Weird interpolation for vectors in different time base

[sdaau]

As I don't have the possibility right now to register an account at KDE Bug Tracking System, I'll try posting here. This is possibly not a bug, but a problem in my expectations - but I thought I'd report it anyway.

I generate some data using the following Perl script:

    $step = 0.01;

    open my $F1, '>/tmp/dat1.txt';
    open my $F2, '>/tmp/dat2.txt';

    for ($ix=0;$ix<100;$ix++) {
      $it = $ix * $step;
      if ($ix%3==0) {
        print {$F1} "$it $ix\n";
      } else {
        print {$F2} "$it\n";
      }
    }

    close $F1;
    close $F2;

`dat1.txt` represents some data with few samples, timestamps in column 1, and values in column 2; `dat2.txt` represents some different values in time I'd like to interpolate `dat1.txt` onto. As I had some problems with subtracting other data fitting this format, I originally thought that `kst` interpolates vectors sample by sample, as discussed in this post:

http://stackoverflow.com/questions/20448790/subtracting-two-interleaved-differently-based-time-series-arrays-with-numpy

... but actually it seems to be due to this "weird" result of interpolation. In the screenshot in the attachment below, you can see `dat1` with blue line/points; and the interpolation should be in red - I'd expect the interpolation to be the exact same blue line, just defined in different points - but you can see some distance between the lines; and further, the red line has a sort of a step behavior. In any case, I don't see why the red points should deviate from that line at all.

This is the procedure I used to generate the image: since due to

[https://bugs.launchpad.net/ubuntu/+source/kst/+bug/1258869 Bug #1258869 “KST cpu hog (100%) on Gnome natty/lucid " : Bugs : “kst” package : Ubuntu]

... I use Windows version of `kst`under `wine`, first I try to load up both data files:

    wine ~/.wine/dosdevices/c\:/Program\ Files/Kst\ 2.0.7/bin/kst2.exe /tmp/dat1.txt -x 1 -d -y 2 /tmp/dat2.txt -x 2

... this will only import dat1.txt and plot it (with points); so dat2.txt has to be imported manually, through the GUI:

# Click Create/Vector;
  # Name: uncheck Auto, name as "ts2"
  # Filename: /tmp/dat2.txt
  # Field or column: Column 1
  # Read to End + Ok
# Click Data Manager:
  # Double-click "Column 1" from dat1.txt, uncheck Auto for name, rename as "ts1"
  # Double-click "Column 2" from dat1.txt, uncheck Auto for name, rename as "val1"
  # Close Data Manager
# ts1, val1 should be 34 in length; ts2 should be 66 in length
# Click Create/Equation
  # Equation: type/choose [val1 (V2)]
  # X vector: ts2 (V3)
  # Check: Interpolate to highest resolution vector
  # From Appearance, choose Lines and Points; make color red
  # From Placement - place in existing plot
  # Click OK
# Right-click plot - Edit curve - val1 vs ts1
  # From Appearance - it's already on points, maybe choose another; add lines; make it blue
# File/Export as image(s)... (png)

Maybe I expect too much of the equation "[val1 (V2)]"/"ts2 (V3)" - but seeing "interpolate" there makes me think it would do a proper interpolation; and seemingly it doesn't. Please forgive (and explain) if it's not a bug :)
Cheers!