Re: global fit

*To*: mathgroup at smc.vnet.net*Subject*: [mg57496] Re: global fit*From*: "Pascal" <Pascal.Plaza at ens.fr>*Date*: Sun, 29 May 2005 21:00:14 -0400 (EDT)*References*: <d79dra$l1l$1@smc.vnet.net>*Sender*: owner-wri-mathgroup at wolfram.com

I should probably have mentioned the form of that function I wish to find the minimum of: g[a_, b_, c_, d_] := ( A = {Table[f1[n, a, c, d], {n, 1, 60}], Table[f2[n, b, c, d], {n, 1, 60}], Table[f3[n, c, d], {n, 1, 60}], Table[f4[n, c, d], {n, 1, 60}], Table[1, {n, 1, 60}]}; invA = PseudoInverse[A]; X = A.invA; residue = B - X.A; Norm[Flatten[residue]] ) This function is certainly a bit slow but it does work. I could check with ContourPlot the existence of a minimum along a and b (c and d being fixed). I tried FindMinimum just on g[a,b0,c0,d0] with b, c and d fixed and giving a very good starting point and limits for a but it also failed.