Professional Documents
Culture Documents
1. Introduction
The Legendre, Laguerre, and Hermite equations have many real
world practical uses which we will not discuss here. We will only focus
on the methods of solution and use in a mathematical sense. In solving
these equations explicit solutions cannot be found. That is solutions
in in terms of elementary functions cannot be found. In many cases it
is easier to find a numerical or series solution.
There is a generalized Fourier series theory which allows one to write
a function f (x) as a linear combination of an orthogonal system of
functions φ1 (x),φ2 (x),...,φn (x),... on [a, b]. The series produced is called
the Fourier series with respect to the orthogonal system.R While the
b
f (x)φn (x)dx
coefficients ,which can be determined by the formula cn = a
Rb
2
,
a φn (x)dx
are called the Fourier coefficients with respect to the orthogonal system.
We are concerned only with showing that the Legendre, Laguerre, and
Hermite polynomial solutions are orthogonal and can thus be used to
form a Fourier series. In order to proceed we must define an inner
product and define what it means for a linear operator to be self-
adjoint.
Rb
Definition 1. We define an inner product hy1 |y2 i = a y1 (x)y2 (x)dx
where y1 , y2 ∈ C 2 [a, b] and two functions aresaid to be orthogonal if
Rb
(y1 |y2 ) = a y1 (x)y2 (x) = 0.
Definition 2. A linear operator L is self-adjoint if hLy1 |y2 i = hy1 |Ly2 i
for all y1 ,y2 .
1
T
2 HOMAS COVERSON, SAVARNIK DIXIT, ALYSHA HARBOUR, AND TYLER OTTO
Ly = −λry
h−λry|yi = hy| − λryi
−hλry|yi = −hy|λryi
hλry|yi = hy|λryi
λhry|yi = λhy|ryi
Z b Z b
λ yyr(x)dx = λ yyr(x)dx
a a
Z b Z b
2
λ |y(x)| r(x)dx = λ |y(x)|2 r(x)dx
a a
λ=λ
∴λ∈R
With this equality, we have a new inner product called the weighted
inner product
Z b
(2.4) hy1 |y2 ir = y1 (x)y2 (x)r(x)dx
a
Therefore y1 and y2 are orthogonal.
∞
X ∞
X ∞
X
(1 − x2 ) ak k(k − 1)xk−2 − 2x ak kxk−1 + n(n + 1) ak xk = 0.
k=2 k=1 k=0
which gives
∞
X ∞
X ∞
X
k−2 k
ak k(k − 1)x − ak k(k − 1)x − 2 ak kxk
k=2 k=2 k=1
∞
X
+ n(n + 1) ak x k = 0
k=0
∞
X ∞
X ∞
X
ak+2 (k + 2)(k + 1)xk − ak k(k − 1)xk − 2 ak (k)xk
k=0 k=0 k=0
∞
X
+ n(n + 1)ak xk = 0
k=0
simplify
∞
X
[(k + 2)(k + 1)ak+2 − (k)(k − 1)ak − 2kak + n(n + 1)ak ]xk = 0
k=0
equating coefficients
k(k + 1) − n(n + 1)
(3.2) ak+2 = ak .
(k + 2)(k + 1)
T
6 HOMAS COVERSON, SAVARNIK DIXIT, ALYSHA HARBOUR, AND TYLER OTTO
L
X
ak xk = a0 + a1 x + a2 x2 + · · · + aL + 0xL+1 + 0xL+2 + · · ·
k=0
L(L + 1) − n(n − 1)
aL+2 = aL =0
(L + 2)(L + 1)
L(L + 1) − n(n − 1)
=0
(L + 2)(L + 1)
L(L + 1) − n(n + 1) = 0
L(L + 1) = n(n + 1)
L = n or L = −(n + 1)
So L = n is our solution because all terms after n+1 are zero. Therefore
the degree of our polynomial solution is n where n is an integer.
Where both solutions are valid for x ∈ (−1, 1). Finding the Le-
gendre polynomials can be very long and difficult. There are many
methods including Rodrigue’s Formula that are useful in finding these
ORTHOGONAL FUNCTIONS: THE LEGENDRE, LAGUERRE, AND HERMITE POLYNOMIALS
7
P0 (x) = 1
P1 (x) = x
1
P2 (x) = (3x2 − 1)
2
1
P3 (x) = x(5x2 − 3)
2
1
P4 (x) = (35x4 − 30x2 + 3)
8
·
·
·
∞
X
r
(4.3) y1 (x) = x ak xk , a0 6= 0
k=0
(4.4) r(r − 1) + p0 r + q0 .
We compare the Laguerre equation(4.1) to our standard form equation
(4.2). We multiply (4.1) by x to obtain the equation
ORTHOGONAL FUNCTIONS: THE LEGENDRE, LAGUERRE, AND HERMITE POLYNOMIALS
9
We see that
r(r − 1) + r = 0
or
r2 = 0.
Thus the roots of the indcial equation are r1 = r2 = 0. Both roots are
equal so we will have a second linearly independent equation, which we
will not use, of the the form
∞
X
r1
(4.6) y2 (x) = y1 (x) ln x + x bk x k .
k=0
We will now find the recurrence realtion for the coefficients for y1 (x)
by direct substitution of y1 (x) into equation (4.1).
∞
X ∞
X
r k
y(x) = x ak x = ak x(k+r)
k=0 k=0
∞
X
y 0 (x) = (k + r)ak x(k+r−1)
k=1
X∞
y 00 (x) = (k + r)(k + r − 1)ak x(k+r−2)
k=2
∞
X ∞
X
(k+r−2)
x (k + r)(k + r − 1)ak x + (1 − x) (k + r)ak x(k+r−1)
k=2 k=1
∞
X
+n ak x(k+r) = 0
k=0
which gives us
∞
X ∞
X
(k + r)(k + r − 1)ak x(k+r−1) + (k + r)ak x(k+r−1)
k=2 k=1
∞
X
− (k + r)ak x(k+r)
k=1
∞
X
+ nak x(k+r) = 0
k=0
∞
X ∞
X
(k+r)
(k + r + 1)(k + r)ak+1 x + (k + r + 1)ak+1 x(k+r)
k=1 k=0
X∞
− (k + r)ak x(k+r)
k=1
X∞
+ nak x(k+r) = 0
k=0
∞
X ∞
X
(k + 1)(k)ak+1 xk + (k + 1)ak+1 xk
k=1 k=0
∞
X
− kak xk
k=1
∞
X
+ nak xk = 0.
k=0
for the terms with k = 1 we see that the k = 0 terms add nothing so
∞
X
[(k + 1)(k)ak+1 + (k + 1)ak+1 − kak + nak ]xk = 0.
k=0
a1 = −na0
(1 − n)a1 n(n − 1)a0
a2 = =
4 22
(2 − n)a2 −n(n − 1)(n − 2)a0
a3 = =
9 22 · 32
(3 − n)a3 n(n − 1)(n − 2)(n − 3)a0
a4 = =
16 22 · 32 · 42
.
.
.
[(k − 1) − n]ak−1 (−1)k n(n − 1)(n − 2)...(n − k + 1)
ak = =
k2 22 · 32 · 42 · · · k 2
k
(−1) n!a0
=
(k!)2 (n − k)!
Which is
∞
X (−1)k n!a0 k
(4.8) yn (x) = x , k = 0, 1, 2, ...
k=0
(k!)2 (n − k)!
When n = 0, 1, 2, 3, ... the series (4.7) ends. That is the terms after the
nth term are zero. If we take a0 to be k! then we get polynomials. We
call these polynomials Laguerre polyomials. The Laguerre Polynomials
ORTHOGONAL FUNCTIONS: THE LEGENDRE, LAGUERRE, AND HERMITE POLYNOMIALS
13
are
(4.9) L0 (x) = 1
L1 (x) = −x + 1
L2 (x) = x2 − 4x + 2
L3 (x) = −x3 + 9x2 − 18x + 6
L4 (x) = x4 − 16x3 + 72x2 − 96x + 24
·
·
·
n
X (−1)k (n!)2 k
Ln (x) = 2 (n − k)!
x
k=0
(k!)
Therefore
r(x) = ce−x .
Equation (4.10) now becomes
where
(4.16) L = (e−x xy 0 )0
5. Hermite Polynomials
The Hermite differential equation is
(5.1) y 00 − 2xy 0 + 2ny = 0n ∈ R, x ∈ (−∞, ∞)
We know that x=0 is an ordinary point of equation (5.1). We may use
the power series method to find the polynomial solutions.
∞
X
y= ak x k
k=0
X∞ ∞
X
0 k−1
y = ak kx = ak+1 (k + 1)xk
k=0 k=0
X∞ ∞
X
00 k−1
y = ak+1 (k + 1)x = ak+2 (k + 2)(k + 1)xk
k=0 k=0
By substitution,
∞
X ∞
X ∞
X
k k
ak+2 (k + 2)(k + 1)x − 2x ak+1 (k + 1)x + 2n ak x k = 0
k=0 k=0 k=0
X∞ ∞
X ∞
X
ak+2 (k + 2)(k + 1)xk − 2 ak+1 (k + 1)xk+1 + 2nak xk = 0
k=0 k=0 k=0
∞
X ∞
X ∞
X
ak+2 (k + 2)(k + 1)xk + 2nak xk − 2 ak kxk = 0
k=0 k=0 k=0
∞
X
[ak+2 (k + 2)(k + 1)xk + 2ak (n − k)]xk = 0
k=0
THOMAS
16 COVERSON, SAVARNIK DIXIT, ALYSHA HARBOUR, AND TYLER OTTO
−2(n − k)
ak+2 = ak
(k + 2)(k + 1)
L
X
ak xk = a0 + a1 x + a2 x2 + · · · + aL + 0xL+1 + 0xL+2 + · · ·
k=0
−2(n − L)
aL+2 = aL
(L + 2)(L + 1)
−2(n − L)
=0
(L + 2)(L + 1)
−2(n − L) = 0
n−L=0
n=L
So L = n is our solution because all terms after n+1 are zero. Therefore
the degree of our polynomial solution is n where n is an integer.So we
can rewrite a polynomial of degree L as
n
X
ak x k
k=0
P0 (x) = 1
P1 (x) = 2x
P2 (x) = 4x2 − 2
P3 (x) = 8x3 − 12x
P4 (x) = 16x4 − 48x2 + 12
·
·
·
So
Z ∞
hLf |gi − hf |Lgi = Lf (x)g(x) − f (x)Lg(x)dx = 0
−∞
2
For Hermite solutions, Ly = [e−x y 0 ]0 , so we want to place restrictions
on f and g so that
Z ∞
Lf (x)g(x) − f (x)Lg(x)dx = 0
−∞
THOMAS
18 COVERSON, SAVARNIK DIXIT, ALYSHA HARBOUR, AND TYLER OTTO
We want
2
lim e−x f (x)g 0 (x) = 0
x→±∞
and
2 /2
lim e−x h0 (x) = 0
x→±∞
6. Acknowledgements
We would like to thank Dr. Davidson and Matthew Dawson for their
patience and wisdom through this project.
References
[1] Tolstov, G.P., Fourier Series, Dover Publications, New York, 1976, pg. 1–65.
[2] Morris, T., Harry P., Ordinary Differential Equations, Dover Publications,
New York,1985, pg.531-630.
[3] Goode, W.S., Annin, S.A., Differential Equations and Linear Algebra Pearson
Education, Upper Saddle River, 2007, pg.671-722.
[4] Adkins, W.A., Davidson, M.G., Ordinary Differential Equations, Springer,
Berlin, 2009
ORTHOGONAL FUNCTIONS: THE LEGENDRE, LAGUERRE, AND HERMITE POLYNOMIALS
19