Leibniz integral rule with variable limits

3 replies · 2K views · last reply
terryphi
Messages
57
Reaction score
0
I'm going through the book "Elementry Differnetial Equations With Boundary Value Problems" 4th Eddition by William R. Derrick and Stanley I. Grossman.

On Page 138 (below)
Printed textbook page titled Green's functions showing integral forms of a particular solution to a nonhomogeneous linear equation, with equations for y_p(x), its derivative y_p'(x), and the Green's function solution y(x).
)

The authors take the derivative of a definite integral and end up with a definite integral plus another term. How did they do this?
 
Use Leibniz integral rule
$$\frac{\mathrm{d}}{\mathrm{d}x}\int_{a(x)}^{b(x)} \! f(x,t) \, \mathrm{d}x=\frac{\mathrm{d}b(x)}{\mathrm{d}x}f(x,b(x))-\frac{\mathrm{d}a(x)}{\mathrm{d}x}f(x,a(x))+\int_{a(x)}^{b(x)} \! \frac{\mathrm{\partial}}{\mathrm{\partial}x} f(x,t) \, \mathrm{d}x$$
https://en.wikipedia.org/wiki/Leibniz_integral_rule
 
It is because they differentiate with respect to ##x## and there are at least three occurrences of ##x## in the formula that is being integrated: one as an upper integration limit, and two as function arguments in the integrand. There may be up to two more, being the ##y_1,y_2## in the denominator of the integrand. The excerpt does not make clear whether they also depend on ##x##, so I will assume they do not, although they could, since there is a ##y_1(x)## and ##y_2(x)## in the numerator.

Let the function ##A:\mathbb R^2\to\mathbb R## be an antiderivative (indefinite integral) of the integrand with respect to ##t##, ie ##\frac{\partial }{\partial t}A(t,x)=a(t,x)## where ##a(t,x)## is the integrand.

Then the definite integral is
$$A(x,x)-A(x_0,x)$$
Differentiating this wrt ##x## and using the total differential rule gives
\begin{align*}
\frac d{dx}\left(A(x,x)-A(x_0,x)\right)
&=D_1A(x,x)\frac{dx}{dx} + D_2A(x,x)\frac{dx}{dx}
-D_1A(x_0,x)\frac{dx_0}{dx}-D_2A(x_0,x)\frac{dx}{dx}\\
&=D_1A(x,x)\cdot 1 + D_2A(x,x)\cdot 1
-D_1A(x_0,x)\cdot 0-D_2A(x_0,x)\cdot 1\\
&=D_1A(x,x) + D_2A(x,x)-D_2A(x_0,x)
\end{align*}
where ##D_kA## indicates the partial derivative of function ##A## wrt its ##k##th argument.
This is equal to
\begin{align*}
a(x,x) + D_u\bigg(A(x,u)-A(x_0,u)\bigg)\bigg|_{u=x}
&=
a(x,x) + D_u \int_{x_0}^x a(t,u)dt\bigg|_{u=x}
\end{align*}
where ##D_u## indicates partial differentiation wrt the variable ##u##
In the second term we can, given certain mild assumptions, move the differentiation operator inside the integral sign to obtain:
\begin{align*}
a(x,x) + \int_{x_0}^x D_ua(t,u)dt\bigg|_{u=x}
=a(x,x) + \int_{x_0}^x D_xa(t,x)dt
\end{align*}

EDIT: Which is a derivation of the Leibniz integration rule referenced in the above post, which appeared on my screen when I posted this.
 
Last edited:
andrewkirk said:
The excerpt does not make clear whether they also depend on ##x##, so I will assume they do not, although they could, since there is a ##y_1(x)## and ##y_2(x)## in the numerator.
The y's in the denominator of the integral do not depend on x. After differentiating The t's in the denominator turn to x's.
 
  • Like
Likes   Reactions: terryphi