Confused by nonlinear least square fitting example

Confused by nonlinear least square fitting example

[Simon]

I'm a newbie to GSL so please be nice if this is a dumb question.

In the main part of the example program for the Levenberg-Marquardt
fitter the pointer to the parameters for the gsl_multifit_funtion_fdf f
appears to be pointing at the data rather than the parameters:
e.g. f.params = &d;
where d is a data struct.

I don't get this.  Surely it should point to a vector of guesses at the
parameters that fit the model?

Simon

-- 
*******************************************************************************
* Dr Simon Martin			s.martin@shef.ac.uk		     *
* Department of Physics and Astronomy
* University of Sheffield		tel: 0114 2224285
* Hounsfield Road			Fax: 0114 2728079
* Sheffield S3 7RH
* United Kingdom
******************************************************************************

Re: Confused by nonlinear least square fitting example

[Joakim Hove]

Simon <s.martin@sheffield.ac.uk> writes:

> I'm a newbie to GSL so please be nice if this is a dumb question.

So am I - but anyway.

> I don't get this.  Surely it should point to a vector of guesses at the
> parameters that fit the model?

That was also my first reaction, I guess it is due to different ideas
of nomenclature.

In the Levenberg-Marquardt functions (and the example) the word
"parameter" is used to indicate "general purpose" information about
the dataset in question, whereas "x" in the fitting functions (expb_f
& expb_df in the example) is a vector of the unknown coefficients, and
*not* the independent variable of the dataset (to add to the confusion
I find the word parameter suitable for these coefficients).

All in all I sum up the GSL nomenclature in this way:

  _Parameter_: Miscellaneous information about the dataset at hand.

  _x_ : A vector of the unknown coefficient which we want to
        determine.

Which is at least partly orthogonal to my intuitive understanding of
"param" and "x". Hopefully this did not only add to the confusion :-)


Joakim


-- 
==== Joakim Hove      www.phys.ntnu.no/~hove/   =======================
|| Institutt for fysikk  (735) 93637 / E3-166  |  Skøyensgate 10D    ||
|| N - 7491 Trondheim    hove@phys.ntnu.no     |  N - 7030 Trondheim ||
================================================= 73 93 31 68 =========

Confused by nonlinear least square fitting example

[Simon]

I'm a newbie to GSL so please be nice if this is a dumb question.

In the main part of the example program for the Levenberg-Marquardt
fitter the pointer to the parameters for the gsl_multifit_funtion_fdf f
appears to be pointing at the data rather than the parameters:
e.g. f.params = &d;
where d is a data struct.

I don't get this.  Surely it should point to a vector of guesses at the
parameters that fit the model?

Simon

-- 
*******************************************************************************
* Dr Simon Martin			s.martin@shef.ac.uk		     *
* Department of Physics and Astronomy
* University of Sheffield		tel: 0114 2224285
* Hounsfield Road			Fax: 0114 2728079
* Sheffield S3 7RH
* United Kingdom
******************************************************************************

Re: Confused by nonlinear least square fitting example

[Joakim Hove]

Simon <s.martin@sheffield.ac.uk> writes:

> I'm a newbie to GSL so please be nice if this is a dumb question.

So am I - but anyway.

> I don't get this.  Surely it should point to a vector of guesses at the
> parameters that fit the model?

That was also my first reaction, I guess it is due to different ideas
of nomenclature.

In the Levenberg-Marquardt functions (and the example) the word
"parameter" is used to indicate "general purpose" information about
the dataset in question, whereas "x" in the fitting functions (expb_f
& expb_df in the example) is a vector of the unknown coefficients, and
*not* the independent variable of the dataset (to add to the confusion
I find the word parameter suitable for these coefficients).

All in all I sum up the GSL nomenclature in this way:

  _Parameter_: Miscellaneous information about the dataset at hand.

  _x_ : A vector of the unknown coefficient which we want to
        determine.

Which is at least partly orthogonal to my intuitive understanding of
"param" and "x". Hopefully this did not only add to the confusion :-)


Joakim


-- 
==== Joakim Hove      www.phys.ntnu.no/~hove/   =======================
|| Institutt for fysikk  (735) 93637 / E3-166  |  Skøyensgate 10D    ||
|| N - 7491 Trondheim    hove@phys.ntnu.no     |  N - 7030 Trondheim ||
================================================= 73 93 31 68 =========