Back to index

We know how to solve second order equations where a, b, c are constants. Even if they're not constant some can be expressed as a linear equation. But not always will they be solvable. However, there is one class of second order equations with non constant coefficients that are always solvable.

Cauchy-Euler equations

If it has a name in it, its very important, if it has 2 names, its very very important!

cauchy-euler equations are equations in the form:

$$ax^2{\frac{d^2y}{dx^2}+bx{\frac{ dy }{ dx }}+cy=f(x)},\ x>0$$

where $a,\ b,\ c$ are still constants and $\in \mathbb{R}$

Note: $x=0$ is not interesting as the derivative terms disappear.

How to solve? There are two approaches:

Textbook only use 2nd method, prof doesn't like this. You can find both methods in the profs notes. Btw, do you know Stewart? Multimillionaire, he's living in a mansion in Ontario.

introduce change of variables:

$x=e^t\Rightarrow t=\ln x$ (x is always +)

(do $x=-e^t$ if you need it to be negative.)

find derivatives with respect to t now. y is a function of t which is a function of x.

$\frac{dy}{dx}=\frac{dy}{dt}{\frac{dt}{dx}}=\frac{ dy }{ dt }{\frac{1}{x}}\Rightarrow \underset{ \text{Important} }{ x\frac{dy}{dx}=\frac{dy}{dt} }$

using $\frac{dy}{dx}=\frac{dy}{dt}{\frac{dt}{dx}}$ and the chain rule, compute 2nd derivative of y wrt to x:

$\frac{d^2y}{dx^2}=\frac{d^2y}{dt^2} \frac{dt}{dx}\cdot \frac{dt}{dx}+\frac{dy}{dt}{\frac{d^2t}{dx^2}}=\frac{1}{x^2}{\frac{d^2y}{dt^2}}-\frac{1}{x^2}\frac{dy}{dt}$

$\underset{ \text{Important} }{ x^2{\frac{d^2y}{dx^2}}=\frac{d^2y}{dt^2}-\frac{dy}{dt} }$

plugging those derivatives in we get: remember

$$a\frac{d^2y}{dt^2}+(b-a){\frac{dy}{dt}}+cy(t)=f(e^t)$$

^ this is a constant coefficient second order non-homogenous equation now! We can solve it now using prior tools.

If you make the substitution $x=-e^t$ and go through the derivation, you get:

$a\frac{d^2y}{dt^2}+(b-a){\frac{dy}{dt}}+cy(t)=f(-e^t)$ Very nice that it's so similar, makes it easy to remember.

Example:

ex second_order second_order_nonhomogenous cauchy-euler

Find the general solution for:

$$x^2{\frac{d^2y}{dx^2}}+3x{\frac{dy}{dx}}+y=x^{-1},\ x>0$$

substitute: $x=e^t$

transform using the technique we showed just earlier:

$\frac{d^2y}{dt^2}+2{\frac{dy}{dt}}+y=e^{-t}$

  1. $r^2+2r+1=0$

$r_{1,2}=-1$

$y_{h}(t)=c_{1}e^{-t}+c_{2}te^{-t}$

  1. $y_{p}(t)=At^2e^{-t}$ using method of undetermined coefficients

$\underbrace{ \cancel{ At^2e^{-t} }+\cancel{ A 2t(-e^{-t}) }+2Ae^{-t}\cancel{ -2Ate^{-t} } }_{ y_{p}'' }\quad+\underbrace{ \cancel{ 2At^2(-e^{-t}) }+\cancel{ 2A 2te^{-t} } }_{ 2y_{p}' }\quad+\underbrace{\cancel{ At^2e^{-t} } }_{ y_{p} }=e^{-t}$

$2Ae^{-t}=e^{-t}$

$A=\frac{1}{2}$

general solution in terms of $t$:

$y(t)=c_{1}e^{-t}+c_{2}te^{-t}+\frac{1}{2}t^2e^{-t}$

but we want solution in terms of $x$:

$y(x)=c_{1}e^{-\ln(x)}+c_{2}\ln(x)e^{-\ln(x)}+\frac{1}{2}\ln(x)^2e^{-\ln(x)}$ This is rather lousy notation, the y here isn't the same as the y above. Conceptually though, it's all oke doke.

$$y(x)=c_{1}x^{-1}+c_{2}\ln(x)x^{-1}+\frac{1}{2}{\ln(x)^2}x^{-1}$$

We are done.

end of lecture 10 start of lecture 11

Last lecture we did Cauchy Euler equations:

$$ax^2{\frac{d^2y}{dx^2}+bx{\frac{ dy }{ dx }}+cy=f(x)} \qquad x>0$$

where $a,\ b,\ c$ are constants and $\in \mathbb{R}$

substitute $x=e^t$

$a{\frac{d^2y}{dt^2}}+(b-a){\frac{dy}{dt}}+cy=f(e^t)$ lousy notation, the y here isn't quite the same as in the above definition.

substitute: $y=x^r$

after calculating derivatives, plugging in, and simplifying we obtain the polynomial equation:

$ar^2+(b-a)r+C=0$

Three cases:

(i) $r_1\ne r_{2}$ then:

$y_{h}(t)=c_{1}e^{rt}+c_{2}e^{rt}$

$y_{h}(x)=c_{1}x^{r_{1}}+c_{2}x^{r_{2}}$ (lousy notation, because the two $y_{h}$ do not equal each other)

(ii) $r_{1}=r_{2}=r$ then:

$y_{h}(t)=c_{1}e^{rt}+c_{2}te^{rt}$

$y_{h}(x)=c_{1}x^r+c_{2}x^r\ln(x)$ (derived by reduction of order.)

(iii) $r_{1,2}=\alpha\pm i\beta$ then:

$y_{h}=e^\alpha(c_{1}\cos(\beta t)+c_{2}\sin(\beta t))$

$y_{h}(x)=x^\alpha(c_{1}\cos(\beta\ln x)+c_{2}\sin(\beta \ln x))$

Now compute your particular solution, $y_{p}$, and combine with $y_{h}$ to obtain your general solution.

Back to index