<t>I would suggest the classicalM. Reed and B. Simon, Methods of Modern Mathematical Physics I: Functional Analysis, Academic Press, ISBN: 0-12-585050-6M. Reed and B. Simon, Methods of Modern Mathematical Physics II. Fourier Analysis , Self-Adjointness, Academic Press, ISBN: 0-12-585002-6and W. Rudi...

<t>QuoteOriginally posted by: AaronI much prefer to use a T matrix with structure so that I have many fewer than n*(n+1)/2 parameters. ...You want a T matrix that (a) you can estimate reliably with known confidence intervals, (b) you can compute robustly without <i>ad hoc</i> techniques or numerical...

<t>There is a nice M.Sc. thesis from Craig Lucas (a student of Nick Higham) on "Computing Nearest Covariance and Correlation Matrices" (Oct. 2001, should be available on the net). I don't remember the details, but I remember it was interesting and was specifically on the problem of "which is the cor...