Initial transient effects in the frequency domain

The effect of an additive initial transient on the harmonic estimates used in frequency-domain simulation factor screening experiments is discussed. It is shown that under reasonable conditions, the harmonic effect is asymptotically negligible. This implies that the initial transient effect is no worse, asymptotically, in the frequency domain than in the time domain. For finite sample sizes, it is shown how the additive initial transient effect changes the distribution of the signal-to-noise ratio, which in turn may affect the power of the statistic for rejecting the null hypothesis. In order to reduce an additive initial transient effect, the difference between corresponding observations from the signal and noise runs can be considered. If the transient is identical for both runs, then it is eliminated in this difference between the signal and noise runs.