Rendering of TF1

Dear Experts

Is there any possibility to improve the rendering of a TF1 generated from a TGraph object? When drawing the TF1 in the latest version of ROOT I have installed on my mac, I noticed that the precision of the drawing of the TF1 was reduced, I believe this is a problem of under-sampling, where you do not have enough points to realisticly reconstruct the function behaviour. Likely it is using only ~100 points or so. Is there a possibility to improve this manually?

  • blue = input data
  • magenta = TF1
  • red = TF1 evaluated in all x-positions of the input data

I attached a sample program that reproduces the issue and that you can run by root -l TF1Rendering.C

I recently upgraded my mac from 11.7 to 15.04 (I know it was long overdue). This forced me to reinstall root from source, and I noticed the rendering got (much) worse with my latest root w.r.t. the past. However I do not remember which was my version of gcc and clang in the previous configuration.

Thanks a lot
Kind regards
Piet Verwilligen
TF1Rendering.C (4.3 KB)


ROOT Version: 6.32.12
Platform: macosx64
Compiler: apple clang version 17.0.0


Hi Piet,

You can add following line to your script:

LumiFunction->SetNpx(1000);

It configures number of points used by TF1 for drawing.
100 is default value and does not changed since a while.

Regards,
Sergey