Numerical solution for Ordinary diferential ecuation

[chire]

I started to solve some ordinary diferential ecuations (ODE) with matlab for some specific research topic.
im using the internal solver algorithm called ode45

i am not familiar with the advanced concepts of numerical methods so i ask the following.

is it possible that the numerical solution computes "wrong" and result should not be trusted??

ill explain myself, for an specific ODE ecuation, im getting a weird curve as the numerical solution. This does not happen on other simpler cases.

more specific: this is what im getting

http://dl.dropbox.com/u/6380744/untitled.jpg

plotted from 0 to pi/2.

as it approaches pi/2 it grows exponentially too fast for our context, and as consecuence the rest of the curve cannot be appreciated and looks flat.
we are not sure if this numerical solution can be trusted
any help or explanation, is thanked.

if you want to reproduce the case i can give you the ODE and the initial conditions

[nuzzle]

I'd say Matlab can be trusted. It's a very solid package. If you get weird results it's more likely a mistake you did.

And no one can tell just by looking at a curve whether it's the correct solution to an unknown ODE. That's pretty obvious isn't it.

[HochiminhCity]

I don't know about ode45,
I only know classic method Newton to resolve differential equations numerically. I use win32 API to draw the curve or OpenGl as an alternative graphics solution

[chire]

thanks,

i was checking and it seems we have a situation on the ecuation, some complex numbers appear
in this cases, i cannot blame numerical solutions
Cristobal