0
$\begingroup$

Hi guys i'm reading a paper in which the authors have two coupled integral equation for the function $f(x)$ and $g(x)$, in order to solve this problem they employ a power series expansion of these function in powers of a small parameter $\epsilon$ $f(x)=\epsilon f^{(1)}(x)+\epsilon^{2} f^{(2)}(x)+O(\epsilon^{3})$ and the same for $g(x)$. I don't understand how to proceed to the solution once you take this approximation, so if you could help me with a simplified problem to let me understand how to proceed it would be great...

Let's consider the following equation (which represent a simplified version of my problem)

$K = \frac{4}{\epsilon}f(x)+f(x)\ln(1-\frac{x^{2}}{l^{2}})+\int_{-l}^{l}(f(x)-f(s))(\frac{2}{(x-s)^{2}})ds$

Now if i take the approximation of the function as a power series and i truncate to second order for example, how do i proceed? (K and are real positive numbers)

$\endgroup$
1
$\begingroup$

By direct substitution we have $$\begin{array}{l}K = \frac{4}{\varepsilon }\left( {\varepsilon {f^{(1)}}(x) + {\varepsilon ^2}{f^{(2)}}(x) + O({\varepsilon ^3})} \right) + \\\left( {\varepsilon {f^{(1)}}(x) + {\varepsilon ^2}{f^{(2)}}(x) + O({\varepsilon ^3})} \right)\ln (1 - \frac{{{x^2}}}{{{l^2}}}) + \int\limits_{ - l}^{ + l} {\left( \begin{array}{l}\varepsilon {f^{(1)}}(x) + {\varepsilon ^2}{f^{(2)}}(x) + O({\varepsilon ^3}) - \\\varepsilon {f^{(1)}}(s) - {\varepsilon ^2}{f^{(2)}}(s) - O({\varepsilon ^3})\end{array} \right)\frac{2}{{{{(x - s)}^2}}}ds} \end{array}$$Now, separating the zero and first order terms in $\varepsilon$ yields $$\left\{ \begin{array}{l}K = 4{f^{(1)}}(x)\\0 = 4{f^{(2)}}(x) + {f^{(1)}}(x)\ln (1 - \frac{{{x^2}}}{{{l^2}}}) + \int\limits_{ - l}^{ + l} {\left( {{f^{(1)}}(x) - {f^{(1)}}(s)} \right)\frac{2}{{{{(x - s)}^2}}}ds} \end{array} \right.$$ which is easily solved to yield $$\left\{ \begin{array}{l}{f^{(1)}}(x) = \frac{K}{4}\\{f^{(2)}}(x) = - \frac{K}{{16}}\ln (1 - \frac{{{x^2}}}{{{l^2}}})\end{array} \right.$$Hope it helps ;)

$\endgroup$
1
$\begingroup$

What is being performed here is the so called ¨naive expansion¨. What you have to do is to substitute for $f(x)$ and collect all terms of equal order in $\epsilon$ and solve them separately.

For example, upon substitution, the only $\mathcal{O}(1)$-term on the right hand side is $4f^{(1)}(x)$. You have to set it equal to the $\mathcal{O}(1)$-component in $K$ and solve the equation. When you collect all the $\mathcal{O}(\epsilon)$-terms on both sides, you shall see that there is a $f^{(1)}(x)$-dependence in the equation that you acquire. Therefore, you have to have $f^{(1)}(x)$ at your disposal to solve for $f^{(2)}(x)$. In general you may need all the $f^{(n)}(x)$ with $n \leq m-1$ in order to solve for $f^{(m)}(x)$.

So, if you are only interested in the solution up to $\mathcal{O}(\epsilon^2)$, you can simply stop after $f^{(2)}(x)$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.