Discussion Overview
The discussion revolves around exploring autocorrelation functions in MATLAB, particularly focusing on how to compute autocorrelation for different lags and the differences between various MATLAB functions like autocorr and xcorr. Participants express confusion regarding notation and the implications of using different functions for autocorrelation.
Discussion Character
- Technical explanation
- Debate/contested
- Mathematical reasoning
Main Points Raised
- Mike expresses confusion about the definition of autocorrelation in MATLAB, particularly regarding the notation x(n-l) * x(n-k) and whether MATLAB can compute this.
- One participant clarifies that x(n-l) * x(n-k) can be expressed as x(n) * x(n-(k-l)), suggesting that MATLAB routines can handle this by providing autocorrelation at different lags.
- Mike references matrix forms of autocorrelation and questions the differences between the two representations provided in his book, noting that both matrices seem to describe stationary processes.
- Another participant explains that the MATLAB autocorr function returns a vector of lag autocorrelation values and points out the indexing differences in MATLAB, where indices start at 1.
- Mike mentions using the xcorr() function and describes how it provides values for a Toeplitz matrix, but he is seeking specific lags that are not directly provided by this function.
- A participant questions the use of cross-correlation for autocorrelation, suggesting that many elements in the matrix may be identical and expressing uncertainty about the calculation of these values.
- This participant also asserts that there is no concept of autocorrelation for lags rx(l,k) as proposed by Mike, recommending the use of the autocorr function instead.
Areas of Agreement / Disagreement
Participants do not reach a consensus on the best approach to compute autocorrelation for different lags, with some advocating for the use of the autocorr function while others discuss the implications of using xcorr. There is ongoing confusion and debate regarding the notation and the interpretation of results.
Contextual Notes
There are unresolved questions regarding the differences between the autocorrelation matrices described in the literature and the outputs from MATLAB functions. Additionally, the implications of using cross-correlation in the context of autocorrelation remain unclear.