Second order DE

by fluidistic
Tags: order
 PF Gold P: 3,188 1. The problem statement, all variables and given/known data Find the general solution to ##A(x)y''+A'(x)y'+\frac{y}{A(x)}=0## where A(x) is a known function and y(x) is the unknown one. Hint:Eliminate the term that contains the first derivative. 2. Relevant equations Not sure. 3. The attempt at a solution So I don't really know how to tackle it. I guess that the hint suggests a change of variable that would get rid of the y' term. So I tried z=y'A, z=Ay and z=y/A. All failed to express the ODE as a function of z and its derivative(s) and A and its derivative(s). I am therefore stuck. Any other hint will be welcome!
 Mentor P: 18,240 What if you try the following substition $y=Bz$, where B is a known function that is yet to be decided. What do you get if you perform this substitution? What should we take for B if we want the z' term to vanish?
PF Gold
P: 3,188
 Quote by micromass What if you try the following substition $y=Bz$, where B is a known function that is yet to be decided. What do you get if you perform this substitution?
Good idea! ##x''A'B'+z'(AB+AB'+A'B)+z (A'B'+B/A)=0##.

 What should we take for B if we want the z' term to vanish?
##B=Ce^{-x}/A##. I checked out and indeed the terms in front of ##z'## vanishes.
I'm left to solve ##z''A'Ce^{-x} \left ( \frac{1}{A}+ \frac{A'}{A^2} \right ) + z \left ( \frac{A'}{A} + \frac{A'^2}{A^2}- \frac{Ce^{-x}}{A^2} \right ) =0##. That really does not look beautiful/easy to me.

Mentor
P: 18,240
Second order DE

 Quote by fluidistic Good idea! ##x''A'B'+z'(AB+AB'+A'B)+z (A'B'+B/A)=0##.
I get something different here. I get that the term of z' is
$$A^\prime B + 2A B^\prime$$

The form of my B is also much simpler.
PF Gold
P: 3,188
 Quote by micromass I get something different here. I get that the term of z' is $$A^\prime B + 2A B^\prime$$ The form of my B is also much simpler.
I see, I made a mistake; I don't know why I make so many of them. I reach the same as yours, so ##B=kA^{-1/2}## where k is a constant.

Edit: I might have rushed through the math but I reach that ##z''+z \left ( 1-\frac{2}{A'} \right ) =0##. Which has different solutions depending on the sign of the term in front of z. I guess I made another mistakes.
Mentor
P: 18,240
 Quote by fluidistic I see, I made a mistake; I don't know why I make so many of them. I reach the same as yours, so ##B=kA^{-1/2}## where k is a constant.
OK, I get the same B. But there is no real need of the constant. You just need to find one single B that makes z' disappear, so you can take k=1.

Anyway, with that choice of B, what does your equation become then?
PF Gold
P: 3,188
 Quote by micromass OK, I get the same B. But there is no real need of the constant. You just need to find one single B that makes z' disappear, so you can take k=1. Anyway, with that choice of B, what does your equation become then?
I now get ##B'=-\frac{1}{2}A^{-3/2}##, ##B''=\frac{3}{4}A^{-5/2}A'^2-\frac{A^{-3/2}A''}{2}##.
The equation becomes ##z''AB+z(AB''+A'B'+B/A)=0##. Now I have to replace the B and its derivatives in that.
Edit: I've just done it and it's not beautiful so far.

 Related Discussions Calculus & Beyond Homework 3 Differential Equations 6 Calculus & Beyond Homework 10 Differential Equations 1 Calculus & Beyond Homework 0