Nama : Geovani Novelina Sihombing (190801010)

Delima br Waruwu (190801060)

Christine Feby Febyola S (190801066)
Nelly L Ompusunggu (190801076)
Desima Roida Glori Simorangkir (190801088)

18.1.1 Assume that the Hermite polynomials are known to be solutions of the Hermite ODE, Eq.
(18.1). Assume further that the recurrence relation, Eq. (18.3), and the values of Hn(0) are
also known. Given the existence of a generating function
∞
tn
g ( x , t ) =∑ H n ( x )
n=0 n!
(a) Differentiate g(x,t) with respect to x and using the recurrence relation develop a first-
order PDE for g(x,t).
(b) Integrate with respect to x, holding t fixed.
(c) Evaluate g(0,t) using the known values of Hn(0).
(d) Finally, show that g ( x , t )=exp ⁡(−t 2+2 tx)

Answer :
It is convenient to use the formula in Eq. (18.4) for H 1n, but it was not one of the relation
ships whose validity was to be assumed. However, Eq. (18.4) can be derived from the
ODE, assuming the validity for all x of the basic recurrence formula, Eq. (18.3). Proceed
by forming
'' '
( H n +1−2 x H n +2 n H n−1 ) −2 x ( H n +1−2 x H n +2 n H n−1) +¿
2 n ( H n+1−2 x H n+ 2n H n−1 )=0

then expand the parenthesized derivatives and use the Hermite ODE to cancel as many
terms as possible. In this way we reach
−2 H n+1 +4 x H n +4 n H n−1−4 H ' n=0 ,
Equivalent, again using Eq.(18.3), to
H 'n=2 n H n−1
(a) Now differentiate g(x, t) with respect to x and use the above derivative formula
 ∞ ∞
∂g tn tn
=∑ H 'n (x ¿) =∑ 2 nH 'n−1 =2 t g ( x ,t ) . ¿
∂ x n=0 n! n=0 n!
(b) This is a separable first-order equation with solution
g ( x , t )=e 2 tx f ( t ) .
n
(c) Find f ( t ) by setting x=0 and using H 2 n ( 0 )=(−1 ) ( 2 n ) ! /n ! And H 2 n+1 ( 0 )=0. We get
∞
(−1)n t 2 n −t 2

g ( x , 0 ) =f ( t )=∑ =e
n=0 n!
2

(d) The final result is g ( x , t )=e 2 xt −t

18.1.3 In developing the properties of the Hermite polynomials, start at a number of different
points, such as:
1. Hermite’s ODE, Eq. (18.1),
2. Rodrigues’s formula, Eq. (18.8),
3. Integral representation, Eq. (18.10),
4. Generating function, Eq. (18.2),
5. Gram-Schmidt construction of a complete set of orthogonal polynomials over (−∞, ∞)
with a weighting factor of exp(−x2 ) (Section 5.2).
Outline how you can go from any one of these starting points to all the other points.

Answer :
The connection of these starting points can, of course, be accomplished in many ways.
Starting with the ODE, one can derive a Rodrigues formula as shown in Eq. (18.8), and
therefrom, as in Eq. (12.18), a Schlaefli integral representation. Applying Eq. (12.18), we
can use the Schlaefli integral to obtain a generating function, which can in turn be applied,
e.g., by developing recurrence formulas and then using them as in the text after Eq. (18.7),
to recover the Hermite ODE. Since these steps form a closed loop, we can regard any of
the four relationships as a valid starting point.
From the Rodrigues formula we can also, by repeated integrations by parts, establish the
orthogonality of the Hermite polynomials on (−∞, ∞) and note the associated weighting
factor; conversely, the weighting factor suffices to determine the ODE that yields the
orthogonal polynomials. These last relationships mean that we can begin an analysis from
any of the five listed starting points.
18.1.3 Prove that |H n ( x )|≤|H n ( ix )|.
Answer :
From Eq. (18.9) we see that

[ n /2 ] [ n/ 2]
n !(4 x 2)−s 2 −s
−n −n s n ! (4 x ) −n
i ( 2 x ) H n ( ix )=∑ ≥ ∑ (−1) =( 2 x ) H n ( x)
s=0 ( n−2 s ) ! s ! s=0 ( n−2 s ) ! s !
because the first sum has only positive terms. Taking the absolute values of the first and

last members of the above relation and multiplying through by |( 2 x )n|| we obtain the
desired inequality.

18.1.4 Rewrite the series form of H n ( x ) , Eq. (18.9), as an ascending power series.
Answer :
n
n ( 2 n) !
H 2 n ( x )=(−1) ∑ (−1)2 s (2 x )2 s ( 2 s ) ! ( n−s ) ! ,
s=0

n
n ( 2n ) !
H 2 n+1 ( x )=(−1) ∑ (−1)2 s (2 x)2 s+ 1 ( 2 s ) ! ( n−s ) ! ,
s=0

18.1.5 (a) Expand x 2 r in a series of even-order Hermite polynomials.

(b) Expand x 2 r+1 in a series of odd-order Hermite polynomials.
Answer :
(a)
2r(2 r )! r H 2 n (x)
x = 2r ∑
2 n=0 ( 2n ) ! ( r −n ) !

2 r+1 ( 2 r +1 ) ! r H2n( x )
(b) x = 2 r+1 ∑ r=0,1,2
2 n=0 ( 2 n+1 ) ! ( r −n ) !

Hint. Use a Rodrigues representation and integrate by parts.

18.1.6 Show that

 (a)

∞
−x 2
∫ H n ( x ) exp
−∞
[ ] {
2
dx = 2 πn! / ( n/ 2 ) ! , n even
0 , n odd .
(b)
∞ 0 , n even .

−∞
−x 2
[ ]
∫ H n ( x ) exp 2 dx = 2 π
{ n+1
(( n+1)/2 ) !
, n odd

Answer
The answers in the text are incorrect. In each answer, replace 2π by (2π) 1/2 .
(a) Using the generating function, form
∞ ∞ ∞ ∞
−x
2/2 2
−t +2 tx−x
2/2 2/2
t2
∫e g ( x ,t ) dx=¿ ∫ e dx=¿ ∑ ∫ e−x H n ( x ) dx ¿ ¿
−∞ −∞ n=0 −∞ n!
Now convert the central member of this equation to obtain a power series in t, by first
completing the square in the exponent and performing the x integration, and then
expanding the resultant function of t. Setting y = x − 2t we get
∞ ∞
2 2 /2

∫ e−t +2 tx− x dx=¿ ∫ e−¿ ¿ ¿ ¿

−∞ −∞

¿¿
We now equate the coefficients of equal powers of t in the last two equations, noting from
the second of these equations that the integrals involving H n of odd n vanish, and that
those of even n correspond to the equations
∞
−x
2/2 H2m (x ) (2 π )1 /2
∫e (2 m) !
dx=
m!
−∞

This formula is equivalent to the corrected form of the answer.

(b) This part of the exercise can be treated in a way similar to that of part (a). The relevant
equation set is
∞ ∞ ∞
tn 2/2

∑ n ! ∫ e−x x H n ( x ) dx= ∫ e−¿¿ ¿

n=0 −∞ −∞

∞
2 2 /2
t
¿e ∫ e− y ( y+ 2t )dy =¿ ¿
−∞
 The term of the integral containing a linear factor y vanishes due to its odd symmetry.
The left-hand side of this equation must have zero coefficients for even n, while for
odd n ( set¿ 2m+1) we get
∞ 2
1 2

∫ e− x x H 2 m +1 ( x ) dx=¿
( 2m+1 ) ! −∞
which easily rearranges into the corrected form of the answer.

18.1.7 (a) Using the Cauchy integral formula, develop an integral representation of H n ( x ) based
on Eq. (18.2) with the contour enclosing the point z = −x
(b) Show by direct substitution that this result satisfies the Hermite equation.
Answer :
(a)
2

n! x 2
e−z
H n ( x) = e ∮ dz
2 πi ( z + x)n +1
(b) Defining z + x=w , z 2−x 2=w 2−2 xw we have
2 2

n ! e−w + 2 xw n ! e−w +2 xw
H n ( x) = ∮ dw , H ' n ( x ) =2 ∮ wn d w
2 πi wn +1 2 πi
2

n ! e−w +2 xw
H ' ' n ( x )=4 ∮ wn−1 dw
2 πi
And
2

' n! e−w +2 xw (
H ' ' n ( x )−2 x H n ( x ) +2 n H n ( x )= ∮ n+1
2n−4 xw+4 w2 ) dw ,
2 πi w
2

n! d e−w +2 xw
¿−2 ∮
2 πi dw wn (dw=0 )
where a zero value is obtained because the start and end points of the contour coincide
and the quantity being differentiated is analytic at all points of the contour

d n
(
18.2.1. Prove that 2 x−
dx )
1=H n ( x )

Hint. Check out the cases n = 0 and n = 1 and then use mathematical induction (Section
Answer:
Combining the recursion formulas in Eqs. (18.3) and (18.4), we have
 d
(
H n+ 1 ( x )= 2 x− )
H (x ) .
dx n
This formula suggests that we use mathematical induction, since it shows that if the
formula of this exercise is true for H n ( x ) ., it is also true for H n+ 1 ( x ). To complete a proof
we need to verify that the formula is valid for n = 0, i.e., that
H 1 ( x )=2 x H 0 ( x )−H ' 0 ( x ) .
Since H 0 ( x ) =1 and H 1 ( x )=2 x , this equation is clearly satisfied.
∞
2

18.2.2 Show that ∫ x m e−x H n ( x ) dx=0 for m an integer 0 ≤ m≤ n−1

−∞

Answer :
m
m
Introduce the expansion x ∑ ai H i ( x ) . Then, for m<n, using orthogonality we have
i=1

∞
2

∫ e−x H n ( x ) dx=0.
−∞

18.2.3 The transition probability between two oscillator states m and n depends on
∞
2

∫ x e−x H n ( x ) H m ( x ) dx
−∞

Show that this integral equals π 1/ 2 2n +1 n ! δ m , n−1+ π 1 /2 2n+1 ( n+1 ) ! δ m , n+1. This result shows
that such transitions can occur only between states of adjacent energy levels,m=n± 1
Hint. Multiply the generating function, Eq. (18.2), by itself using two different sets of
variables (x,s) and (x,t). Alternatively, the factor x may be eliminated by the recurrence
relation, Eq. (18.3)
Answer :
1
Using Eq. (18.3) to replace xH n ( x ) by H ( x )+ nH n−1( x), , then invoking orthogonality
2 n+1
and inserting the normalization integral from Eq. (18.15)
∞
2

∫ e−x xH n ( x ) H m ( x ) dx=0
−∞

∞ ∞
1 2 2

¿ ∫ e−x H n +1 ( x ) H m ( x ) dx+ n ∫ e− x H n−1 ( x ) H m ( x ) dx

2 −∞ −∞

1 1
¿ [ 2 ] [ ]
δ n+1 , m+ n δ n−1 , m π 1 /2 2m m!=2n n! π 1 /2 ( n+1 ) δ m ,n +1 δ n−1 ,m .
2
 CHEBYSHEV POLYNOMIALS

18.4.1 By evalulating the generating function for special vallues of x, verify the special values
T n ( 1 )=1 , T n (−1 )n , T 2 n ( 0 )=(−1 )n , T 2 n +1 ( 0 ) =0
Answer:
For x=1 , x=−1 ,∧x=0 the generating function g(x , t ) for T n becomes
∞
1−t 2 1+ t
g ( 1 ,t )= 2
= =1+2 ( t +t 2 +t 3 +… ) =1+∑ 2 t n ,
1−2 t+t 1−t 1

∞
1−t 2 1−t n
g (−1 ,t )= 2
= =1+2 (−t−t 2−t 3+ … )=1+ ∑ (−1 ) 2 t n ,
1+2 t+t 1+t 1

∞
1−t 2 n
g (0 , t)= 2
=1+2 (−t 2+t 4 −t 6 + … )=1+ ∑ (−1 ) 2 t 2 n .
1+t 1

∞
n
Comparing the above with g ( x , t )=T 0 ( x ) + ∑ 2 T n ( x ) t ,we see that
n=1

T n ( 1 )=1 , T n (−1 )n , T 2 n ( 0 )=(−1 )n , T 2 n +1 ( 0 ) =0

18.4.2 By evaluating the generating function for special values of x, verify the special values
U n ( 1 )=n+1 , U n (−1 )=(−1 )n ( n+1 ) ,U 2 n ( 0 ) =(−1 )n ,U 2 n +1 ( 0 )=0
Answer:
For x=1 , x=−1 ,∧x=0 the generating function g ( x , t ) for U n becomes
1 1 d 1 d
g ( 1 ,t )= 2
= 2
=
1−2 t+t (1−t ) dt 1−t dt ( )
= ( 1+t +t 2+ t 3 +… )
 ∞
¿ 0+1+2 t+ 3t +…=∑ ( n+1 ) t n ,
2

n=0

1 1 −d 1 d
g (−1 ,t )= 2
=
1+2 t+t ( 1+t ) 2
= = (1−t +t 2−…)
dt 1+ t dt ( )
∞
n
¿ 0+1−2t +3 t 2 −…=∑ (−1 ) ( n+ 1 ) t n ,
n=0

∞
1 n
g (0 , t)= = (−1 ) t 2 n .
2 ∑
1+t 1

∞
n
Comparing with g ( x , t )=∑ U n ( x ) t , we see that
n=0

U n ( 1 )=n+1 , U n (−1 )=(−1 )n ( n+1 ) ,U 2 n ( 0 ) =(−1 )n ,U 2 n +1 ( 0 )=0

18.4.3 Another Chebyshev generating function is
∞
1−xt
= X n ( x ) t n ,|t|<1
2 ∑
1−2 xt+t n=0
How is X n( x) relate to T n ( x ) and U n ( x )?
Answer:
X n ( x )=T n ( x )

18.4.4 Given
( 1−x 2 ) U 'n' ( x )−3 x U 'n ( x ) +n ( n+2 ) U n ( x )=0
Show that V n ( x ) , Eq. (18.106), satisfies
( 1−x 2 ) V 'n' ( x )−x V 'n ( x )+ n2 V n ( x )=0
Which is Chebyshev’s equation.
Answer:
Using Eq. (18.109), evaluate the terms of the ODE for V n in terms of U n−1 and its derivatives,
with the aim of showing that the ODE suggested for V n is automatically satified, given the ODE
given for U n−1. Specifically ew have
1 /2
V n ( x )=( 1−x 2 ) U n−1(x ),
1
' 2 2 ' x
V ( x )=( 1−x ) U n−1 ( x )−
n 1
U n−1 ( x ) ,
2 2
( 1−x )
 1
'' 2 2 '' 2x '1
V ( x )=( 1−x ) U n−1 ( x )−
n 1
U 'n −1 ( x )− 3
U n−1 ( x ) .
2 2 2 2
( 1−x ) ( 1−x )
Using the above, form
3 1
2 ' ' 2 2 2 ''
( 1−x ) V ( x ) −x V ( x ) + n V n ( x )=( 1−x ) U
n n n−1 ( x )−3 x ( 1−x ) U 'n−1 ( x )
2 2

1
2 2 2
+ ( n −1 ) ( 1−x ) U n−1 ( x )
1 /2
¿ ( 1−x 2 ) 2 ''
[ ( 1−x ) U n−1 ( x )−3 x U 'n−1 ( x ) + ( n−1 )( n+1 ) U n−1 ( x ) ] .
The experession within the square brackets vanishes because of the equation satisfied by
U n−1 ( x ) , confirming that V nsatisfies the suggested ODE.

18.4.5 Show that the Wronskian of T n ( x ) and V n ( x ) is given by

n
T n ( x ) V 'n ( x )−T 'n ( x ) V n ( x )= 1 /2
( 1−x2 )
This verifies that T n and V n (n ≠ 0) are independent solutions of Eq. (18.97). Conversely,
for n=o, we do not have linear independence. What happens at n=0? Where is the
“second” solution?
Answer:
Writing the ODE for T n and V n in self-adjoinr form, we obtain (calling the dependent variable y)
n2 y '
[ ( 1−x 2 ) ¿ ¿ 1/2 y ' ] ' + 1 /2
= [ p ( x ) y ' ] + q ( x ) y=0 ,¿
( 1−x 2 )
We use the fact that the Wronskian of any two solutions to this ODE has the general form A/ p ( x ),
in this case
An
W ( T n ,V n ) =T n ( x ) V 'n ( x ) V n ( x )= 1/ 2
( 1−x 2 )
To find the value of An , evalute the Wronskian at x=0 where its value will be An . From Eq.

(18.95), we note that T 'n ( 0 )=n T n−1 ( 0 ) . From Eq.(18.109) we identify V n ( 0 )=U n−1( 0);

differentiating Eq.(18.109), we also find V 'n ( 0 )=nU n−2 ( 0 ) . Our Wronskian therefore (forx=0 ¿ be
written
T n ( 0 ) V 'n ( 0 )−T 'n ( 0 ) V n ( 0 )=n [ T n ( 0 ) U n −2 ( 0 )−T n−1 ( 0 ) U n−1 ( 0 ) ]=A n
All the function values on the left-hand side of this equation are given in Eq. (18.100); the
quantity in square brackets evaluates to -1 for both even and odd n, so
 An n(−1)
W (T ¿ ¿ n , V n)= = ¿
2 1/ 2 1 /2
( 1−x ) ( 1−x 2 )

−1
18.4.6 Show that W ( x )=( 1− x2 ) 2
T n+1 (x) is a solution of
n

¿
Answer:
We know that T n (x) satisfies the ODE
''
( 1−x 2 ) T n ( x )−¿
And the exercise gives W n ( x ) in terms of T n+1 . we can verify the asserted ODE for W n by
rewriting it is an equation in T n+1and comparing with the above ODE, To do so, we need
−1
T n+1 ( x ),
2 2
W n ( x )=( 1− x )
1 x
W 'n ( x )= T 'n +1 ( x ) + T 'n +1 ( x )
1 3 ,
2 2 2
( 1−x ) ( 1−x )
1 x 1+ 2 x 2
W 'n' ( x )= 1
T 'n+1 ( x ) + 3
T 'n+1 ( x ) + 5
T n+1 ( x).
( 1−x ) 2 ( 1−x 2 ) 2 2 2
( 1−x )
Using the above, form
1
( 1−x ) W ( x )−3 x W ( x ) +n ( n+2 ) W n ( x )=( 1−x ) T 'n'+1 ( x )
2 ''
n
'
n
2 2

−1 1+ n ( n+2 )
T 'n+1
'
2 1/ 2
( x )+ 1
T n+1 (x)
( 1−x ) 2 2
( 1−x )
−1
2 2
¿ ( 1−x ) ¿
The expression in square brackets vanishes because of the equation satisfied by T n+1 ( x),
confirming the ODE proposed for W n .

−1
18.4.7 Evaluate the Wronskian of U n ( x ) and W ( x )=( 1− x2 ) 2
T n+1 (x)
n

Answer:
Compare with the method of solution to exercise 18.4.5. The ODE for U n and W n is in self-
adjoint form,
3 '

[(1−x ) y ] + n( n+ 2 ) (1−x )
2 2 ' 2 1 /2
y=0
So the Wronskian has the functional form
An
W ( U n , W n ) =U n ( x ) W 'n ( x )−U 'n ( x ) W n ( x ) W n ( x )= 3 /2
( 1−x 2 )
At x=0 , U 'n' ( 0 )=( n+1 ) U n−1 ( 0 ) ,and U n (0) is given in Eq. (18.100). In addition,

W n ( 0 ) =T n +1 ( 0 ) , W 'n ( 0 )= ( n+1 ) T n ( 0 ) , with T n ( 0 ) also given in Eq. (18.100). Thus

U n ( 0 ) W 'n ( 0 )−U 'n ( 0 ) W n ( 0 ) =( n+1 ) [ U n ( 0 ) T n ( 0 )−U n−1 ( 0 ) T n +1 ( 0 ) ]=n+1
Note that we get the same result for both even and odd n, and that it fixes the constant An as n+1.
Therefore,
n+1
W (U n , W n)= 3/ 2
( 1−x 2)

1
18.4.8 V ( x )=( 1−x 2 ) 2 U ( x ) is not defined for n=0. Show that a second independent solution
n n−1

of the Chebychev differential equation for T n ( x ) ( n=0 ) is V 0 ( x )=arcos x ( ¿ arcsin x ) .

Answer:
Work with the ODE for T 0 in the form given in Eq. (18.104), but written as y ' ' =0 to imply that
we will search for the general solution. Thus,
d2 y
=0 , with solutions y=c0 ∧ y=c 1 θ
d θ2
The solution y=c 0is just c 0 T 0 ( x ) ; Since x=cos θ , the solution y=c1 θ is equaivalen to

π
y=c1 arscosx . Since −arccos x is a linear combination of these two solutions, it is also a
2
solution, more compactly written as arcsin x.

18.4.9 Show that V n ( x ) satisfies the same three-tern recurrence relation as T n (x), Eq.(18.92).
Answer:
Write the proposed recurrence relation for V n with V nwritten in terms of U n−1 according to Eq.
(18.109). Because U n and T n satisfy the same recurrence formula, and it is independent of n, so
also does V n.

18.4.10Transform the series form of T n ( x ) Eq(18.114), into an ascending power series.

n
n m ( n+ m−1 ) ! 2m
ANS T 2 n ( x )=(−1 ) n ∑ (−1 ) (2 x ) ,n ≥ 1 ,
m=0 ( n−m ) ! ( 2 m) !
 n
2 n+1 (−1 )m+n ( n+m ) !
T 2 n+1 ( x )= ∑ ( 2 x )2 m +1 .
2 m=0 ( n−m ) ! ( 2 m+1 ) !

18.4.11Rewrite the series form of U n ( x ) , Eq. (18.115), as an ascending power series.

n
n ( n+m ) !
ANS U 2 n ( x )=(−1 ) ∑ (−1 )m ( n−m ) ! ( 2 m ) ! ( 2 x )2 m ,
m=0

n
n ( n+m+1 ) !
U 2 n+ 1 ( x )=(−1 ) ∑ (−1 )m ( n−m ) ! ( 2m+1 ) ! ( 2 x )2m +1 .
m=0

18.4.12(a) From the differential equation for T n(in self-adjoint form) show that
1
d T m ( x) d T n ( x ) 1 /2
∫ ( 1−x 2 ) dx=0 , m≠ n .
−1 dx dx
(b) Confirm the preceding result by showing that
d T n (x )
=n U n−1 ( x ) .
dx

18.4.13The substitution x=2 x ' −1 converts T n ( x) into the shifted Chebyshev polynomials T ¿n ( x ' ) .
Verify that this produces the shifed polynomials shown in table 18.5 and that
Table 18.5 Shifed Type Chebyshev Polynomials
T ¿0=1
T ¿1=2 x −1
T ¿2=8 x 2−8 x+ 1
T ¿3=32 x 3 −48 x 2+32 x +1
T ¿4=128 x 4 −256 x3 +160 x 2−32 x+1
T ¿5=512 x 5−1280 x 4 +120 x 3−400 x 2 +50 x−1
T ¿6=2048 x 6−6144 x 5 +69124 −3584 x 3+ 840 x 2−72 x+ 1

They satisfy the orthonomality condition

1
−1 /2 δ mn π
∫ T ¿n ( x' ) T m ( x ' ) [ x ( 1−x ) ] dx=
2−δ π 0
.
0

Answer:
(a) The equations forT n and T m can be written
 1

[( 1−x ) T ( x ) ]= −n
2 2 '
n
T ( x)
( 1−x )
n
2 1 /2

[( 1−x ) T ( x ) ]= −m
2 2 '
m
T ( x)
( 1−x )
m
2 1 /2

Multiplying the first of these equations by m 2 T m (x ) and the second by n2 T n( x), then
subtracting the first from the second and integrating from −1 ¿ 1, we reach
1 1 ' 1 1 '

n 2
∫ T n ( x ) [ ( 1−x ) 2 2 '
]
T ( x ) dx−m
m
2
∫ T m ( x ) [( 1−x )
2 2 '
]
T ( x ) dx =0
n
−1 −1

Note that the right-hand-side integrals are convergent and their cancellation is there
legitimate.
Now integrate the integrate the integrals in the above equation by parts, integrating the
explicit derivatives and differetiating the order Chebyshev polynomials. The presence of the
1 /2
factor ( 1−x 2 ) causes all the endpoint terms to cancel, and we are left with
1
1 /2
(m −n ) ∫ T 'm ( x ) T 'n ( x ) ( 1−x 2 ) dx=0
2 2

−1

The left-hand side of this equation is zero whether or not m=n, but only if m≠ n does it show
that the integral vanishes.

(b) Differentiating Eq.(18.105), we get, noting that x=cosθand comparing the result with Eq.
(18.107)
d T n ( x ) d T n θ dθ
= =¿
dx dθ dx
18.4.14The expansion of a power of x in a Chebyshev series leads to the integral
1
dx
I mn=∫ x m T n ( x ) .
−1 √1−x 2
(a) Show that this integral vanishes for m<n.
(b) Show that this integral vanishes m+n odd.
Answer:
Note that all instances of thix and dx in the orthogonality integral should have been primed, i. e .,
written x ' or d x '
The shift compresses the original T n into half the original rannge; this would decrease the
orthogonality integral by a factor of 2. However ,
 1 1
→ ,
2 1/ 2
( 1−x ) 2 √ x (1− x)

So the weight factor given in the exercise is twice that of the original T n . These factors of 2
cancel, so the orthogonality condition given for the shifted T nhas the same value as that given for
the T n .

18.4.15Evaluate the integral

1
dx
I mn=∫ x m T n ( x )
−1 √1−x 2
For m ≥n and m+n even by each of two methode
(a) Replacing T n ( x) by it’s Rodrigues representation.
(b) Using x=cos θ to transform th eintegral to a form with θ as variable.
m ! ( m−n−1 ) ‼
ANS I mn=π , m≥ n , m+n even.
( m−n ) ! ( m+ n ) ‼

18.4.16Establish the following bounds, −1 ≤ x ≤1 ;

(a) |U n ( x )|≤n+1 ,

(b) |dxd T (x)|≤ n .

n
2

Answer:
(a) Introducing the Rodrigues formula for T n
1 1
(−1 )n d n n+
I mn=
1
∫ xm ( ) dx
( 1−x 2 ) 2 dx
( )
2n Γ n+
2
−1

1
Integrating by parts n π 2 times, and noting that the endpoint integrated terms vanish, we

reach
1
1
π 2 m!
m! n+1 /2
I mn= ∫ x m−n ( 1−x 2) dx
1 (m−n) −1
n
2 Γ n+
2 ( )
1
2
π m! m! m−n+1 1
¿
1 ( m−n ) !
B
2
, n+
2 ( )
( )
2n Γ n+
2
 Here B is a beta function; the integral was evaluated using Eq.(13.50) Inserting the value
of B in terms of gamma function, we reach

I mn=
π m!
1
2
Γ ( m−n+1
2 ),
n
2 ( m−n ) !
( m−n
2 )
!

Which when written in terms of double factorials can seen equivalent to the answer in the
text.
(b) From Eq.(18.105), write I mn in terms of θ as
π
I mn=∫ cos θ cos nθ dθ
0

This integral can be evaluated by writting it entirely in terms of complex exponentials,

expanding the mth power by the binomial theorem, and noting that nearly all the resulting
integrals cancel. Alternatively, it can be recognized as a case of formula 3.631(9) in
Gradshteyn & Ryzhik, table of Integrals, Series, and Products, 6th ed.

18.4.17 (a) Show that for −1 ≤ x ≤1 ,|V n ( x )|≤ 1.

(b) Show that W n ( x ) is unbounded in −1 ≤ x ≤1.
Answer:
(a) Use the trigonometric form of U n given in Eq. (18.107), and expand sin ( n+1 ) θ :
sin ( n+1 ) θ sin nθ sin nθ
|U n|= | sin θ
=||
sinθ || |
+cos nθ ≤
sinθ
+1=|U n−1|+1

Applying this result successively to U n−1 ,U n−2 , … , we conclude that |U n| ≤|U 0|+n=n+1,
as required
(b) From the result of exercise 18.4.13(b), we know that

|dtd T ( x)|=n|U
n n−1 |
2
Applting the result from part (a), n|U n−1|≤ n , as given in the text.
18.4.18Verify the orthogonality-normalization integrals for
(a) T m ( x ) , T n ( x ) ,
(b) V m ( x ) , V n ( x ) ,
(c) U m ( x ) ,U n ( x ) ,
 (d) W m ( x ) ,W n ( x ) ,
Hint. All these can be converted to trigonometric integrals.
Answer:
(a) From Eqs. (18.107) and (18.109), write V n=sin nθ . This form is clearly bounded by ± 1
(b) Equation (18.110) shows that W n becomes infinite at x=± 1, so it is clearly unbounded on
the specified range

18.4.19Show whether
(a) T m ( x ) and V n ( x ) are ot are not orthogonal over the interval [-1,1] with respect to
1
the weighting factor ( 1−x 2 ) 2

(b) U m ( x ) and W n ( x ) are ot are not orthogonal over the interval [-1,1] with respect to
1
the weighting factor ( 1−x 2 ) 2

18.4.20Derive
(a) T n+1 ( x ) +T n−1 ( x )=2 xTn(x )
(b) T m+n ( x )+ T m−n ( x )=2 T m ( x ) T n (x), from the “cprresponding” cosine identities.
Answer:
(a) Using the techniques in parts (a) and (b) of Exercise !8.4.19, we get
1 π
1/ 2
∫ U m ( x ) U n ( x ) ( 1−x 2) dx=∫ cos mθ sin nθ dθ
−1 0

Which for many values of mm≠ n is nonzero. (The interval (0,π) is nor an interval of
orthogonality for this function set).
(b) Orthogonality is not obtained for all U m and W n . Set the solution to part (a)

18.4.21A number of equation relate the two types of Chebyshev Polynomials. As examples show
that
T n ( x )=U n ( x )−x U n−1 (x)
And
( 1−x 2 ) U n ( x )=x T n+1 ( x ) −T n +2 ( x ) .
Answer:
 (a) Using Eq.(18.105), this recurrence formula is equivalent to
cos ( n+ 1 ) θ+ cos(n−1)θ=2 cosθ cos nθ
Writing
cos (n ±1) θ=cos θ cos nθ ∓sin θ sin nθ
The verification is immediate.
(b) Since the equation of this part can be written
cos ( m−n ) θ +cos( m−n)θ=2 cos mθ cos nθ
It can be confiermed by the approach of part (a) using the formulas for cos (m± n)θ

18.4.22Show that
d V n( x ) T (x)
=−n n 2
dx √1−x
(a) Using the trigonometric forms of V n and T n ,
(b) Using the Rodrigues representation.
Answer:
Equation (18.91), the generating function for the T n, can also be used to develop a formulas
involving the U n . In particular,
∞
1−t 2
2
=T 0 +2 ∑ T n ( x ) t n ,
1−2 xt+t n=1

∞
(1−t 2 ) ∑ U n ( x ) t n
n=0

18.4.23Starting with x=cosθ and T n cosθ=cosnθ , expand

k
eiθ +e−iθ
k
x= (2 )
And show that
1
T k ( x )+ k T k−2 ( x ) + k T k−4 +…
xk=
2 −1k [ 1 () () 2 ]
The series in brackets terminating after the term containing T 1∨T 0 .
Answer:
(a) Differentiate the trigonometric form for V n , Eq. (18.106). The result is
d V n( x ) d V n (θ) −1 T (x)
dx
=
dθ
=( ncos nθ )( )
sin θ
=−n n 2
√ 1−x
 (b) Using Eq.(18.109), we write the Rodrigues formula for V n by multiplying that for U n−1Eq.
1 /2
(18.103) by ( 1−x 2 ) . Thus
n−1 1
n π 1 /2 d n−1
V n=
(−1 )
1 dx n −1
n−
( 1−x 2) 2 [ ]
2n Γ n+( )
2

18.4.24Develop the following Chebyshev expansions (for [-1,1]):

2 1 /2 2
(a) ( 1−x ) = ¿
π
∞
+1 ,0 < x ≤ 1 = 4
(b)
−1 ,−1 ≤ x <0 π s=0
∑ s
} −1
(−1 ) ( 2 s +1 ) T 2 s +1 ( x )

Answer:
Making the binomial expansion of x k , and changing the sign, we identify the result as n times the
Rodrigus formula for T n
k k
eiθ +e−iθ 1
( = k ∑ k einθ e−i(k−n)θ
) ()
k
x=
2 2 n=0 n
1 k
1
k ∑ () k [¿ e i(2 n−k)θ +e i(k−2 n)θ ]+ k
2 0 ≤n ≤ k/ 2 n
2 k /2
0, {(
, k even ,

k odd ,
¿ )
The exponenyials in the sum combine to form 2 cos ( k−2 n ) θ=2 T k −2 n ( x ) , tielding final result
similar to that shown in the text. However, the text did not make clear taht when k is even, the
final tern of the expansion

18.4.25 (a) for the interval [-1,1] show that

∞
1 s +1 ( 2 s−3 ) ‼
|x|= +∑ (−1 ) ( 4 s +1 ) P2 s ( x )
2 s=1 ( 2 s +2 ) ‼
∞
2 4 s +1 1
¿ + ∑ (−1 ) 2
T 2 s ( x ).
π π s =1 4 s −1
(b) Show that ratio of the coefficient of T 2 s (x) to that of P2 s ( x ) approaches ( πs ¿−1 as
s → ∞ . This ilustrates the relatively rapid convergence of the Chebyshev series.
Hint. With the Legendre recurrence relations, rewrite x P n ( x ) as a linear combination
of derivatives. The trigonometric substitution x=cosθ ,T n ( x ) =cos nθ is most helpful
for the Chebyshev part.
Answer:
(a) Here we need the Chebyshev expansion of sin θ . Letting c l be the coefficient og T l in the
expancsion, we have
π π
1 cos θ 2
c 0= ∫ sin θdθ=−
π 0
=
π 0 π |
(b) This sine function is ± 1 for l=1,5 , …∧−1 for l=3,7 ,… ; changing the index from l to
2 s +1 we can write
s
4 (−1 )
c 2 s +1=
π 2 s +1
18.4.26Show that
∞
π2 −2
=1+2 ∑ ( 4 s 2−1 )
8 s=1

Hint. Apply Parseval’s identity (or the completeness relation) to the results of Exercise
18.4.26.
Answer:

(a) The lagrange expansion of |x| is of the form |z|=∑ c 2 s P2 s ( x ) , where

s

1
1
c 0=∫ xdx =
0 2
And for nonzero s,
1 1
c 2 s=(4 s +1) ∫ x P2 s ( x ) dx =∫ ¿ ¿¿
0 0

1
¿∫ ¿ ¿
0

The final result in the llimit of large s is therefore

Ratio ≈ ( πs )−1 ( πs )1 /2=( πs )−1/ 2

18.4.27Show that
∞
π 4
−1 1
(a) cos x= − ∑ T ( x)
2 π n=0 ( 2 n+1 )2 2 n+1
∞
−1 4 1
(b) sin x= ∑ T (x)
π n=0 ( 2n+ 1 )2 2 n+1
Answer:
From
1 −1 π
2 π
∫|x| ( 1−x 2 ) 2 dx=∫ cos θdθ= , 2
−1 0

And the also calculate the same quantity using the expansion of |x| in Chebyshev polynomials
from Exercise 18.4.26. we have
1 −1 1 ∞ 2 −1
π
2 −1
2

−1
2 4
=∫ |x| ( 1−x 2) 2 dx =∫ + ∑ (−1 )
π π n=1
s +1
[ 1
2
4 s −1
T 2 s ( x ) ( 1−x 2 ) 2 dx
]
Using the orthogonality of the T n and the values of their normalization integrals, the above
equation reduces to
∞
π 4 16 1 π
= ( π )+ 2 ∑
2 π2 2
π n=1 ( 4 s −1 ) 2
2

π
When this equation is multiplied through by we get the given in the text
4

18.4.28show that
∞
−1 π 4 1
(a) cos x= − ∑ T (x)
2 π n=0 ( 2 n+1 )2 2 n+1
∞
−1 4 1
(b) sin x= ∑ T (x)
π n=0 ( 2n+ 1 )2 2 n+1
Answer:
(a) Taking x=cos θ , this equation is seen equivalent to
∞
π 4 1
θ= − 2 ∑ T ¿¿
2 π n=0 ( 2 n+1 )2 2 n+1
Making a change of the index variable from l to 2 n+1 , the expansion becomes
π
θ= T 0 ¿
2
Equivalent to the answer in the text
1 −1 π
(b) We note that sin x=sin (cosθ)= −θ . This observation leads immediatedly to the
2
expansion in the text.