gnuplot-info — For questions and discussion about gnuplot

[Gnuplot-info] I need call gnuplot from Windows SO

[felipe r. <fel...@gm...>]

Hi Everybody,

I try run gnuplot from a fortran program in Windows SO. I have a file with
sentences in gnuplot, in the example proc.txt. In gnuplot run this file with
load("proc.txt"). I've managed to display the gnuplot terminal but not run the
file.

PROGRAM principal
CALL ESCRIBIR
END PROGRAM

SUBROUTINE ESCRIBIR
WRITE(*,*)'HOLA VAMOS A INTENTAR LLAMAR A GNUPLOT Y QUE PLOTEE ALGO'
CALL SYSTEM('gnuplot> load"proc.txt"')
!call system(gnuplot)
!read(*,*)
END SUBROUTINE ESCRIBIR

In the same folder I have the gnuplot.exe and executable program principal.
Please, I need help. Thanks very much!

Regards,



-- 
----------------------------------------------------------
FELIPE RODRÍGUEZ FONTE
ESTUDIANTE DE ING.INDUSTRIAL
UNIVERSIDAD DE LAS PALMAS DE G.C
fel...@gm...

Re: [Gnuplot-info] how to change sequence in the key?

[Hans-Bernhard B. <HBB...@t-...>]

On 25.05.2011 17:37, Jon wrote:

> I specify titles in the plot command, and let them be displayed in the key
> box. But I would like to change the sequence of the titles in the key box,
> say, the title to the second plot be shown as the first in the key box.

Hmmm.... and how would the result differ from that you can get just 
changing the sequence of datasets inside the 'plot' command?

[Gnuplot-info] how to change sequence in the key?

[Jon <dev...@gm...>]

Dear All,

I specify titles in the plot command, and let them be displayed in the key
box. But I would like to change the sequence of the titles in the key box,
say, the title to the second plot be shown as the first in the key box. I
looked through set key but don't find the relevant options, thus come over
to ask whether this is possible.

Thanks,

Sincerely,
Jon

Re: [Gnuplot-info] how to set arrow according to y2 axis?

[Jon <dev...@gm...>]

Thanks for your comment, I overlooked the first and second option in
coordinate explanation.

Sincerely,
Jon


2011/5/22 Hans-Bernhard Bröker <HBB...@t-...>

> On 22.05.2011 20:45, Jon wrote:
>
>  I would like to set arrow to label something which is drawn according to
>> y2
>> axis, thus if I can use coordinate y2 for locating arrow, that would be
>> very
>> convenient if I readjust coordinate of y2, which will make the plot shift
>> as
>> well.
>>
>
> Given that the information you're looking for is right there in the 3rd
> paragraph of "help set arrow", I really wonder how you might have managed to
> miss it when you looked there, before asking here...
>

Re: [Gnuplot-info] Multiple line in titles

[<mc...@gm...>]

Hello,

I have to add that I was having the problems about not being able to do line breaks only when using something like

splot 'file' u 1:2:3 w l title columnhead(1)

If instead I just set

set title "line 1\nline 2"

before, then it works.

But now I unfortunately ran into the next problem I was not able to solve: I use special characters and exponents, and then the two lines are too close to each other. Is there a way to increase the line spacing somehow? \n\n would be too much.

Thanks very much for your help, and best regards
Yukti

[Gnuplot-info] Multiple line in titles

[<mc...@gm...>]

Dear gnuplot community,

does anybody of you know how I can set my plot title to span multiple rows? It simply is too long for one row. Unfortunately, \n doesn't seem to work, and I already googled on that for quite some time with no success.

Thank you very much and best regards!
Yukti

Re: [Gnuplot-info] how to set arrow according to y2 axis?

[Thomas S. <t.s...@fz...>]

set arrow 1 from <x1>, second <y1> to <x2>, second <y2>

see
help coordinates


bsmile wrote:
> 
> Dear All,
> 
> I would like to set arrow to label something which is drawn according to
> y2
> axis, thus if I can use coordinate y2 for locating arrow, that would be
> very
> convenient if I readjust coordinate of y2, which will make the plot shift
> as
> well.
> 
> Thanks,
> 
> Sincerely,
> Jon
> ------------------------------------------------------------------------------
> What Every C/C++ and Fortran developer Should Know!
> Read this article and learn how Intel has extended the reach of its 
> next-generation tools to help Windows* and Linux* C/C++ and Fortran 
> developers boost performance applications - including clusters. 
> http://p.sf.net/sfu/intel-dev2devmay
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
> 
> 

-- 
View this message in context: http://old.nabble.com/how-to-set-arrow-according-to-y2-axis--tp31676651p31677601.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] How to plot a colored distribution map?

[Amir S. <ami...@gm...>]

Hi, have you tryied pm3d?

http://t16web.lanl.gov/Kawano/gnuplot/plotpm3d-e.html

Amir


On 23 May 2011 08:58, Xi Liu <jas...@gm...> wrote:

> Hi all:
> I got a 3 dimension data set whose x and y values both range from 1 to 200,
> z value varies between -5000 to 5000,  I want to plot a 2d colored
> distribution map drawing 40000 points with different color depending on
> their z value, for example the point with -5000 get pure white, and 5000
> get
> pure black, and the color will change smoothly from . So I can tell where
> the most black area goes during plotting several consecutive date sets and
> decide which x,y pair I should use in the future.
> But I don't find corresponding command to solve my problem, so I use the
> following commands:
>
> set palette model RGB defined (0 'white', 1 'black')
> set view map
> splot "file"  u 1:2:3 with points palette
>
> a stupid solution, draw the 3d picture and look down from above. this kind
> of solve my problem except the picture are not that smooth and points are
> cross-shaped. I think I must have missed some important command in gnuplot,
> would someone give me a hint how to solve this in a gnuplot way?
>
> and BTW, when I open there pipes to three gnuplot processes  issuing
>  commands plotting the pictures I described above, the whole process takes
> about five seconds to draw the whole 3 pics. Is that a little abnormally
> slow?
>
> Thanks for your time on my stupid question.
>
> ------------------------------------------------------------------------------
> What Every C/C++ and Fortran developer Should Know!
> Read this article and learn how Intel has extended the reach of its
> next-generation tools to help Windows* and Linux* C/C++ and Fortran
> developers boost performance applications - including clusters.
> http://p.sf.net/sfu/intel-dev2devmay
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>



-- 
**************************************************************
Amir Sabbagh
Il mio sito / My site: http://almucantarat.altervista.org
**************************************************************

[Gnuplot-info] How to plot a colored distribution map?

[Xi L. <jas...@gm...>]

Hi all:
I got a 3 dimension data set whose x and y values both range from 1 to 200,
z value varies between -5000 to 5000,  I want to plot a 2d colored
distribution map drawing 40000 points with different color depending on
their z value, for example the point with -5000 get pure white, and 5000 get
pure black, and the color will change smoothly from . So I can tell where
the most black area goes during plotting several consecutive date sets and
decide which x,y pair I should use in the future.
But I don't find corresponding command to solve my problem, so I use the
following commands:

set palette model RGB defined (0 'white', 1 'black')
set view map
splot "file"  u 1:2:3 with points palette

a stupid solution, draw the 3d picture and look down from above. this kind
of solve my problem except the picture are not that smooth and points are
cross-shaped. I think I must have missed some important command in gnuplot,
would someone give me a hint how to solve this in a gnuplot way?

and BTW, when I open there pipes to three gnuplot processes  issuing
 commands plotting the pictures I described above, the whole process takes
about five seconds to draw the whole 3 pics. Is that a little abnormally
slow?

Thanks for your time on my stupid question.

Re: [Gnuplot-info] how to set arrow according to y2 axis?

[Hans-Bernhard B. <HBB...@t-...>]

On 22.05.2011 20:45, Jon wrote:

> I would like to set arrow to label something which is drawn according to y2
> axis, thus if I can use coordinate y2 for locating arrow, that would be very
> convenient if I readjust coordinate of y2, which will make the plot shift as
> well.

Given that the information you're looking for is right there in the 3rd 
paragraph of "help set arrow", I really wonder how you might have 
managed to miss it when you looked there, before asking here...

Re: [Gnuplot-info] Symbol font for pngcairo

[<rhu...@ih...>]

On Sun, 22 May 2011 10:59:31 +0200
Nunzio Losacco <nun...@gm...> wrote:

> Hi all,
> 
> I'm really not happy with the default Symbol font used by Gnuplot on my
> linux system. That is an ugly Sans greek font while I'd like some Roman
> Greek font.
> I've found a "CM Roman Greek" font on my system through fc-list. but
> apparently gnuplot doesn't understand it:
> 
> if I do:
> 
> term pngcairo ...... font "CM Roman"
> 
> it uses the correct Roman font, but if I do
> 
> term pngcairo ...... font "CM Roman Greek" it reverts to using the default
> Sans (no greek!!) font.
> 
> What am I missing?

Say, which version do you use?
Have you tried enhanced syntax?
Does help gd help?

[Gnuplot-info] how to set arrow according to y2 axis?

[Jon <dev...@gm...>]

Dear All,

I would like to set arrow to label something which is drawn according to y2
axis, thus if I can use coordinate y2 for locating arrow, that would be very
convenient if I readjust coordinate of y2, which will make the plot shift as
well.

Thanks,

Sincerely,
Jon

[Gnuplot-info] Symbol font for pngcairo

[Nunzio L. <nun...@gm...>]

Hi all,

I'm really not happy with the default Symbol font used by Gnuplot on my
linux system. That is an ugly Sans greek font while I'd like some Roman
Greek font.
I've found a "CM Roman Greek" font on my system through fc-list. but
apparently gnuplot doesn't understand it:

if I do:

term pngcairo ...... font "CM Roman"

it uses the correct Roman font, but if I do

term pngcairo ...... font "CM Roman Greek" it reverts to using the default
Sans (no greek!!) font.

What am I missing?

Thanks very much for your help.

NL

Re: [Gnuplot-info] Palette boundary colors

[Thomas S. <t.s...@fz...>]

you are searching for
set cbrange


SiSkin wrote:
> 
> Good time of day!
> 
> I'm new to gnuplot. I'm dealing with an evolutionary process and have a
> bunch of gnuplot data files corresponding to different time steps. I
> output them into png-files and make a video animation. The problem is that
> two boundary colors are always subject to current min and max, so the same
> color does not remain indicating the same value throughout the process. I
> tried to use
> 
> set palette defined (0 "blue", 1 "red", 2 "yellow")
> 
> but the issue remains, as seen from the picture (which is supposed to be
> all in shades of red):
> 
>  http://old.nabble.com/file/p31665935/1.png 
> 
> How do I set the lower and upper boundaries for the palette instead of
> having them set automatically? I beg your pardon if my question has
> already been answered but I failed to find a solution.
> 
> Thank you for your help!
> 
> 

-- 
View this message in context: http://old.nabble.com/Palette-boundary-colors-tp31665935p31671685.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] Plot value from another column above histogram bar

[dave101 <da...@ya...>]

You code works perfectly.  Thank you!

-- 
View this message in context: http://old.nabble.com/Plot-value-from-another-column-above-histogram-bar-tp31652478p31664360.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] Plot value from another column above histogram bar

[dave101 <da...@ya...>]

Thank you Thomas.  

Looks like I need to upgrade to 4.4 to use the for loop syntax.  My upgrade
installs are not going well, so it may be a bit before I can test it...
-- 
View this message in context: http://old.nabble.com/Plot-value-from-another-column-above-histogram-bar-tp31652478p31656013.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] Negative logscale axis

[Tue B. <al...@gm...>]

Thank you very much for the fast and helpfull reply!

The trick was to use the two in conjunction and then manually set a minus
sign on the axis with set format.

Below is the working code:


#!/usr/bin/python
from numpy import *
import Gnuplot, Gnuplot.funcutils
import os, sys, string
import glob
import struct as st
import math
import time


for datafile in glob.glob('*.txt' ):
    f = open(datafile, 'r')
    data = map(string.split, f.readlines())
    g = Gnuplot.Gnuplot(debug=1)
    g.title('Plot of generation ') # (optional)
    g('set xrange[1:102]')
    g('set yrange [0.01:10] reverse')
    g('set format y "-%.3f"')
    g('set logscale xy')
    g('set terminal gif')
    g('set output "'+str(datafile[:-4])+'.gif"')
    g('y(r)=r0+z/r')
    g('r0=0')
    g('z=1')
    g('fit [10:100] y(x) "'+str(datafile)+'" using 1:(-$2) via r0,z')
    g('plot "'+str(datafile)+'" using 1:(-$2), y(x)')


2011/5/19 Hans-Bernhard Bröker <HBB...@t-...>

> On 19.05.2011 13:04, alyflex wrote:
>
>  Hi, I got some data that I want to plot on a double logarithmic scale, but
>> since all the y values are negative, I need a negative logarithmic scale,
>> but I have thus far been unable to make one. Does anyone know how to do
>> this?
>>
>
> Negate the data on creation or on-the-fly (".. using 1:(-$2) .."), and
> revert the y axis ("set yrange reverse").
>
>

Re: [Gnuplot-info] Negative logscale axis

[Hans-Bernhard B. <HBB...@t-...>]

On 19.05.2011 13:04, alyflex wrote:

> Hi, I got some data that I want to plot on a double logarithmic scale, but
> since all the y values are negative, I need a negative logarithmic scale,
> but I have thus far been unable to make one. Does anyone know how to do
> this?

Negate the data on creation or on-the-fly (".. using 1:(-$2) .."), and 
revert the y axis ("set yrange reverse").

[Gnuplot-info] Negative logscale axis

[alyflex <al...@gm...>]

Hi, I got some data that I want to plot on a double logarithmic scale, but
since all the y values are negative, I need a negative logarithmic scale,
but I have thus far been unable to make one. Does anyone know how to do
this?

The python script I'm using is the following:

from numpy import *
import Gnuplot, Gnuplot.funcutils
import os, sys, string
import glob
import struct as st
import math
import time

for datafile in glob.glob('*.txt' ):
    g.title('Plot')
    g('set xrange[0.1:100]')
    g('set yrange[-10:-1]')
    g('set terminal gif')
    g('set logscale xy')
    g('set output "'+str(datafile[:-4])+'.gif"')
    g('y(r)=r0+z/r')
    g('r0=0')
    g('z=1')
    g('fit [15:100] y(x) "'+str(datafile)+'" using 1:2 via r0, z')
    g('plot "'str(datafile)+" using 1:2, y(x)')    


And the datafiles look like this:

  0.000000000000000E+000  0.000000000000000E+000
  9.765625000000000E-002  -2.05577406839801     
  0.195312500000000       -1.80843071757366     
  0.292968750000000       -1.59497223855229     
  0.390625000000000       -1.41038643815402     
  0.488281250000000       -1.24415910487246     
  0.585937500000000       -1.09103728309075     
  0.683593750000000      -0.948432020417034     
  0.781250000000000      -0.814990671718170     
  0.878906250000000      -0.690058106734278     
  0.976562500000000      -0.573781621676318     
   1.07421875000000      -0.468133842918065     
   1.17187500000000      -0.380214853165435     
   1.26953125000000      -0.327276191990924     
   1.36718750000000      -0.322127989028048     
   1.46484375000000      -0.347499023048631     
   1.56250000000000      -0.383040585116026     
   1.66015625000000      -0.418903776691791     
   1.75781250000000      -0.451056613640834     
   1.85546875000000      -0.478246521157531     
   1.95312500000000      -0.500379189881943     
   2.05078125000000      -0.517785215206294     
   2.14843750000000      -0.530920312130199     
   2.24609375000000      -0.540249563775338     
   2.34375000000000      -0.546208272940876     
   2.44140625000000      -0.549192158402245     
   2.53906250000000      -0.549557488932987     
   2.63671875000000      -0.547623221032223     
   2.73437500000000      -0.543674138702906     
   2.83203125000000      -0.537964470887115     
   2.92968750000000      -0.530721834409276     
   3.02734375000000      -0.522150927843818     
   3.12500000000000      -0.512436786265136     
   3.22265625000000      -0.501747552410168     
   3.32031250000000      -0.490236961648138     
   3.41796875000000      -0.478046512570940     
   3.51562500000000      -0.465307507703403     
   3.61328125000000      -0.452142758785914     
   3.71093750000000      -0.438668278702490     
   3.80859375000000      -0.424994625544276     
   3.90625000000000      -0.411228394904050     
   4.00390625000000      -0.397473277288847     
   4.10156250000000      -0.383831372110201     
   4.19921875000000      -0.370403910478473     
   4.29687500000000      -0.357292281706303     
   4.39453125000000      -0.344598174399981     
   4.49218750000000      -0.332423974672431     
   4.58984375000000      -0.320871802060817     
   4.68750000000000      -0.310042639379781     
...



-- 
View this message in context: http://old.nabble.com/Negative-logscale-axis-tp31654923p31654923.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] Plot value from another column above histogram bar

[Thomas S. <t.s...@fz...>]

does the following script produce the plot you want?
(change 'nw' for the actual number of weeks)

data:

data   R1   S1   R2  S2   R3  S3   R4  S4   R5  S5
line1  1000 0.1  900 0.2  800 0.3  700 0.4  600 0.5
line2  900  0.1  800 0.2  700 0.3  600 0.4  500 0.5
line3  800  0.1  700 0.2  600 0.3  500 0.4  400 0.5
line4  700  0.1  600 0.2  500 0.3  400 0.4  300 0.5
line5  600  0.1  500 0.2  400 0.3  300 0.4  200 0.5

script:

set macros
nw=5 ; # number of weeks
gs="1" ; # gap between clusters
b=1./(nw+@gs)
o=-0.5*(nw-1)*b
w = nw-1
set xrange [0:6]
set yrange [0:1050]
set style histogram clustered gap @gs
set style data histogram
plot \
   for [i=0:w] 'data.txt' using (column(2+i*2)):xticlabels(1) title
columnheader,\
   for [i=0:w] '' using ($0+1):(column(2+i*2)):(stringcolumn(3+i*2)) \
      every ::1 \
      with labels center offset first o+i*b, graph 0.01 notitle


dave101 wrote:
> 
> Say I have 5 manufacturing lines.  Each week, I collect 2 quality metrics
> R and S.  After week 2, I have a table like below.  I would like to graph
> a clustered histogram chart using the data from the R# columns and then
> put the number from the corresponding S# column above each bar.
> 
> So far I have below which can plot the S# points, but i can't make it just
> put the S numbers and all of the S# points get put on the first bar in the
> cluster, not on each bar in the cluster.  (i.e.  I need to use a separate
> xtic for each S#.
> 
> I need the solution to be scalable -- each week I will add 2 new columns
> for R and S.  So I need to be able to calculate the xtics somehow based on
> number of columns?
> 
> plot "data.txt" using 2:xticlabels(1) ti col , '' using 4 ti col, '' using
> 3 ti col axis x1y2 with points, '' using 5 ti col axis x1y2 with points
> 
> data    R1      S1      R2      S2
> line1   1000    0.2     900     0.1
> line2   900     0.2     800     0.1
> line3   800     0.2     700     0.1
> line4   700     0.2     600     0.1
> line5   600     0.2     500     0.1
> 
> 

-- 
View this message in context: http://old.nabble.com/Plot-value-from-another-column-above-histogram-bar-tp31652478p31654616.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

[Gnuplot-info] Plot value from another column above histogram bar

[dave101 <da...@ya...>]

Say I have 5 manufacturing lines.  Each week, I collect 2 quality metrics R
and S.  After week 2, I have a table like below.  I would like to graph a
clustered histogram chart using the data from the R# columns and then put
the number from the corresponding S# column above each bar.

So far I have below which can plot the S# points, but i can't make it just
put the S numbers and all of the S# points get put on the first bar in the
cluster, not on each bar in the cluster.  (i.e.  I need to use a separate
xtic for each S#.

I need the solution to be scalable -- each week I will add 2 new columns for
R and S.  So I need to be able to calculate the xtics somehow based on
number of columns?

plot "data.txt" using 2:xticlabels(1) ti col , '' using 4 ti col, '' using 3
ti col axis x1y2 with points, '' using 5 ti col axis x1y2 with points

data    R1      S1      R2      S2
line1   1000    0.2     900     0.1
line2   900     0.2     800     0.1
line3   800     0.2     700     0.1
line4   700     0.2     600     0.1
line5   600     0.2     500     0.1

-- 
View this message in context: http://old.nabble.com/Plot-value-from-another-column-above-histogram-bar-tp31652478p31652478.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

[Gnuplot-info] Plotting non-continuous data using Lines

[Daniel E. <nuc...@gm...>]

Hi all,

I am attempting to plot a series of train times on a graph, with time on the
x-axis, and location on the y-axis. The plot works fine for trains
travelling through all the locations listed, but GNUPlot doesn't like trains
which start at an intermediate location, or trains where there is only a
passing time, rather than an arrival/departure time. Also, is there a way to
class the Y axis by labels, rather than by numbers? I've currently had to
throw in the "Stn No" column to get it to work.

My plotting code is:
set title "Timetable D"
set style data lines
set xlabel "Time"
set timefmt "%H:%M"
set xdata time
set xrange ["08:00":"20:00"]
set format x "%H:%M"
set ylabel "Location"
set grid
set key left
plot for [i=3:4] 'SVRTableD.dat' using i:2:yticlabels(1) title column(i)

A couple of lines of example data are:
Station    Stn No    1055 BH    1000 BY
KR    1    12:05
BY    2    11:54
BY    2    11:51    10:00
AY    3    11:38
AY    3    11:35    10:12
HY    4    11:27
HY    4    11:24    10:19
HL    5    11:16    10:26
HL    5    11:10    10:31
BH    6    10:55    10:46

The 1055 BH plots correctly, however, the 1000 BY line starts itself at KR,
as well as giving strange results for stations where there is no arrival
time (the train does not stop at a number of locations). The latter could be
fixed by putting in dummy arrival times that are the same as the departure
one, but I'd rather not have to edit all my data to do that!

Regards,
Daniel

Re: [Gnuplot-info] How to draw a vertical line

[pban92 <pb...@ya...>]

Thank you Thomas! That's what I did as the last resort.



pban92 wrote:
> 
> I failed with set parametric to draw a vertical line at x = 100. Any one
> pls? Thank you in advance! http://old.nabble.com/file/p31622738/vert.jpg 
> 

-- 
View this message in context: http://old.nabble.com/How-to-draw-a-vertical-line-tp31622738p31646928.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

[Gnuplot-info] The best way to deal with excel columns with eneven width for gnuplot

[pban92 <pb...@ya...>]

Whenever I copy excel columns to text file (like Notepad or Notepad++), I
need to make sure that the columns have fix widths. Otherwise multiple
columns will look terribly messed up once exported. Can anybody pls tell me
how to avoid this sickening situation without manual modification on the
columns? Saving as csv file did not work as well! 

For example, when I want to plot 1:5 or 6:10 on excel, I see a nice
plotting. But when I copy paste the columns to text file it becomes ugly.
Here is the file:  http://old.nabble.com/file/p31646665/testcopy.xls
testcopy.xls  

Many thanks for solving my years' long problem!
-- 
View this message in context: http://old.nabble.com/The-best-way-to-deal-with-excel-columns-with-eneven-width-for-gnuplot-tp31646665p31646665.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] Help request - linear regression giving screwy results

[Eric R. <es...@th...>]

Thomas Sefzick <t.s...@fz...>:
> to make the fit algorithm happy, apply a factor of e.g. 1.e-7 to 'm'
> 
> f(x) = 1.e-7 * m *x + b
> 
> and the fit algorithm will converge after a few iterations because
> the effect of a step in 'm' or 'b' on the deviation of the regression
> function from the data is now nearly equal.
> 
> not necessary, but also helpful: shift the origin of the regression
> function into the x-range defined by the data points (=subtract
> 10 years from 'x'), this makes the fit algorithm faster.

OK, not a floating-point misconvergence but something more subtle.  
Let's see if I can recast this into a FAQ emtry.  Criticize, please.

-----------------------------------------------------------------------------
Q: I'm using time xdata and my curve-fit produces absurd results. How do I 
fix this?

A: Change the coefficients on your fit function so that incrementing any
of them produces a difference in the value of y that is about the same as
incrementing any other.

Say, for example, that you are trying a linear fit on some time-based 
data that looks like this:

2009-12-00 5.2
2011-03-00 34.7

Because dates in gnuplot are calculated as 'seconds since beginning
of the gnuplot epoch' which is '2000-01-01 00:00:00', your data
actually looks to the curve-fitter like this:

312940800.0 5.2
352252800.0 34.7

It's the large difference in magnitude between x and y values that causes 
the problem.  The fit algorithm will make small steps in the 1.e-8 range
when varying 'm', but large steps in the 1.e-1 range when varying
'b'. This difference in step size makes it (nearly) impossible for the
fit algorithm to converge properly, so it will end up with a regression
function which crosses your data points but with a wrong slope.

To fix this, apply a scale factor to the m coefficient that makes values of
mx comparable in magnitude to values of y.  In this case fitting to 
f(x) = 1.e-7 * m *x + b will work much better.

Shifting the origin of the regression function to inside the x-range defined 
by the data points is not necessary for convergence, but it will make the fit 
algorithm run faster.  In this case, try subtracting 10 years, like so:

offset=10*365*24*60*60
f(x)=1.e-7*m*(x-offset)+b

The fix (and the optimization) generalizes to nonlinear fit functions.
The curve-fitting algorithm is going to walk through tuples of
coefficients looking for a tuples with small residuals.  If there are
large enough variations in the magnitude of changes in y as different
coefficience are tweaked, that search may converge on a wrong tuple.
To fix this, tweak the function so the search has roughly equal grain
in all directions.
-----------------------------------------------------------------------------

-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>