Orthogonality of Legendre Polynomials
📂Functions Orthogonality of Legendre Polynomials Theorem In the interval [ − 1 , 1 ] [-1,\ 1] [ − 1 , 1 ] , Legendre polynomials form an orthogonal set .
∫ − 1 1 P l ( x ) P m ( x ) d x = 2 2 l + 1 δ l m ( l , m = 0 , 1 , 2 , … )
\int_{-1}^{1} P_{l}(x)P_{m}(x) dx =\frac{2}{2l+1}\delta_{lm} \quad (l, m = 0, 1, 2, \dots)
∫ − 1 1 P l ( x ) P m ( x ) d x = 2 l + 1 2 δ l m ( l , m = 0 , 1 , 2 , … )
Proof Case 1: l ≠ m l \ne m l = m Legendre Differential Equation
The following differential equation is called the Legendre differential equation .
d d x [ ( 1 − x ) 2 d y d x ] + l ( l + 1 ) y = 0
\dfrac{d}{d x}\left[ (1-x)^{2} \dfrac{d y}{d x} \right] +l(l+1)y = 0
d x d [ ( 1 − x ) 2 d x d y ] + l ( l + 1 ) y = 0
Since Legendre polynomials are solutions to the Legendre differential equation, they satisfy the above equation. Substituting P l P_{l} P l and P m P_{m} P m into the equation, we get,
d d x [ ( 1 − x 2 ) P l ′ ( x ) ] + l ( l + 1 ) P l ( x ) = 0 d d x [ ( 1 − x 2 ) P m ′ ( x ) ] + m ( m + 1 ) P m ( x ) = 0
\begin{align*}
\dfrac{d}{dx} \left[ (1-x^2)P^{\prime}_{l}(x) \right] + l(l+1)P_{l}(x) &= 0 \\
\dfrac{d}{dx} \left[ (1-x^2)P^{\prime}_{m}(x) \right] + m(m+1)P_{m}(x) &= 0
\end{align*}
d x d [ ( 1 − x 2 ) P l ′ ( x ) ] + l ( l + 1 ) P l ( x ) d x d [ ( 1 − x 2 ) P m ′ ( x ) ] + m ( m + 1 ) P m ( x ) = 0 = 0
Multiplying both equations by P m ( x ) P_{m}(x) P m ( x ) and P l ( x ) P_{l}(x) P l ( x ) respectively and subtracting, we obtain the following.
P m d d x [ ( 1 − x 2 ) P l ′ ] − P l d d x [ ( 1 − x 2 ) P m ′ ] + [ l ( l + 1 ) − m ( m + 1 ) ] P l P m = 0
\begin{equation}
P_{m}\dfrac{d}{dx}\left[ (1-x^2)P^{\prime}_{l} \right] - P_{l}\dfrac{d}{dx}[(1-x^2)P^{\prime}_{m}] + [l(l+1)-m(m+1)]P_{l}P_{m} = 0
\end{equation}
P m d x d [ ( 1 − x 2 ) P l ′ ] − P l d x d [( 1 − x 2 ) P m ′ ] + [ l ( l + 1 ) − m ( m + 1 )] P l P m = 0
Meanwhile, the following equation holds.
d d x [ ( 1 − x 2 ) ( P m P l ′ − P l P m ′ ) ] = d d x [ ( 1 − x 2 ) P l ′ P m ] − d d x [ ( 1 − x 2 ) P m ′ P l ]
\begin{align*}
& \dfrac{d}{dx}[(1-x^2)(P_{m}P^{\prime}_{l}-P_{l}P^{\prime}_{m})] \\
&= \dfrac{d}{dx}[{\color{blue}(1-x^2)P^{\prime}_{l}}P_{m}] -\dfrac{d}{dx}[{\color{blue}(1-x^2)P^{\prime}_{m}}P_{l}]
\end{align*}
d x d [( 1 − x 2 ) ( P m P l ′ − P l P m ′ )] = d x d [ ( 1 − x 2 ) P l ′ P m ] − d x d [ ( 1 − x 2 ) P m ′ P l ]
Thinking of the blue-painted part as one function and expanding the equation with the product rule gives,
d d x [ ( 1 − x 2 ) P l ′ ] P m + ( 1 − x 2 ) P l ′ P m ′ − d d x [ ( 1 − x 2 ) P m ′ ] P l − ( 1 − x 2 ) P m ′ P l ′ = d d x [ ( 1 − x 2 ) P l ′ ] P m − d d x [ ( 1 − x 2 ) P m ′ ] P l
\begin{align*}
& \dfrac{d}{dx}[(1-x^2)P^{\prime}_{l}]P_{m}+(1-x^2)P^{\prime}_{l}P^{\prime}_{m}-\dfrac{d}{dx}[(1-x^2)P^{\prime}_{m}]P_{l}-(1-x^2)P^{\prime}_{m}P^{\prime}_{l} \\
&= \dfrac{d}{dx}[(1-x^2)P^{\prime}_{l}]P_{m}-\dfrac{d}{dx}[(1-x^2)P^{\prime}_{m}]P_{l}
\end{align*}
d x d [( 1 − x 2 ) P l ′ ] P m + ( 1 − x 2 ) P l ′ P m ′ − d x d [( 1 − x 2 ) P m ′ ] P l − ( 1 − x 2 ) P m ′ P l ′ = d x d [( 1 − x 2 ) P l ′ ] P m − d x d [( 1 − x 2 ) P m ′ ] P l
This is the same as the first two terms of ( 1 ) (1) ( 1 ) . Thus, ( 1 ) (1) ( 1 ) can be organized as follows.
d d x [ ( 1 − x 2 ) ( P m P l ′ − P l P m ′ ) ] + [ l ( l + 1 ) − m ( m + 1 ) ] P l P m = 0
\dfrac{d}{dx}[(1-x^2)(P_{m}P^{\prime}_{l}-P_{l}P^{\prime}_{m})]+ [l(l+1)-m(m+1)]P_{l}P_{m}=0
d x d [( 1 − x 2 ) ( P m P l ′ − P l P m ′ )] + [ l ( l + 1 ) − m ( m + 1 )] P l P m = 0
Integrating both sides over the interval [ − 1 , 1 ] [-1, 1] [ − 1 , 1 ] yields,
( 1 − x 2 ) ( P m P l ′ − P l P m ′ ) ∣ − 1 1 + [ l ( l + 1 ) − m ( m + 1 ) ] ∫ − 1 1 P l ( x ) P m ( x ) d x = 0
(1-x^2)(P_{m}P^{\prime}_{l}-P_{l}P^{\prime}_{m})\Big|_{-1}^{1} +[l(l+1)-m(m+1)]\int_{-1}^{1}P_{l}(x)P_{m}(x)dx=0
( 1 − x 2 ) ( P m P l ′ − P l P m ′ ) − 1 1 + [ l ( l + 1 ) − m ( m + 1 )] ∫ − 1 1 P l ( x ) P m ( x ) d x = 0
The first term is ( 1 − x 2 ) ∣ x = ± 1 = 0 (1-x^2)\Big|_{x = \pm 1}=0 ( 1 − x 2 ) x = ± 1 = 0 , so it becomes 0 0 0 . Due to the condition of l , m l, m l , m , the constant in front of the integration in the second term absolutely cannot be 0 0 0 . Therefore, we get the following.
∫ − 1 1 P l ( x ) P m ( x ) d x = 0
\int_{-1}^{1}P_{l}(x)P_{m}(x)dx=0
∫ − 1 1 P l ( x ) P m ( x ) d x = 0
■
Case 2: l = m l = m l = m Recurrence relation of Legendre polynomials
l P l ( x ) = x P l ′ ( x ) − P l − 1 ′ ( x )
lP_{l}(x) = xP^{\prime}_{l}(x) - P^{\prime}_{l-1}(x)
l P l ( x ) = x P l ′ ( x ) − P l − 1 ′ ( x )
Multiplying both sides of the above equation by P l ( x ) P_{l}(x) P l ( x ) and integrating gives,
l ∫ − 1 1 [ P l ( x ) ] 2 d x = ∫ − 1 1 x P l ( x ) P l ′ ( x ) d x − ∫ − 1 1 P l ( x ) P l − 1 ′ ( x ) d x .
\begin{equation}
l\int_{-1}^{1}[P_{l}(x)]^{2} dx= \int_{-1}^{1}xP_{l}(x)P^{\prime}_{l}(x)dx -\int_{-1}^{1} P_{l}(x)P^{\prime}_{l-1}(x)dx.
\end{equation}
l ∫ − 1 1 [ P l ( x ) ] 2 d x = ∫ − 1 1 x P l ( x ) P l ′ ( x ) d x − ∫ − 1 1 P l ( x ) P l − 1 ′ ( x ) d x .
Here, P l − 1 ′ ( x ) P^{\prime}_{l-1}(x) P l − 1 ′ ( x ) is a polynomial of degree l − 2 l-2 l − 2 , and since Legendre polynomials are orthogonal to polynomials of lower degree , the last term on the right-hand side is 0 0 0 . The first term on the right-hand side can be solved with integration by parts .
∫ − 1 1 x P l ( x ) P l ′ ( x ) d x = ∫ − 1 1 x 2 [ 2 P l ( x ) P l ′ ( x ) ] d x = x 2 [ P l ( x ) ] 2 ∣ − 1 1 − 1 2 ∫ − 1 1 [ P l ( x ) ] 2 d x = 1 − 1 2 ∫ − 1 1 [ P l ( x ) ] 2 d x .
\begin{align}
\int_{-1}^{1}xP_{l}(x)P^{\prime}_{l}(x)dx &= \int_{-1}^{1}\frac{ x}{2}[2P_{l}(x)P^{\prime}_{l}(x)]dx \nonumber \\
&= \frac{ x}{2}[P_{l}(x)]^{2}\bigg|_{-1}^{1}-\frac{1}{2}\int_{-1}^{1}[P_{l}(x)]^{2}dx \nonumber \\
&= 1-\frac{1}{2}\int_{-1}^{1}[P_{l}(x)]^{2}dx.
\end{align}
∫ − 1 1 x P l ( x ) P l ′ ( x ) d x = ∫ − 1 1 2 x [ 2 P l ( x ) P l ′ ( x )] d x = 2 x [ P l ( x ) ] 2 − 1 1 − 2 1 ∫ − 1 1 [ P l ( x ) ] 2 d x = 1 − 2 1 ∫ − 1 1 [ P l ( x ) ] 2 d x .
The third equation is valid due to P l ( 1 ) = 1 P_{l}(1)=1 P l ( 1 ) = 1 . Therefore, substituting ( 3 ) (3) ( 3 ) into ( 2 ) (2) ( 2 ) gives,
l ∫ − 1 1 [ P l ( x ) ] 2 d x = 1 − 1 2 ∫ − 1 1 [ P l ( x ) ] 2 d x ⟹ ∫ − 1 1 [ P l ( x ) ] 2 d x = 2 2 l + 1
\begin{align*}
&& l\int_{-1}^{1}[P_{l}(x)]^{2} dx &= 1-\frac{1}{2}\int_{-1}^{1}[P_{l}(x)]^{2}dx \\
\implies && \int_{-1}^{1}[P_{l}(x)]^{2} dx &= \frac{2}{2l+1}
\end{align*}
⟹ l ∫ − 1 1 [ P l ( x ) ] 2 d x ∫ − 1 1 [ P l ( x ) ] 2 d x = 1 − 2 1 ∫ − 1 1 [ P l ( x ) ] 2 d x = 2 l + 1 2
■