Converting a p.d.f. for log(X) into a p.d.f. for X


I have a total p.d.f. which is the RooAddPdf of two RooKeysPdfs. I fit this total p.d.f. to a histogram of data as a function of LogX, which is just log(X), where X is another observable in my list.

After the fit, I can view the fit to the distribution of LogX and it looks fine.

Now I would like to plot X, and then show the total p.d.f. as a function of X plotted on the data points.

Plotting X is of course easy.

Plotting the p.d.f. seems a bit harder. I think one method is to recall:

P(x) = P(y) dy/dx for a p.d.f. P

I may define y = log(X) = LogX, so P(X) = P(log(X))/X

However, I cannot seem to write anything that returns P(X)

I have tried using RooClassFactory, but RooAbsPdf won’t accept another p.d.f. as input, for example.

I am hoping there is a simple way to do it, as it would be really useful. Any advice?

Thanks in advance,

~ Paras