
www.Usenet.com
| <-- __Chronological__ --> | <-- __Thread__ --> |
> If I have a M times N matrix A which only contains complex random > values, all selected from an Gaussian distribution with a given mean > and variance, then are there any expressions available who give the > mean value of the largest/smallest singular values of A ? > > It is easy to numerically compute the mean of largest/smallest > singular value in for example Matlab but I would suspect an > analytical answer may be available in the literature. Random > Matrix theory ? Any references would be appreciated. Thanks! You may find the answer in Alan Edelman's thesis (which you can find on his home page, http://www-math.mit.edu/~edelman/). See also: Eigenvalues and condition numbers of random matrices, by A. Edelman. SIAM Journal on Matrix Analysis and Applications 9 (1988), 543--560. where you can find pdf's for the smallest and largest singular values and simple formulas for the expected values if m,n->oo. Daniel
| <-- __Chronological__ --> | <-- __Thread__ --> |