Gaussian peak with a left tail

[WladimirKazantz]

Ok, here one more problem

I imported some data and i want to fit a gaussian peak

That is not a problem. I used genfit with a gaussian distribution function and everything is fine.

But in some peaks I have left and in some others, right tails. I tried many ways by adding to the gaussian function others,

but the results are not good enough. Are ther any sugestions?

Also I would like to ask if it possible to add more fit functions, in Mathcad Prime.

Sometimes genfit take a lot of time to fix it, when the data are not complicated enough.

Thank you for your time.

[RichardJ]

But in some peaks I have left and in some others, right tails. I tried many ways by adding to the gaussian function others,

but the results are not good enough. Are ther any sugestions?

Could you post a worksheet with the problem data and the fit that is not good enough..

[WladimirKazantz]

Please do not pay attention on the second peak that is on the left of the big one.

If there is something you can't understand please let me explain.

[AlanStevens]

Might be simpler to use Given ... Minerr. See attached.

Alan

[RichardJ]

I agree 100% that it's better to use minerr than genfit. It's much easier to set up and gives identical results. However, note that if you minimize a vector of residuals in the solve block you should right click on minerr and force it to the L-M algorithm. The other algorithms are not appropriate in this case. In version 14 and 15 it defaults to this algorithm anyway, but that is not the cad in earlier versions.

[RichardJ]

The baseline is not going to compensate for that asymmetry, and I don't think your G function will either. There are asymmetric functions that might, but what is your end goal? What do you want to get out of the fit? For example, if the parameters of the Gaussian have some physical meaning, and that's what you want to get, there is little point is fitting with any function that is too far from a Gaussian!

[WladimirKazantz]

I am sorry for my too late reply, but i had problems with the net.

Ok, my goal is to measure the surface of my peak.

This is a result of a nuclear reaction and I want to know the number of counts.

This is why i am trying to fit it.

I don't care about the left peak. Let me be more specific.

There are 5 peaks, I want to fit only the last one at the right side, the bigger one.

If you see at the bottom of this, you can notice that there is a tail which can't be fitted by a normal distribution.

I know that this tail has a physical explanation, and obey on poisson distribution.

On the attached file you can notice that the tail i talking about is much more bigger.

[ValeryOchkov]

Wladimir Kazantzidis wrote:

Ok, here one more problem

I imported some data and i want to fit a gaussian peak

That is not a problem. I used genfit with a gaussian distribution function and everything is fine.

But in some peaks I have left and in some others, right tails. I tried many ways by adding to the gaussian function others,

but the results are not good enough. Are ther any sugestions?

Also I would like to ask if it possible to add more fit functions, in Mathcad Prime.

Sometimes genfit take a lot of time to fix it, when the data are not complicated enough.

Thank you for your time.

I think the result depends strongly on guess values.