gnuplot-info

[Gnuplot-info] GNUPLOT: PARAMETERS AND ERRORS

[<ema...@al...>]

I'm Emanuele Di Gloria, a Ubuntu user.=20
I want use gnuplot for my work. I noticed that after every iteration, =
the program gives a value for every parameter used for my function.
For example, i use the function f(x)=3D A+(B*x), with A=3D0 and B=3D3, =
and I have this file, from which I can read the data;


1.0000    3.0000   0   0.002
2.0000    6.0000   0   0.004
3.0000    9.0000   0   0.003
4.0000    12.000   0   0.002
5.0000    15.000   0   0.001
6.0000    18.000   0   0.004
7.0000    21.000   0   0.003
8.0000    24.000   0   0.002
9.0000    27.000   0   0.006

After "interation 5", the program gives=20

A=3D 1,0023e-15
B=3D 1,574e-26  =20

the numbers are invented). What do this two values mean? If they are not =
errors, how can I find errors on A and B? Can you help me?   Thanks.

[Gnuplot-info] GNUPLOT: PARAMETERS AND ERRORS

[<ema...@al...>]

I'm Emanuele Di Gloria, a Ubuntu user.
I want use gnuplot for my work. I noticed that after every iteration, =
the program gives a value for every parameter used for my function.
For example, i use the function f(x)=3D A+(B*x), with A=3D0 and B=3D3, =
and I have this file, from which I can read the data;


1.0000    3.0000   0   0.002
2.0000    6.0000   0   0.004
3.0000    9.0000   0   0.003
4.0000    12.000   0   0.002
5.0000    15.000   0   0.001
6.0000    18.000   0   0.004
7.0000    21.000   0   0.003
8.0000    24.000   0   0.002
9.0000    27.000   0   0.006

After "interation 5", the program gives

A=3D 1,0023e-15
B=3D 1,574e-26 =20

the numbers are invented). What do this two values mean? If they are not =
errors, how can I find errors on A and B? Can you help me?   Thanks.

Re: [Gnuplot-info] GNUPLOT: PARAMETERS AND ERRORS

[<HBB...@t-...>]

ema...@al... wrote:

> I want use gnuplot for my work. I noticed that after every iteration, 
> the program gives a value for every parameter used for my function.
[...]
> After "interation 5", the program gives

> A= 1,0023e-15
> B= 1,574e-26  

> the numbers are invented). What do this two values mean?

I really wonder how that could fail to be obvious.  Those are the values 
for the fitted parameters 'A' and 'B' that are the program's current 
shot at the solution.  It's called an iteration because it walks through 
several such attempts to find the best one.

 > If they are not errors, how can I find errors on A and B?

For single iterations you can't, because gnuplot doesn't compute errors 
every time.  The errors on A and B are printed at the end, for the final 
solution.

Re: [Gnuplot-info] GNUPLOT: PARAMETERS AND ERRORS

[Lars H. <la...@ho...>]

> I want use gnuplot for my work. I noticed that after
> every iteration, the program gives a value for every
> parameter used for my function. For example, i use the
> function f(x)= A+(B*x), with A=0 and B=3, and I have this
> file, from which I can read the data;
>
> [8-< snip]
>
> After "interation 5", the program gives
>
> A= 1,0023e-15
> B= 1,574e-26
>
> the numbers are invented). What do this two values mean?
> If they are not errors, how can I find errors on A and B?
> Can you help me?   Thanks.

Unfortuanetly you provide only very few informations e.g. 
about which commands you are using or your version of 
gnuplot. I assume, that you are using the fit-command, at 
least your interpolation problem indicates that.

I suppose, you've already studied the help (help fit) very 
well. A and B are the free parameters, that gnuplot uses to 
fit your given function f(x) to the points of your dataset. 
The algorithm, that is working in the backgound, needs not 
only a function with these free paramaters, but also an 
'idea' of the ranges these parameters could lie in. So my 
advice is to give gnuplot suitable initial values (the 
numbers are invented):

f(x)= A+(B*x)
A=26.5
B=-6.5
fit f(x) 'datafile' via A,B

Now gnuplot should be able to iterate the fit values for A 
and B within only a few steps. The iteration process stops, 
when the weighed sum of the squared residuals becomes 
smaller than a fixed (very small) value. Then gnuplot 
prints the parametres with their standard errors to the 
screen. 

Did I get your problem right? If not, please write some more 
details.
Regards
Lars Hoegen

P.S.: We're not deaf, so please stop shouting.