comments ...in reverse order...I think you mean to plot a response graph as a result in a harmonic linear dynamic study. Its known as a gain or a bode plot in audio circles. As for what to expect the reactions to be ...as in equal and opposite...I'd sooner expect them to be unequal and not opposite at step 60 ie. 50 khz. At high frequency the step may be near a resonance that is excited by the load. So check the mode list to see what modes lie around 50khz. Then, display the mode shape to get an idea of whether the load you apply would excite the mode shape. My lean is toward there being reaction forces to a surprising behavior of the geometry at 50khz which in order to keep the fixed items fixed, the reactions diverge. On the other end of the spectrum, in the low frequency range below the fundamental frequency ie. in a quas- static case, I'd expect the simulation to approximate reactions as equal and opposite. If experimental measurement confirms the reactions are equal and opposite at 50khz, one possibility is adjusting damping to null out the resonance peak and realize your equal and opposite expectation.