Chapter 3: Linear Prediction Extrapolation

Linear prediction models each sample in a discretely sampled signal as a linear combination of the preceding samples. It implicitly corresponds to treating the signal as a sum of sinusoids and provides a straightforward means for extrapolating the signal beyond the measured interval. It was the first such method employed to ameliorate short data records in the indirect dimensions of multidimensional NMR experiments and remains one of the most widely used. In this chapter we describe the history, theory, strengths and weaknesses of linear prediction extrapolation.