Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo
Close
From: Andrey V. Komolkin <komolkin@es...>  20070317 15:20:16

Dear colleagues, I am frequent user and enthusiast of "gnuplot", but not a developer. Let me propose some new features for the program. I hope these are new, because I did not find any references to such features in documentation for new version 4.3.  First. In the "plot" command I propose to allow to setup range of 'x' (independent) variable to each of functions/files. For instance: set xrange [0:10] plot sin(x), cos(x) range [1:9], 'file.dat' range [1:9] with lines In this example, sin(x) should be shown in range [0:10], but cos(x) and data from 'file.dat' in different ranges, which are set individually for these graphs. This may be useful in the case we like to plot tabulated data and its approximation (analytical function) in the same plot. Imagine, the data we like to approximate have two parts: f1(x)=a*x+b fit [0:5] f1(x) 'file.dat' via a,b f2(x)=c*x+d fit [6:10] f2(x) 'file.dat' via c,d set xr [0:10] plot 'file.dat' w p, [0:5] f1(x), [6:10] f2(x) I do not like to plot analytical function f1(x) outside the range [0:5], where it was fitted and (in general) defined. And vice versa, f2(x) was defined in range [6:10]...  Second. Let me propose "dashed" ranges: [0:10,20:30]. This feature may be useful for approximation of experimental data. Imagine, experimental data g(x) contains several peaks f(x) on the "basement" 0.02*x+1.0. We need to remove trend (0.02*x+1.0) from the data. set xr [10:10] set yr [0:7] f(x,a,x0,dx)=a*exp((xx0)**2/dx**2) g(x)=f(x,5,1,1)+f(x,2,3,0.5)+1.00.02*x This trend can not be determined on the whole range [10:10], but on two ends of the range: t(x)=a*x+b fit [*:5,5:*] t(x) 'file.dat' via a,b set xr [10:10] plot t(x) range [10:5,5:10], 'file.dat' with points  Conclusion. I proposed in this letter two new features for the 'gnuplot'. First: separate ranges for each functions/files in 'plot' command Second: "dashed" ("united" or however it will be called) ranges which consist of several (two, three or even more) subranges.  I like to receive comments on my proposals. Best wishes, Andrei Komolkin. ====================================================================== Dr. Andrei V. Komolkin Senior Lecturer (University Reader) V. A. Fock Institute of Physics Tel.: +7(812)4284479 SaintPetersburg State University Fax: +7(812)4287240 198504, SaintPetersburg Email: komolkin@... Russia 
From: Lutz Maibaum <maibaum@be...>  20070318 01:10:51

Dear Andrey, I believe the features you are proposing might already be available in gnuplot, albeit with a less concise syntax. On Saturday 17 March 2007 11:40, Andrey V. Komolkin wrote: > First. In the "plot" command I propose to allow to setup range of 'x' > (independent) variable to each of functions/files. For instance: > > set xrange [0:10] > plot sin(x), cos(x) range [1:9], 'file.dat' range [1:9] with lines This can be done using set xrange [0:10] plot sin(x), (x >= 1 && x <= 9 ? cos(x) : 1/0), 'file.dat' using 1:($1 >= 1 && $1 <= 9 ? $2 : 1/0) with lines (This assumes that file.dat contains rows of (x,y) pairs. If instead it contains rows of single numbers, replace $1 by $0 and $2 by $1). > Imagine, the data we like to approximate have two parts: > > f1(x)=a*x+b > fit [0:5] f1(x) 'file.dat' via a,b > f2(x)=c*x+d > fit [6:10] f2(x) 'file.dat' via c,d > set xr [0:10] > plot 'file.dat' w p, [0:5] f1(x), [6:10] f2(x) Try plot 'file.dat' w p, (x >= 0 && x <= 5 ? f1(x) : 1/0), (x >= 6 && x <= 10 ? f2(x) : 1/0) > Second. Let me propose "dashed" ranges: [0:10,20:30]. This feature may > be useful for approximation of experimental data. > t(x)=a*x+b > fit [*:5,5:*] t(x) 'file.dat' via a,b > set xr [10:10] > plot t(x) range [10:5,5:10], 'file.dat' with points t(x)=a*x+b fit t(x) 'file.dat' using 1:($1 < 5  $1 > 5 ? $2 : 1/0) via a,b set xr [10:10] plot (x < 5  x > 5 ? t(x) : 1/0), 'file.dat' with points You can find the details in "help using" and "help ternary". Hope this helps, Lutz 