摘要
Wavelets are applied to a regression model with an additive stationary noise. By checking the empirical wavelet coefficients with significantly large absolute values across fine scale levels, the jump points are detected first. Then the cusp points are identified by checking the wavelet coefficients with significantly large absolute values which are secondly large only to the previous wavelet coefficient across fine scale levels. All estimators are shown to be consistent.
Wavelets are applied to a regression model with an additive stationary noise. By checking the empirical wavelet coefficients with significantly large absolute values across fine scale levels, the jump points are detected first. Then the cusp points are identified by checking the wavelet coefficients with significantly large absolute values which are secondly large only to the previous wavelet coefficient across fine scale levels. All estimators are shown to be consistent.
