how to smooth this curve

Share icon

In the attached experiment, the plotted curve is quite ugly. I tried to smooth the waves but I didn't improve it!

How can I smooth the curve?

Thanks a lot Experiment_7.pxp (808.53 KB)
It's a particle size distribution experiment. The waves are given by the device as float, but when I load them in Igor they are taken as integer (I think it's because the device uses a comma instead of a dot for floats). I've discussed with my advisor and he said that it's ok like this.
But I'm wonder how could I get rid of the small steps and make it more "continuous".

Thanks in advance.

March 12, 2015 at 02:16 am - Permalink

I think your problem is you have too many data points for your resolution - the data increase stepwise with many points describing the steps. I suggest you try reducing the number of points you have and then running a simple smooth. For example:
make/O/D/N=(DimSize(Counts_Interpl_smth,0)/200) Counts_Interpl_smth_dec, Size_Interpl_smth_dec
Counts_Interpl_smth_dec[]=Counts_Interpl_smth[p*200]
Size_Interpl_smth_dec[]=Size_Interpl_smth[p*200]
Duplicate/O Counts_Interpl_smth_dec,Counts_Interpl_smth_dec_smth
Smooth 11, Counts_Interpl_smth_dec_smth


Hope this helps,
Kurt

March 12, 2015 at 03:12 am - Permalink

Quote:
but when I load them in Igor they are taken as integer (I think it's because the device uses a comma instead of a dot for floats)


Igor can handle comma as decimal separator. Choose Data->Load Waves->Load Waves. In the resulting dialog, set all options to taste. Click the Tweaks button and choose Comma from the decimal separator popup menu. Click Return and then Do It.

March 12, 2015 at 08:03 am - Permalink

Hm... particle sizes... I have attached a copy of your experiment file with a LogNormal fit of the data. It's astonishingly good- is this fake data? :)

Anyway, I wouldn't alter the data- that's sort of like fraud, isn't it? But the fit is very good and the smooth fit curve can serve as a smooth representation of what the data is about. Just don't publish without saying that!

I have also attached a PNG picture of the graph so that everyone can see that it's a really good fit!

John Weeks
WaveMetrics, Inc.
support@wavemetrics.com
LogNormalFit.pxp (835.91 KB) LogNormalFit.png (18.26 KB)

March 12, 2015 at 01:13 pm - Permalink

I will use this last approach. Thanks for the replies guys!!

P.S.: They are not fake data :D

March 17, 2015 at 08:38 am - Permalink