inconsistency in fitting results (error of parameters) when using fit wizard with Custom functions

[fbuatier-cyber]

Hello I am an instructor in LabCourse at the physics department in Genova.
We tested intensively the fitting routine (fit Wizard) comparing the numerical results of a linear fit with those of Root, of a least squares routine, and Curve.fit . We compared Chi2 and the value of the fit parameters/uncertintise finding the following bug.

If we use fit_wizard + built-in function (polinomial, degree 1) the errors of the parameters are consistent withe the reference software, if we choose fit_wizard+Custom function (we build a linear function) instead the errors of the parameters are systematically smaller. Reading on the web it seems there is a wrong normalization of the parameters when choosing the Custom function option .

Is it possible to fix this bug? We would like to use the software with our students but in these conditions the inconsistency is too evident.
Best regards
Francesco Buatier

[narunlifescience]

Hi, Sorry! Unfortunately, I no longer develop the software due to time constraints. However, if you can point me to the specific piece of code where the issue occurs, I can try to assist you. Regards, Arun

…

On Wed, 1 Oct 2025 at 3:08 AM, fbuatier-cyber ***@***.***> wrote: *fbuatier-cyber* created an issue (narunlifescience/AlphaPlot#48) <#48> Hello I am an instructor in LabCourse at the physics department in Genova. We tested intensively the fitting routine (fit Wizard) comparing the numerical results of a linear fit with those of Root, of a least squares routine, and Curve.fit . We compared Chi2 and the value of the fit parameters/uncertintise finding the following bug. If we use fit_wizard + built-in function (polinomial, degree 1) the errors of the parameters are consistent withe the reference software, if we choose fit_wizard+Custom function (we build a linear function) instead the errors of the parameters are systematically smaller. Reading on the web it seems there is a wrong normalization of the parameters when choosing the Custom function option . Is it possible to fix this bug? We would like to use the software with our students but in these conditions the inconsistency is too evident. Best regards Francesco Buatier — Reply to this email directly, view it on GitHub <#48>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AC3FJTCYCAXLJFZH26HVSZT3VLZ4DAVCNFSM6AAAAACH6EOHRSVHI2DSMVQWIX3LMV43ASLTON2WKOZTGQ3TCMJUGQ3TCNA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[fbuatier-cyber]

Hi Arun thanks for providing feedback. The issues probably is not too complex. It is a pity because otherwise the program is working fine. Apparently the bug is in the fit_wizard+ when you choose a Custom function, rather than using the preset and built-in fit functions. In the case you use the Fit wizard+custom fit function, the fit works properly, but the results (chi^2 and error of the fit parameters are underestimated with respect to the fit performed with Fit Wizard+ built in functions. In our case we tested with a custom linear function compared to a built-in polinomial degree 1) , but it happens also Reading on some forum on the web it seems there is a wrong normalization of the parameters, and Chi^2 when choosing the Custom function option . Could you simply fix the way the error bars are calculate at the end of the fits? let me know idf I can assist further regards Francesco

________________________________ Da: Arun Narayanankutty ***@***.***> Inviato: domenica 2 novembre 2025 10:52 A: narunlifescience/AlphaPlot ***@***.***> Cc: fbuatier ***@***.***>; Author ***@***.***> Oggetto: Re: [narunlifescience/AlphaPlot] inconsistency in fitting results (error of parameters) when using fit wizard with Custom functions (Issue #48) [https://avatars.githubusercontent.com/u/11949260?s=20&v=4]narunlifescience left a comment (narunlifescience/AlphaPlot#48)<https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnarunlifescience%2FAlphaPlot%2Fissues%2F48%23issuecomment-3477808038&e=ed7a584b&h=1c8beb04&f=y&p=y> Hi, Sorry! Unfortunately, I no longer develop the software due to time constraints. However, if you can point me to the specific piece of code where the issue occurs, I can try to assist you. Regards, Arun

On Wed, 1 Oct 2025 at 3:08 AM, fbuatier-cyber ***@***.***> wrote: *fbuatier-cyber* created an issue (narunlifescience/AlphaPlot#48) <#48><https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnarunlifescience%2FAlphaPlot%2Fissues%2F48%26gt%3B&e=ed7a584b&h=1ff541ab&f=y&p=y> Hello I am an instructor in LabCourse at the physics department in Genova. We tested intensively the fitting routine (fit Wizard) comparing the numerical results of a linear fit with those of Root, of a least squares routine, and Curve.fit . We compared Chi2 and the value of the fit parameters/uncertintise finding the following bug. If we use fit_wizard + built-in function (polinomial, degree 1) the errors of the parameters are consistent withe the reference software, if we choose fit_wizard+Custom function (we build a linear function) instead the errors of the parameters are systematically smaller. Reading on the web it seems there is a wrong normalization of the parameters when choosing the Custom function option . Is it possible to fix this bug? We would like to use the software with our students but in these conditions the inconsistency is too evident. Best regards Francesco Buatier — Reply to this email directly, view it on GitHub <#48><https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnarunlifescience%2FAlphaPlot%2Fissues%2F48%26gt%3B&e=ed7a584b&h=1ff541ab&f=y&p=y>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AC3FJTCYCAXLJFZH26HVSZT3VLZ4DAVCNFSM6AAAAACH6EOHRSVHI2DSMVQWIX3LMV43ASLTON2WKOZTGQ3TCMJUGQ3TCNA><https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAC3FJTCYCAXLJFZH26HVSZT3VLZ4DAVCNFSM6AAAAACH6EOHRSVHI2DSMVQWIX3LMV43ASLTON2WKOZTGQ3TCMJUGQ3TCNA%26gt%3B&e=ed7a584b&h=4e1f38ed&f=y&p=y> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

— Reply to this email directly, view it on GitHub<https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnarunlifescience%2FAlphaPlot%2Fissues%2F48%23issuecomment-3477808038&e=ed7a584b&h=1c8beb04&f=y&p=y>, or unsubscribe<https://urlsand.esvalabs.com/?u=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FBYD5IHPIXP4QEZDHUYNPUYD32XH6LAVCNFSM6AAAAACH6EOHRSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTINZXHAYDQMBTHA&e=ed7a584b&h=1a778cf6&f=y&p=y>. You are receiving this because you authored the thread.Message ID: ***@***.***>