2

[/]

n

k = 0

(

12 ) (

kn

k

n

2

)!

kt

nk 2

2

n

!(

k

)!(

n

2

k

)!

/ 2

n

(

n

1

)/

2

n even

n odd

21.1.2 Generating Function

∞   ∑ P ( ts ) n s < 1 n 1 wts
P (
ts
)
n s
< 1
n
1
wts (,
) =
=
n = 0
2
1 −
2 st
+
s
n
− 1
P (
ts
)
− > 1
s
n
 
n = 0
w ( –t,–s ) = w ( t,s )

21.1.3 Rodrigues Formula

P

n

1

d

n

( ) =−=(

t

2

n

n ! dt

n

t

1

)

n

2

n

012 L

,,

21.1.4 Recursive Formulas

1.

2.

(n

+

1

)P

n

+

1

()t

−+

2

(

n

1

P

()

t

tP

nn1

+

()

t

=

(

n

+

© 1999 by CRC Press LLC

)tP ()t

+

nP

nn

1)

P

n

()

t

(

P

1

()t

=

0

n

=

12

,

, L

()

t

=

derivative of

()

P t

n

=

generating function

012,, ,L

3.

4.

5.

6.

()

tP

P

P

t

nn

()

t

P

1

()

t

=

nP

n

()

t

=+ (2

()

t

=

n

1)

Pt

n

nP

()

()t

ntP ()t

() =

nn 1

t

nn1

+

1

2

(t

Pt

1

0

() = 1

)P ()t

n

Pt

1

TABLE 21.1 Legendre Polynomials

P 0

P 1

P

2

= 1

= t

=

3

2

t

2

1

2

Ptt

3

=

2

2

5

3

3

Pt

63

Pt

4

=−+

8

8

5

70

3

8

15

8

35

4

30

t

tt

2

3

5

=−+

8

8

Pt

t

P

6

=−+−

16

16

16

7

=−+−

16

16

16

16

429

7

693

5

315

tt

3

35

231

6

315

tt

t

4

105

2

5

16

Figure 21.1 shows a few Legendre functions.

n

n

= 1 2 L

=

12, ,L

,

,

shows a few Legendre functions. n n = 1 2 L = 12, , L ,

FIGURE 21.1

21.1.5 Legendre Differential Equation

If

y

For

=

(

Px

n

t =

cos

) (

ϕ

:

n

= 012,, ,L)

is a solution to the second-order DE

(1

t

2

)

y′′ −

2

1

d

ϕ 

dy

d ϕ 

sin ϕ

d

sin

ϕ

++= nn ( 1 ) y

ty′ ++=n ()10 n

y

0

© 1999 by CRC Press LLC

Example

From (21.1.4.4) and t = 1 implies

For

01

=−

nP

n

( )

nP= 2

,

2

1 ==P 11

1

()

()

and so forth. Hence

11

nP

n

1

( ) or

P n ()1 = 1.

( )

P

=

P

nn

1

( 1 ).

21.1.6 Integral Representation

1. Laplace integral:

2. Mehler-Dirichlet formula:

3. Schläfli integral:

1 Pt ) = cos ϕ ] n d ϕ n ( π 2 cos(
1
Pt
) =
cos
ϕ
]
n d
ϕ
n (
π
2
cos(
n +
1
) ψ
P n (cos
θ )
=
∫ θ
2
π
2 cos
ψ
cos
θ
0
n
1
(
z
2 − 1
)
P
( t
) =
dz
n
n + 1
2 π j
2
n (
z
− t
)
C

d

ψ

For

nP= 1

,

1

0

θπ ,

<<

n

1 ==P 11

()

0

()

=

012 ,,

,

.

where C is any regular, simple, closed curve surrounding t .

21.1.7 Complete Orthonormal System

{[

1

2

(

2n

+

1

)]

1

/

2

( )}

Pt

n

The Legendre polynomials are orthogonal in [–1,1]

and therefore the set

is orthonormal.

1 ∫ P () t P () t dt = 0 n m − 1
1
P
()
t
P
()
t
dt = 0
n
m
1
1
2
[
P
( )]
t
2 dt =
n = 012
,
,
L
n
2
n +
1
1
2
n +
1
() t =
Pt()
n
= 012L
,,
ϕ n
n
2

21.1.8 Asymptotic Representation:

θ

P n (cos )

2 π n sin θ
2
π
n sin
θ

δ = fixed positive number

21.1.9 Series Expansion

If f ( t ) is integrable in [–1,1] then

© 1999 by CRC Press LLC

ft () =

n = 0

aPt ()

n

n

sin

 


n +

1


π 

4

2

θ +

1

−<< t

1

,

n →∞

,

δθπδ

a

n

=

2

n

+

1

2

1

1

f

() ()

P

t

n

t

dt

n = 012 ,,

L

For even f ( t ), the series will contain term P n ( t ) of even index; if f ( t ) is odd, the term of odd index only. If the real function f ( t ) is piecewise smooth in (–1,1) and if it is square integrable in (–1,1), then the series converges to f ( t ) at every continuity point of f ( t ). If there is a discontinuity at t then the series

converges at

[ ft( ++0)

ft( 02)]/

.

21.1.10 Change of Range

If a function f ( t ) is defined in [ a,b ], it is sometimes necessary in the applications to expand the function in a series in the applications to expand the function in a series of orthogonal polynomials in this interval. Clearly the substitution

t =

2 b

+ a

a

x

2

 

 

x

=

b

b

+ a

b

,

,

ab <

a t +

22

transform the interval [ a,b ] of the x-axis into the interval [–1,1] of the t-axis. It is, therefore, sufficient to consider

  

b

a

b

+

a

aPt

2

t +

 

2

=

n = 0

n

n

( )

   

1

=

2

n

+

1

f

b

a t +

b

+

a

n

2

1

22

f

a

P

n

( )

t

dt

The above equation can also be accomplished as follows:

Example

Suppose f ( t ) is given by

Then

© 1999 by CRC Press LLC

∞ ft () = ∑ aX () t n n n = 0 nnn 1
ft
()
=
aX
()
t
n
n
n = 0
nnn
1
dt
(
a
)(
t
b
)
X
( ) =
t
n
n
n
nb
!(
a
)
dt
a
2
n
+
1
a
=
f
()
t
X
() t
dt
n
n
b
a
b
0
−≤
1
t
<
a
f
( ) =
t
1
a
<
t
1
1
2
n +
1
a
=
P
( )
t
dt
n
n
2
a

Using (21.1.4.4), and noting that P n (1) =1 we obtain

a

=−

1

2

[

Pa

(

)

Pa

(

nn

+

11

n

)],

which leads to the expansion

ft

( )

Example

Suppose f ( t ) is given by

1

1

)

2

2

(

1

a

n = 1

[

P

n

f

( ) =

t

+

1

(

a

)

a

0

=

1

2

(1

a

)

( ),

P

1

a

(

)]

Pt

nn

11

10

−≤

t

<

t

<

1

0

−<< 1

t

1

The function is an odd function and, therefore, f ( t ) P n ( t ) is an odd function of P n ( t ) with even index. Hence a n are zero for n = 0,2,4,…. For odd index n , the product f ( t ) P n ( t ) is even and hence

a

1

1

1

=

n

+

f

()

t

P

dt

=

2

n

+

nn

2

2

()

t

1

1

0

P

n

()

t

dt

n = 135 ,

,

,L

Using (21.1.4.4) and setting n = 2 k + 1, k = 0,1,2… we obtain

a 2

k

+

1

k

4

=+

(

=

[

P

k

22

+

1 1

3

)

0

( )

t

P

2

k

[

P

( )

t

dt

=

( )

t

P

++

1

kk

22

2

0

( )]

Pt

2

k

1

0

=− P ( 0

2

)

P

kk

22

+

(

0

)

( )]

t

dt

where we have used the property P n (1) =1 for all n . But

and, thus, we have

The expansion is

© 1999 by CRC Press LLC

a

2

k +

1

f

=

=

 

P

2 n

(

0

)

= −

1

2

 =

(

1

2

)

n

n

(

2 n

2

)!

 

(

1

)

k

(

2 k

)!

n

k

+ 1

()( 1

k

22 +

)

2

(

(

n

!)

1

)

k

(

2 k

)!

1 +

2

k

+

1

(

2

2

k

(

k

!)

2

k

12 )

(

k

)!(

4

2

k

k +

22

+

3

)

[(

k +

1

)!]

2

=

 

2

2

k

(

k

!)

2

2

k

+

2

 

2

2

k +

1

k

!(

k +

1

)!

 

( ) =

t

n = 0

(

k

12 ) (

k

)!(

4

k

+

3

)

2

2

k +

1

k

!(

k +

1

)!

( )

Pt

2

k +

1

−≤≤ 1

t

1

21.1.11 Expansion of Polynomials

If

q m

n

+

m

k = 0

1

( t ) =

1

2

1

q

m

cx

k

k

()

t P

n

is an arbitrary polynomial, then

q

m

() =

t

cP

0

0

() +

t

() ++

cP t

11

L

c

P

mm

()

t

where

()

t

dt

=

0

,

n

=

012 ,,

L.

If q m ( t ) is a polynomial of degree

m

and m

<

r ,

1

1

()

q t

m

P

r

()

t

dt

=

0

,

m

<

r

.

Example

To find P 2 n (0) we use the summation

P

2 n

( ) =

t

with k = 0. Hence

(

1

)

n

n

k = 0

(

k

12

)

(

n

+− k

2

1

)!

2

2

n

1

(

2

knk

)!(

+−

1

)!(

nk

)!

t 2 k

P

2 n

(

0

) =

(

n

12

)

(

n

1

)!

(

n

12

)

n

[(

2

n

1

)!]

(

1

)

n

(

2 n

)!

=

=

2

21

n

(

n

1

)!

n

!

2

2

n

[(

nn

1

)!]

n

!

2

2n

(

n

!)

2

Example

To evaluate

1

0

m ( )

P

t

dt

for m 0 we must consider the two cases: m = odd and m = even.

(a)

m = even and m 0

(b)

∫∫∫ 1

1

0

1

1

P

1

P

()

t

dt

==

()

t

dt

P

mmm

2

1

2

−−

1

1

2

()

t

1

dt

=

1

1

P

m

()

t

P

0

()

t

dt

= 0

The result is due to the orthogonality principle.

m = odd and m 0. From the relation (see Table 21.2)

1

1

P

m

( t )

dt =

with t = 0 we obtain

1

2

m

+

1

[

PtPt

1

+

1

( )

( )]

mm

1

0

P

m

( )

t

dt =

1

2

m

+

1

[

P

1

(

0

)

P

mm

+

1

(

0

)]

Using the results of the previous example, we obtain

   m − 1 m + 1  1 ( − 1 )
m
1
m
+ 1
1
(
1
)
2
(
m
1
)!
(
1
)
2
(
m
+
1
)!
∫ 1
P
( )
t
dt =
m
2
2
2
m +
1 
0
m − 1
m + 1
m
1
m
+ 1
2
!
2
!
2
2
m − 1
m − 1
(
1
)
2
(
m
12 )!(
mm
++ 11 )
1
)(
(
1
)
2
(
m
1
)!
=
=
m
+ 1
mm − 1
+ 1
 
m
+ 1
m − 1
m
+ 1
m
()
2
m
+
12
!
2
!
2
  !  
2
 
2
2
  !  
2

m = odd

© 1999 by CRC Press LLC

c =

n

then

21.2

Legendre Functions of the Second Kind (Second Solution)

21.2.1 Second Kind:

1 1 1. + t Q = ln , t < 1 ; 0 2
1
1
1. + t
Q
=
ln
,
t
< 1 ;
0
2
1
− t
1
1
+ t
2. Qt ( )
=
t
ln
− 1
,
t
< 1 ;
1
2
1
− t
2
n +
1
n
3.
Q
( t )
=
tQ
( t )
Qt ( ),
n+
1
nn−
1
n + 1
n +
1
1
[
(
n − 1
)]
2
241
n
k
4.
Q
()
t
=
PtQ
()
()
t
n
n
0
(
2
k
+
1
)(
nk
k = 0
1
for [
( n − 1
)] see 21.1.1.
2

)

n =

1,2,L

( ),

P tt

n

2

k

1

<

1

,

n

=

12 ,

, L

21.2.2 Recursions

Q n ( t ) satisfies all the recurrence relations of P n ( t ).

21.2.3 Property

1

x

t

=

(

n = 0

2

n

+

21.2.4 Newman Formula

Q

n

( ) =

t

P

1

n

x

2

1

t

x

1

(

)

dx

,

1

)

P () tQ

n

n

(

x

)

n = 012 ,, L

21.3 Associated Legendre Polynomials

21.3.1 Definition

If m is a positive integer and –1 t 1, then

m

Pt

n

()

=−1

(

t

2

)

m /

2

m

dPt

n

( )

dt

m

m

= 12 L

,,

,

n

where

P n m ( t )

is known as the associated Legendre function or Ferrers’ functions .

21.3.2 Rodrigues Formula

P

m

n

( ) =

t

(

1 t

2

)

m /

2

d

n

+

m

2

n

n !

dt

n

+

m

(

t

2

1

n

),

m

=

12 ,,

L

,;

nnm

+≥

21.3.3 Properties

1.

P

n

m

()

t

=

(

1

)

m

(

n

m

)!

(

n

+

m

)!

P

m

n

( )

t

© 1999 by CRC Press LLC

0

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

0 () =

P

n

t

()

Pt

n

1

m

n +

)P

(n

−+ m

1

()t

(

21

n

+

m

)tP ()t

n

++ (n

m

m)P

n

1

()t

=

0

(

1

t

2

)

12

/

m

Pt

n

( )

=

1

2

n

+

1

[

m + 1 n + 1

P

( )

tP

m + 1

n

1

( )]

t

1

(

1

n

2

(

2

)

+

/

12

1

P

n

2

t

/

12

[(

++

n

m

n

n

(21)(

P

m

n

(

t

1

)

=

mt

21

( )

t

[

(

n n

1

+−

(

m

n

n

1

)

( )

t

−+

P

m +

( )

t

)

1)

P

n

m

Pt

n +

1

(

=

( )

=

Pt

n

1

m

(

t

m

t

)

2

m

n

1)

tP

( )

t

12

/

=

n

+

m

+−

n

1

m m

(

m P

)

m

n

1

−−+

n

m

1

)]

( )]

t

1

( )

t

t

2

)

12

/

m

( )

Pt

[(

)(

mn

1

)

P

m 1

( )

t

(

P

m

n

m

Pt

n

()

=

(

t

2

1

)

m /

2

m

dPt

n

( )

dt

m

1

1

1

1

P

m

n

()

t

m

P

k

[

P

m

n

( )]

t

2

()

t

dt

=

0

k

n

dt =

2 n

(

+

m

)!

(

2

n

+

1

)(

nm

)!

21.3.4 Differential Equation

12

)(

n

m

−+

)

m 1 n + 1

P

2 m dP ( t ) m  2  t dP ( t )
2
m
dP
( t )
m
2
t dP
( t )
m
2
n
n
m
(
12
t
)
+
 n ( n
+−
1
)
 P
( t )
2
2
n
dt
dt
(
1 − t
)
21.3.5 Schlafli Formula
2
n
(
n
+
m
)!
(
x
− 1
)
m
2
m /
2
P
( ) =
t
(
1 − t
)
dx
n
n
nm
+
+ 1
2 π jn !
2
(
x
t
)
C

= 0

( )]

t

where C is any regular closed curve surrounding the point t and taking it counterclockwise.

21.4 Bounds for Legendre Polynomials

21.4.1 Stieltjes Theorem

4 1 P n (cos ) γ ≤ 2 , π n sin γ
4
1
P n (cos )
γ
2
,
π
n
sin
γ

0

<<γ

π

,

21.4.2 Second Stieltjes Theorem

4 Pt () − P () t < n n +2 π n + 2
4
Pt
() −
P
()
t
<
n
n +2
π
n +
2

© 1999 by CRC Press LLC

n

=

12 L

,,

21.4.3

21.4.4

dP n ( ) t n < 2 1 , t < , n =
dP
n ( )
t
n
<
2 1
,
t
<
,
n =
12
,, L
2
dt
π 1 − t
62
1
1
1 () t
+ Pt ()
<
,
t
< 1
P n+
n
π
n
1 −
t

21.5 Table of Legendre and Associate Legendre Functions

TABLE 21.2 Properties of Legendre and Associate Legendre Functions [P n (t) = Legendre Functions, = Associate Legendre Functions, Q n (t) = Legendre Functions of the Second Kind]

n m ( )

P

t

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

∞ 1 = ∑ Ptx n ( ) n t ≤< 1 1 x 2
1
= ∑
Ptx n ( )
n t
≤< 1
1
x
2
1
2 tx
+
x
n = 0
[/] n
2
k
nk − 2
(
12 (
)
n
2
kt
)!
n
P
( t
) =
[
n /
2 ]
=
n
n
2
kn
!(
k
)!(
n
2
k
)!
2
k = 0
P
( t ) = 1
0
= −
(
− 1
)
n (
2 n
)!
0 )
P 2
n (
2
 
n
2 1 
  =
2 2
n (
n !)
( 0
) =
0
P 2n 1
+
P
(− ) =
t
Pt
()
P
(− ) =−
t
P
()
t
2nn2
21n
+
21n
+
n
P
(− ) =−( 1)
t
Pt
()
n
n

Pn ()1 = 1

n

= 012,, L;

P

n

(1) =−( 1)

n

1

d

n

n

2

n

n !

()t

tP

)P

(n

P

1

+

n

+

1

+

n 1

t

()

()

()

()

t

t

t

=

=

2

P

P

n 1

n

P

+

n 1

P

P

n

tP

nn1

+

d t

(

t

()

n

21

−+

n

+

P

P

( ) =

t

(

t

() t

tP

P

P

1

nn

nn1

+

()

t

P

1

Rodrigues formula,

1)

() t

+

P

n

1)

P

P

nP

1

()t

0

=

0

2

−=)

1

)tP ()t

+

n

nn

()

t

n

()

t

()

()

t

t

=

nn 1

()

t

()

t

+

+

2

2

tP

()

t

()

() t

t

()

t

=

=

=

nn 1

+

tP

n

nP

(2

(

nn 1

+

()

t

n

n

()

Pt

n

()

t

P

(1

t

2

)

P

() t

=

nP

nn

1

() t

ntP

n

() t

− t 2 ) ′ P () t = nP nn − 1 () t −

< 1,

−<<11

t

P

2 n

( t ) =

(

1

)

n

n

k = 0

(

k

12 ) (

n

+− 2 k )!(

1

)!

nk

2

2

n

1

(

2

k

)!(

)!

nk +−

1

P

(1

1

1

1 1

2

)

t

n

n ( )

P

t

dt

( t )

=+ (

n

= 0

1)[

( t )

tP

nn + 1

P

( t )]

t

2 k

© 1999 by CRC Press LLC

n

=

even ;

[

nn 2

/

]

(

=−

n = 1,2,L

n = 012, ,

n = 012, ,

n = 012, ,

n = 012, ,

n = 012, ,

L

L

L

L

L

n = 1,2,L

n = 1,2,L

n = 1,2,L

n = 1,2,L

n

n

n

t
t

= 012, ,

= 012, ,

L

L

= 1,2,L

1

12 )/

n

=

odd

TABLE 21.2 Properties of Legendre and Associate Legendre Functions [P n (t) = Legendre Functions, = Associate Legendre Functions, Q n (t) = Legendre Functions of the Second Kind] (continued)

n m ( )

P

t

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

1

1

1

1

P

n

()

t

P

m

[P (t)]

n

1

2

()

t

dt

dt =

=

0

2

2

n +

1

n m

n = 012, ,

L

mm ( − 2 )( L m −+ s 2 ) m 1 P m,
mm (
2
)( L
m
−+ s
2
)
m
1 P
m, s = even
2 ∫
t
( t )
dt =
s
( ms ++
1 )(
ms +−
11
)
L
(
m
+
)
1
1
(
m
13 )( m
)
L (
ms −+
2
)
m
2 ∫
1 P
t
( t )
dt =
m, s = odd
s
( ms ++
1 )(
ms +−
12
)
L
(
m
+
)
− 1
1
2 n
t P
n () t
P
() t
dt =
n = 1,2,L
n − 1
2
4
n
− 1
1
1
∫ P
() t
P
()
t
dt
= 2
n = 012, ,
L
n
n + 1
1
1
2 n
t P
n ′ () t
P
() t
dt =
n
= 012, ,
L
n
2
n +
1
1
1
2
∫ (1
− t
) P ′ ()t P ′ ()t dt
= 0
k
≠ n
n
k
1
1
2
2
− 1
/
2
(
1 −
t
)
P
() t
dt
=
n = 012, ,
L
n
2
n +
1
1
1
2
n ( n +
1
)
2
t
P
() t
P
() t
dt =
) n = 1,2,L
n
+
1
n
1
2
(
4
n
12 )(
n
+
3
1
1
2
n ( n +
1
)
2
( t
− 1
)
P
() t
P