perurbation projection vector
I think you have to point out a change in some network variables depending on a pertubation source.
If for instance the pertubation source is a changing device current of an active device in your oscillator then you have to run two simulations. One with, the other w/o the pertubation current on. Then compare the difference over time, typical over a period.
There is also a typical characterisation which is called "impulse sensitivity function". There the pertubation source is a impulse which is a very fraction of the oscillation period. Most setups use a gaussian function. The change of phase depending on the actual position within the period is then plotted. It gives insight at which phases the oscillator is sensitive to pertubations.