https://bugs.kde.org/show_bug.cgi?id=475408
Bug ID: 475408
Summary: Curve fitting on log axis needs extremely many
evaluated points
Classification: Applications
Product: LabPlot2
Version: 2.10.1
Platform: Archlinux
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: general
Assignee: alexander.se...@web.de
Reporter: nuutilaj...@gmail.com
Target Milestone: ---
SUMMARY
I am fitting a curve to my data, which contains 9 data points. Each data point
is separated on the x-axis by 1 order of magnitude (X 0.0001, 0.001, 0.01, 1,
etc.)
By default, the curve fitting tool will evaluate the curve for only the first
1000 or so points, drawing only the upper tip of my data. I have to manually
increase the evaluated points to 100 million to get my curve completely fitted,
at which point the program becomes quite sluggish.
I could of course take log10 of my x-axis values and plot and fit it on linear
axes to make a similar plot, but I'm interested in comparing my coefficients to
other publications.
STEPS TO REPRODUCE
1. Create quick test data with Octave notebook: x = [0.0001 0.001 0.01 0.1 1 10
100 1000 10000]; y = [12 9 15 35 70 85 91 89 94];
2. Plot the xy-curve, set X-Range of the plot area to log10 scale
3. Fit modified Hill equation to the xy-curve: a + (b-a) * ( (x^n) /
((x^n)+(K^n)) )
4. Gradually increase the evaluated points from fit options to 100 million to
get the whole curve plotted
5. Try modifying any of the parameters and coefficients or running the fit
OBSERVED RESULT
Modifying the fit or running the curve fitting tool becomes sluggish because it
needs to evaluate 100 million points.
EXPECTED RESULT
The fitting could be a bit snappier. Could the fitting tool evaluate points
more evenly/sparingly between the maximum and minimum x-axis values when the
axis is on a log scale?
SOFTWARE/OS VERSIONS
Linux/KDE Plasma:
KDE Plasma Version: 5.27.8
KDE Frameworks Version: 5.110.0
Qt Version: 5.15.11
ADDITIONAL INFORMATION
--
You are receiving this mail because:
You are watching all bug changes.
