5

For $\alpha \in \mathbb{R}$, let $y_{\alpha}(x)$ be the solution of the differential equation $$\frac{dy}{dx} + 2 y = \frac{1}{1+x^2},\text{satisfying} \space y(0)=\alpha.$$Then prove that $\lim\limits_{x\to\infty}y_{\alpha}(x)=0$.


Since the above differential equation is a first order linear differential equation with integrating factor $e^{2x}$. Therefore its solution is given by, $$y_{\alpha}(x) = e^{-2x} \left(\int \left(\frac{e^{2x}}{1+x^2}\right) \rm d x + c\right)$$ Now I am stuck with this integral. I have tried integration by parts but because of exponential and $\frac{1}{1+x^2}$ term it is not useful.I tried substitutions but did not get any help. Kindy someone help me to solve this integral or if there is any other method to solve it then please share.

San Ti
  • 391

3 Answers3

7

You don't need to do the integral. Put the $e^{-2x}$ in the denominator and use L'hospital's Rule and Fundamental Theorem of Calculus.

$$\lim_{x \to \infty} \frac{\int_0^x \frac{e^{2t}}{1+t^2}\; dt}{e^{2x}} =\lim_{x\to \infty} \frac{ \frac{e^{2x}}{1+x^2}}{2e^{2x}} =\lim_{x\to \infty} \frac{1}{2(1+x^2)} = 0. $$

B. Goddard
  • 33,728
  • 1
    This is not the right solution to the initial value problem though, as $y(0)=0$ in your case. It should be $\frac{\alpha+\int_{0}^{x}\frac{e^{2t}}{1+t^2} dt}{e^{2x}}$ – Davide Masi Jan 03 '25 at 13:49
  • 1
    @DavideMasi You are right but that term is immaterial as it will become zero after the limit.The main focus was on the integral term and this solution addresses the issue.Thank you for your comment. – San Ti Jan 03 '25 at 15:05
  • 1
    I know this was not the focus, I wrote this comment so that you would not be confused :) – Davide Masi Jan 03 '25 at 15:08
  • 1
    @DavideMasi Thank you so much for the clarification. – San Ti Jan 03 '25 at 15:11
2

By definition of the exponential integral function

$$\int \frac{e^{-x}}{x} d \ x = Ei[x]+c.$$

By the series it's a $\log(x)$ plus an entier function $$\mathop{Ei}(x) \ = \ \log x + \sum_{n=1}^\infty \frac{(-x)^{n}}{n \ n!}$$

For your problem use partial fraction decomposition

$$\frac{1}{1+x^2} = \frac{1}{2}\left( \frac{1}{x +i } + \frac{1}{x-i} \right)$$

and set the integration path and boundaries such that for $x=0$ the value a is assumed.

Roland F
  • 5,122
2

Here is another proof: \begin{align} y_{\alpha}(x)&=e^{-2x}\left(\alpha+\int_0^x\frac{e^{2t}}{1+t^2}\,dt\right) \\ &=e^{-2x}\left(\alpha+\int_0^{x/2}\frac{e^{2t}}{1+t^2}\,dt +\int_{x/2}^{x}\frac{e^{2t}}{1+t^2}\,dt\right), \tag{1} \end{align} hence $$ |y_{\alpha}(x)|\leq e^{-2x}\left(|\alpha|+\left|\int_0^{x/2}\frac{e^{2t}}{1+t^2}\,dt\right|+\left|\int_{x/2}^{x}\frac{e^{2t}}{1+t^2}\,dt\right|\right). \tag{2} $$ For $x>0$ we have $$ \left|\int_0^{x/2}\frac{e^{2t}}{1+t^2}\,dt\right|<\int_0^{x/2}e^{2t}\,dt<\frac{e^x}{2} \tag{3} $$ and $$ \left|\int_{x/2}^x\frac{e^{2t}}{1+t^2}\,dt\right|<\frac{1}{1+(x/2)^2}\int_{x/2}^{x}e^{2t}\,dt<\frac{2e^{2x}}{4+x^2}. \tag{4} $$ Plugging $(3)$ and $(4)$ into $(2)$ we get $$ |y_{\alpha}(x)|<e^{-2x}|\alpha|+\frac{e^{-x}}{2}+\frac{2}{4+x^2}\qquad(x>0), \tag{5} $$ from which follows $$ \lim_{x\to+\infty}y_{\alpha}(x)=0.\qquad\square \tag{6} $$

Gonçalo
  • 15,869
  • I am having trouble with this inequality $$ \left|\int_{x/2}^x\frac{e^{2t}}{1+t^2},dt\right|<\frac{1}{1+(x/2)^2}\int_{x/2}^{x}e^{2t},dt$$ – San Ti Jan 04 '25 at 10:10
  • 1
    Notice that $\frac{1}{1+t^2}<\frac{1}{1+a^2}$ if $t>a>0$. Therefore, $$\int_{a}^b\frac{e^{2t}}{1+t^2},dt<\int_a^b\frac{e^{2t}}{1+a^2},dt=\frac{1}{1+a^2}\int_a^be^{2t},dt$$ if $0<a<b.$ – Gonçalo Jan 04 '25 at 17:28
  • 1
    Clear now.Thank you for this excellent proof. – San Ti Jan 05 '25 at 02:34