Talk:Bispectrum

Poor definition
I'm not sure what the bispectrum is, but the explanation here is poorly written.

1. It is defined here in terms of cumulants, itself a concept whose WP article doesn't promise clarity. (If I were interested in cumulants, perhaps it would be more useful, I can't say. But that's not my current interest.)

2. If I understand correctly, the bispectrum of 2 signals (f,g) is, precisely, cov(f+g,f*g), where "*" = convolution and "cov" = covariance (i.e., correlation x product of standard deviations). Is this true? If so, it would be much more helpful to define it that way before explaining its relation to cumulants and its computation by Fourier transform.

It would be especially helpful to explain the underlying intuition, i.e., how it encodes information that, say, EEG analysts have found to be meaningful, or why one might turn to bispectra when studying certain kinds of related signals. Jmacwiki (talk) 20:43, 2 June 2023 (UTC)