# Identifying a second-order ODE

1. Aug 7, 2012

### Undoubtedly0

Is there a general solution to

$$\frac{d}{dt}\left[p(t)\frac{dx(t)}{dt}\right] + q(t)x(t) = 0$$

for $x(t)$ when $p(t)$ and $q(t)$ are arbitrary functions? Better yet, does this question have a name, or some identifier, that I could look in to? It might appear more familiar written as

$$\left[p(t)x^\prime\right]^\prime + q(t)x = 0$$

2. Aug 8, 2012

### JJacquelin

No, there is no genertal solution
.

#### Attached Files:

• ###### GeneralODE.JPG
File size:
41.7 KB
Views:
88
3. Aug 10, 2012

### Staff: Mentor

Look up Sturm-Liouville problems or equations.

4. Aug 11, 2012

### GarageDweller

I think that's just a ODE with non constant coeffecients, since expanding yields

P(t)x''+P'(t)x'+q(t)x=0

You may be able to solve this with power series if P and q fit them.
Non linear differential equations rarely have closed form solutions.
But that's okay, we have computers

5. Aug 11, 2012

### JJacquelin

Of course, when I say "There is no general solution", I mean "No general analytical solution espressed on a closed form".
Obviously, in some particular cases, with some particular forms of functions p(t) and q(t), the solutions might be known on closed form, and/or be expressed as infinite series.
Even more generally the solutions can be accurately approached thanks to numerical methods.

Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Similar Threads for Identifying second order Date
I Resolution of a PDE with second order Runge-Kutta Oct 3, 2017
Help me identify problem Apr 7, 2015
Help identifying ODE Sep 16, 2008
Help Identifying Diagrams? Sep 2, 2005
DE Identifying/Solving Homo/Bernoullis/Exact/etc Sep 25, 2004