If the circuit is only weakly non-linear, you will tend to get better accuracy with flexible balance compared with shooting (at least, more cheaply) - if you're trying to resolve signals such as 3rd order intermodulation products which will be a long way down when you have small signal inputs. The numerical noise floor when representing a harmonic response with a small number of terms will be lower than when representing the same set of harmonics with a sequence of low-order polynomials.

If the circuit is strongly non-linear (e.g. switching circuits, dividers, circuits driven with square waves, circuits driven into compression), then shooting will tend to be more accurate as you would need hundreds of harmonics to represent the solution with flexible balance - it's much easier to represent something closer to a square wave with a set of low order polynomials.

The above are a bit of a generalisation, but maybe enough to answer your query.

Regards,

Andrew.

No. The transient solution is still a sequence of low-order polynomials trying to represent a sum of sinusoids, so it will have similar accuracy limits to shooting (not surprising, since shooting is transient-based).

Andrew.