bo198214 Wrote:You have to demonstrate it. I *dont* think that there are such stable values except perhaps 1.[update] just got aware about the timestamps of our posting. I seem to have edited the previous *after* your msg, see "update". I withdrew the "b out of range of convergence". I was in error with that, it was when 1/e^e<b<1. [/update]
Just uploaded a b=0.5-document, anyway. I've taken size 4x4 to 32x32, 200 or 400 digits float precision(computation ~ aug 07; don't remember the exact used precision)
see Eigenvalues for b=0.5 (truncations)
Quote:As you see from the graphs in this thread I (maple) also only went to size 30. And that only with help of some university machines (took perhaps 10min there), as my home computer wouldnt make it above 25 in reasonable time. But its not only that the diagonalizaton takes so long I also had to adjust the precision in dependence of the matrix size, which again took rounds of computation.Hmm. In de.sci.mathematik someone told me, that he did it with maple with size 128x128 in some dozen seconds or few minutes. Strange. Maybe he was not correct with the computation. Anyway, that was short before the time when I found an analytical description for the infinite case and I didn't proceed the truncation-path.
Gottfried Helms, Kassel