Peter, I can reproduce your bug, and it seems to be something with gnuplot 3.7.3, which appears not to produce a fit.log file. With gnuplot 3.7.2 (under linux) everything works fine for me.
Does anyone have an older gnuplot version that does the fitting for gstat? -- Edzer
Peter von Bertoldi wrote:
Dear list, A while ago, with some older gstat/gnuplot version I could fit a variogram model using WLS [gnuplot], weights n(h) or WLS [gnuplot], weights n(h)/mod(h)^2 and get a fit.log file (I think that was the name) with some stats in it.
Now, with gstat 2.4.1 (12 March 2003) and G N U P L O T Version 3.7 patchlevel 3 (the newest one) I get the following error message at the bottom of the gstat window: Warning: can't find fit.log -- get gnuplot 3.7 from www.gnuplot.info The rest of the fitting methods work well, as the plotting.
As always gstat and gnuplot are on the same directory. Did I miss something? Has somebody else encountered the same problem and wish to share the solution? Or this gnuplot does not fit anymore?
Thanks for your help
Peter von Bertoldi [EMAIL PROTECTED]
228 Richards Building Phone: (519) 824-4120 x 54264
University of Guelph Fax: (519) 824-5730
GUELPH, Ontario Canada, N1G 2W1 http://www.lrs.uoguelph.ca/