# Analytically solving ODEs with non-constant coefficients for a specific t

Printable View

• Dec 7th 2011, 08:20 PM
thesteve
Analytically solving ODEs with non-constant coefficients for a specific t
Given an ODE in the form of f(t)y''+g(t)y'+h(t)y=0

Since the coefficients are functions of t in order to find a solution for all t this would need to be done numerically. But If all I am looking for is the y(t) at a specific value of t and NOT the general solution, can I just plug in that value of t into the coefficients of the original ODE and then solve it using standard analytically techniques or is a numeric solution the only way?
• Dec 8th 2011, 03:37 PM
Ackbeet
Re: Analytically solving ODEs with non-constant coefficients for a specific t
Quote:

Originally Posted by thesteve
Given an ODE in the form of f(t)y''+g(t)y'+h(t)y=0

Since the coefficients are functions of t in order to find a solution for all t this would need to be done numerically. But If all I am looking for is the y(t) at a specific value of t and NOT the general solution, can I just plug in that value of t into the coefficients of the original ODE and then solve it using standard analytically techniques or is a numeric solution the only way?

Actually, if the functions f(t), g(t), and h(t) have relatively standard Taylor series expansions, you could attempt a series solution. You cannot just plug in one valur of t in order to find y(t) at just that one point, for several reasons. 1. Merely by writing y'(t), you are writing a limit. Limits don't care what happens at a point, they care what happens near a point. Hence, you must have information near that point. 2. You have specified no initial conditions. The DE you wrote down, a linear second-order homogeneous ODE, has two arbitrary constants in its solution (due to the two integrations you are essentially performing) which must be determined by initial conditions. Therefore, there is no way to pin down the solution to the DE, even if you could exhibit it, because it would have two arbitrary constants.

Do you have a specific f(t), g(t), and h(t) in mind? If so, why not post them?