Marco van Leeuwen wrote: > > Hi all, > > I'm writing a polynomial fitting routine, using the least squares > approximation, but I'm encountering some difficulties with the > matrix-inversion. It seems that the problem is sensitive for numerical > roudoff's. > Of course, using the fitting functionality of ROOT is still an alternative, > but if anyone knows the answers to the following, please let me know. > > 1) Is there an improved version TMatrix, in double precision?? > This is currently under discussion. Something will come. I do not understand why you cannot use TH1::Fit or TGraph::Fit with polynomials. You can may be look anyhow at the following static functions defined in the implementation of the class TH1. extern void H1LeastSquareFit(Int_t n, Int_t m, Double_t *a); extern void H1LeastSquareLinearFit(Int_t ndata, Double_t &a0, Double_t &a1, Int_t &ifail); extern void H1LeastSquareSeqnd(Int_t n, Double_t *a, Int_t idim, Int_t &ifail, Int_t k, Double_t *b); > 2) Does anyone know a standard routine for calculating the binomial > coeefficients (i.e. n over k, the number of permutions of k things in a > pool of n: n!/(k!*(n-k)!) ) > > 3) Or: is there already a straight-forward least-square fitting algorithm > for polynomials? Yes, this is the ABC of fitting. As I said above, polynomial fitting is provided by TH1::Fit or TGraph::Fit using the predefined functions pol0, pol1, pol2, etc. Polynimial fitting is also possible via the mouse on a TH1 or TGraph object. select the Fitpanel item in the context menu. Rene Brun
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 00:43:38 MET