Controlled loss of information
The same time series could also stand for a totally different signal (which passes through the same red squares and fits even better), because:
- Amplitudes have been rounded to discrete levels.
- Information about the time course of the signal between the samples is lost.
Nevertheless, the digital representation is accurate enough, provided that we know that:
- The amplitude resolution was sufficient to reproduce the smallest variation of interest.
- The spacing of samples was sufficiently dense in order not to miss any of the signal's variations.
The next two sections, Quantization noise and Sampling theorem, will deal with these two premises.