[Help-gsl] Help with multifit non-linear Least Squares

[petmakris]

Dear everybody,

I am new to gsl and I am trying to use the functions from
gsl_multifit_nlin.h with some data. (If i'm saying something totaly wrong
plz ignore me). I have experimental y[i] and t[i] data for the equation

y= ( t^2 ) / ( t-f )              [1]

which is a rotated hyperbola. I am using aliased the example from the
documentation of gsl, and because i've never done fiting before, could
somebody help me on: is this a fitable function?  I can see the example with
the exponential but (ahem i don't what to do).

----------------------------------------------------------------------
If you wan't to help me somehow you can execute (in bash shell) the
following command and you'll see the program output at once (and where the
problem is)
----------------------------------------------------------------------
wget http://finsler.math.uoa.gr/~petmakris/publish/3.3.tar.gz \
 && tar -xvzf 3.3.tar.gz && cd 3.3/c && make run
----------------------------------------------------------------------


The error output is:


iter:   0 f==     1.00000000  |f(x)| = 4021.56
status = success
iter:   1 f==     6.02518738  |f(x)| = 3691.04
-----------------------------------------------
status = success
iter:   2 f==    16.07556213  |f(x)| = 1806.9
-----------------------------------------------
status = success
iter:   3 f==    17.42672993  |f(x)| = 1065.53
-----------------------------------------------
status = success
iter:   4 f==    18.61374006  |f(x)| = 595.905
-----------------------------------------------
status = success
iter:   5 f==    18.41882458  |f(x)| = 534.914
-----------------------------------------------
status = success
iter:   6 f==    18.40383771  |f(x)| = 534.604
-----------------------------------------------
status = success
iter:   7 f==    18.40348631  |f(x)| = 534.603
-----------------------------------------------
7.01149e-07
gsl: matrix_source.c:31: ERROR: first index out of range
Default GSL error handler invoked.




Thanks in advance,
Petros Makris