0
$\begingroup$

Recently, I have asked a question about variational analysis (First moments of uniform distribution on a curve from (0,0) to (1,1) in two-space). Such the question can be addressed in some cases by solving the following integral function:

$f(x)=\frac{\lambda\int_{x}^1 \sqrt{1+(f(t))^2}dt+c_1}{\int_{x}^1 \sqrt{1+(f(t))^2}dt+c_2}$, where $f(x)\in\mathbf{C}^1[0,1]$, and $\lambda$, $c_1$ and $c_2$ are constant numbers.

How to solve the above nonlinear equation?

$\endgroup$
7
  • $\begingroup$ Thanks for your answer, but I think that something is wrong because $f(0)=(-1)^{-0.5}$ is not allowed. $\endgroup$
    – RyanChan
    Commented May 9, 2020 at 5:03
  • $\begingroup$ Yes. $q>1$ (the integral constant) . $\endgroup$
    – Alapan Das
    Commented May 9, 2020 at 5:05
  • $\begingroup$ Let us put $x=0$ in the equation under consideration. This implies $c=0$ and then $f(x)=\lambda$. $\endgroup$
    – user64494
    Commented May 9, 2020 at 6:19
  • $\begingroup$ Thanks for your nice comment. The original posted question is indeed wrong, and now I have edited it. $\endgroup$
    – RyanChan
    Commented May 9, 2020 at 6:54
  • 5
    $\begingroup$ Both original and edited questions are MSE topics. The integral equation under consideration can be reduced to a certain ODE of the first order by its differentiating and taking into account $$ \int_{x}^1 \sqrt{1+(f(t))^2}dt=-{\frac {f \left( x \right) c_{2}-c_{1}}{f \left( x \right) -\lambda}},\,f(1)= \frac {c_1} {c_2}.$$ Maple 2019.1 solves it in terms of roots of a certain transcendental equation. That huge solution is useless. $\endgroup$
    – user64494
    Commented May 9, 2020 at 7:34

1 Answer 1

1
$\begingroup$

$f(x).\int_{0}^{x} \sqrt{1+{f(t)}^2} dt =c+\lambda =a$

Say, $f(x)=y$

So, $\frac{d (\int_{0}^{x} \sqrt{1+{y(t)}^2} dt)}{dx} =\frac{d\frac{a}{y}}{dx}$

From Newton- Leibniz's formula

$\sqrt{1+y^2}=a\frac{d}{dx}(\frac{1}{y})$

Taking $z=(\frac{1}{y})^2$ , we get

$\sqrt{1+z}=a\frac{dz}{dx}$

or, $\frac{dz}{\sqrt{1+z}}= \frac{dx}{a}$

Or, $y=\frac{1}{\sqrt{(\frac{x}{a}+q)^2-1}}$ , ($q>1$ for $a>1$, $q<1$ for $a<0$)

Putting the result $y=\frac{1}{\sqrt{(\frac{x}{a}+q)^2-1}}$ we can easily see that the integral equation is satisfied.

$\endgroup$
1
  • $\begingroup$ Thanks. I have noticed that $f(x)\cdot\int_0^x\sqrt{1+f^2}dt=c+\lambda$ is used in your answer, which is not implied by the integral equation. $\endgroup$
    – RyanChan
    Commented May 9, 2020 at 5:13

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .