Deconvolution y vs x. Positive and negative values

Hello!

Do you know any algorithm to deconvolve signal, which is sum of overlapped individual signals in case of response with positive and negative values?
For example, I have PMT.
PMT generate gauss signal.
But my amplifier not ideal and distort pulse shape.
I know response for amplifier.

If response has only one sign of data (i.e. all values are positive or all values are negative), I can use TSpectrum::Deconvolution.
But this algorithm does not work if my data has positive and negative values.
This situation very common for physics, because charge-sensitive amplifiers give response with both signs.

Here hepunx.rl.ac.uk/~adye/software/u … nfold.html I found overview of several algorithms, but they work with PDFs, which are positive by definition.

I will be very grateful for any information.
Thank you in advance.

Best regards, Vladislav.

ROOT provides this root.cern.ch/doc/master/group__Unfold.html

Hello!

I have contacted with Stefan Schmitt (author of TUnfold) and he told, that it is possible in principle.
I have not checked yet.

Does root cern have realisation of Wiener deconvolution en.wikipedia.org/wiki/Wiener_deconvolution ?
It is simple to realize, but may be there is ready code.

Best regards, Vladislav.