sigma vs. lambda

[JButtlar]

Hi,

I recently get the following warning (probably after updating to Rssa 0.11):

(converted from warning) the field lambda' is deprecated. usesigma' instead.

I have until now always used ssaresults <- ssa() ssaresult$lambda

to identify situations where I set neig too huge and ssa() returned less eigentriples than neig (in which case ssaresult$lambda was zero). I interpreted lambda to be related to the variance of the respective eigentriple.

Does ssaresult$sigma now return the identical information and can be used in the same manner?

Thanks for your help and thanks for sharing your SSA code in general Jannis

[eodus]

Yes, sigma is absolutely identical to lambda. You can just replace '$lambda' by '$sigma' in your existing code and everything will be ok.

We renamed lambda to sigma because of the fact that in the most of the articles lambda denotes 'eigenvalues of the matrix X^T X' and 'singular values of the trajectory matrix X' is denoted by sigma

[eodus]

You can also use nsigma(ssaresult) for getting the number of really computed singular values and nu(ssaresult) for getting the number of really computed eigenvectors (left singular vectors). And also sigma (former lambda) is not VARIANCE of the eigentriple; it is STANDARD DEVIATION (sq.root of variance)

[JButtlar]

Thanks for your help. You are absolutely right about the variance/std.dev difference.