Orbital simulator and non fixed orbit

[FLOUR]

Hello,
Let's consider a small 3 planet system. One of the planet's is fixed and is a lot heavier than the others. The acelleration is
mP2 * aP2 = - FP3 - FP1
From with the velocity is derived with allows to calculate the positions of the planet's in a simulator program. After running the simulation, The orbit of one of the "moving" planets isn't fixed - some variation occur after each loop.

What can cause this? The other planet (that's not fixed) has an elliptic fixed orbit.
FP = Gravitational Force of Planet
Thanks.

 

[FLOUR]

Hello,
some more data about the situation:

Planet 1 - fixed on the center of the referential
Planet 2 - initial distance = 3 AU (start_x= 3 AU, start_y = 0)
Planet 3 - initial distance = 6 AU (start_x= 6 AU, start_y=0)

So Planet 2 is on the "middle" of the system.

Planet 1 is about 1000 times "heavier" than Planet 2.
Planet 3 is about 30 times "heavier" than Planet 2.

Planet 3's orbit is fixed elliptic. Planet 2's orbit has some variation to the initial elliptic orbit.

1) I was under the impression that orbits were suposed to be fixed...
2) I'm guessing the planet in the "middle" is getting heavy interaction from the other planet's...

Thanks for your comments.