Weird solution to an integral equation

102 Views Asked by At

So according to this wiki link

https://en.m.wikipedia.org/wiki/Integral_equation

The solution to the Fredholm equation of the first type of the form

$$g(s)=s\int_{0}^{\infty}dtK(st)f(t)$$

For a given function $g$ and kernel $K$ is

$$f(t)=\sum_{n=0}^{\infty}{\frac{a_n}{M(n+1)}t^n}$$

Where the coefficients can be found in the expansion of $g$

$$g(s)=\sum_{n=0}^{\infty}a_n s^{-n}$$

and

$$M(n+1)=\int_{0}^{\infty}dtK(t)t^n$$

is the Mellin transform of $K$ if it exists.

Now the identity for $g(s) $ is a little weird. Why is the sum over the negative powers instead?

Also if it is correct (or corrected) please direct me to a proof or something. I can't for the life of me figure out how to derive this.

1

There are 1 best solutions below

1
On BEST ANSWER

Here's a basic solution which I think shows the essential transformative steps; it can be made more rigorous, though perhaps with some restrictions on $f(t)$, $g(s)$, and $K(st)$, by carefully validating operations such as the interchanges of integration and summation and the range of $s$ (e.g., we should perhaps stipulate $s > 0$); nevertheless:

Given the integral equation

$g(s) = s\displaystyle \int_0^\infty K(st) f(t) \; dt, \tag 0$

we set

$u = st; \tag 1$

holding $s$ fixed, we have

$du = sdt, \tag 2$

$t = \dfrac{u}{s} = us^{-1}; \tag 3$

then

$s \displaystyle \int_0^\infty K(st)f(t) \; dt = \int_0^\infty K(st) f(t) s \; dt = \int_0^\infty K(u) f(us^{-1}) \; du; \tag 4$

we expand $f(us^{-1})$ into a power series

$f(us^{-1}) = f(t) = \displaystyle \sum_0^\infty f_n t^n = \sum_0^\infty f_n (us^{-1})^n = \sum_0^n f_n u^n s^{-n}; \tag 5$

we find

$g(s) = s \displaystyle \int_0^\infty K(st)f(t) \; dt$ $= \displaystyle \int_0^\infty K(u) \left ( \sum_0^\infty f_n u^n s^{-n} \right )\; du = \sum_0^\infty f_n \left ( \int_0^\infty K(u) u^n \; du \right ) s^{-n}; \tag 6$

with

$g(s) = \displaystyle \sum_0^\infty a_n s^{-n}, \tag 7$

and

$M(n + 1) = \displaystyle \int_0^\infty K(u) u^n \; du; \tag 8$

we re-assemble (6) via (7) and (8):

$\displaystyle \sum_0^\infty a_n s^{-n} = \sum_0^\infty f_n M(n + 1) s^{-n}; \tag 9$

term-by-term comparison of these series yields

$f_n = \dfrac{a_n}{M(n + 1)}, \tag{10}$

whence

$f(t) = \displaystyle \sum_0^\infty \dfrac{a_n}{M(n + 1)} t^n, \tag{11}$

as per request. $OE\Delta$