From: Hans-Bernhard B. <HBB...@t-...> - 2011-03-18 19:00:16
|
On 18.03.2011 17:01, Douglas Mason wrote: > [This is a question that came up when I was comparing some original data > to an approximation which was computed through an external program. You > can see the original plot in the attachment. My goal is to quantify the > variation between the data and the approximation.] But why are you trying to use gnuplot for that? You need a data processing tool, not a plotting program. When you want to compare data sets in a plotting program, you plot them into a common diagram and look at it. For quantifying differences, you run a data processor. As far as filledcurves is concerned, all it takes would be to transform your data, adding the "extra" points introduced by "steps" or "fsteps". x1 y1 x2 y2 x3 y3 --> x1 y1 # "steps" x2 y1 x2 y2 x3 y2 x3 y3 |
From: Douglas M. <dou...@gm...> - 2011-03-18 19:42:26
|
That is a valid response, however, I don't believe it to be outside the realm of what gnuplot offers. For instance, people often use these techniques for 2d data, using dgrid3d, to interpolate ungridded data so that it can be compared to other data and plotted on a grid. In many ways, my problem is quite similar, only the available interpolate functions aren't able to do it. I would refrain from suggesting gratuitous features, but it seems fairly simple and useful. For instance 'smooth step', it would make plotting "filled steps", a plotting style that would combine filledcurves and steps, far easier than some methods that have been proposed. Douglas 2011/3/18 Hans-Bernhard Bröker <HBB...@t-...> > On 18.03.2011 17:01, Douglas Mason wrote: > > [This is a question that came up when I was comparing some original data >> to an approximation which was computed through an external program. You >> can see the original plot in the attachment. My goal is to quantify the >> variation between the data and the approximation.] >> > > But why are you trying to use gnuplot for that? You need a data processing > tool, not a plotting program. > > When you want to compare data sets in a plotting program, you plot them > into a common diagram and look at it. For quantifying differences, you run > a data processor. > > As far as filledcurves is concerned, all it takes would be to transform > your data, adding the "extra" points introduced by "steps" or "fsteps". > > x1 y1 > x2 y2 > x3 y3 > > --> > > x1 y1 # "steps" > x2 y1 > x2 y2 > x3 y2 > x3 y3 > > |