pm1366 said:
no

,no where talking about that ... it is an article about cylindrical shells , i can give it to anyone who want to have a look...
Look up 'repeated roots' and 'differential equations' in your textbooks or on the internet.
The solution strategy is this:
You have a homogeneous differential equation (ie one that solves Dy = 0, where D is some differential operator).
Taking a function and adding derivatives of itself to get zero is only possible with the exponential function, so it is assumed that the solution will be of the form y=e
λt . Plug that into the differential equation and you have conditions on λ for that solution to work. The condition on λ is called the characteristic equation . Since y=e
λt for a specific λ satisfying the characteristic equation has been found to solve the homogeneous equation, a sum of all e
λt with every λ found must also solve the homogeneous equation because differentiation is linear. So the total homogeneous solution is Ʃ A
ie
λit as your excerpt states.
From other considerations we know an nth order differential equation has n independent functions that satisfy the homogeneous equation. If some of the λ are repeated (eg you have four zeroes for λ), you are not finding all those functions so you have to have another way to find them. That part the article hasn't mentioned but HallsOfIvy gave you the solutions -- the idea is to assume the solution is the form of e
λt, te
λt, t
2e
λt, ... the highest power of t depending on how many repeated roots you have (if you have four repeated roots, you need four equations with that root). You can confirm these also satisfy the differential equation by substitution and the resulting complete solution will be a summation of all such functions you found.
The idea of doing this is called 'variation of parameters'. The notion was Ae
λt is one solution, what if A was allowed to be a function of time, would that work? Plug into the differential equation, and lo and behold, it does with A=t
n for some n.