Prev: Call for Papers:Special Issue on Cognitive Information Processing
Next: image processing
From: Jan on 16 May 2010 14:24
Dear

I want to make a function of a variable, I have a string variable with an equation in it and I want to construct a function which I can use for lsqcurvefit.

I have this:
% x0 are my initial parameters
% ind is a column matrix with x and y values in it

equation = x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4); % = a*x^c+b*y^d
equation = strcat('@(x0,ind)',equation,',''x0'',''ind''');

param = lsqcurvefit(f,x0,ind,z,-100,100,options);

I got the following error:
??? Error using ==> inline.feval at 26
Too many inputs to inline function.

The problem is that matlab doesn't recognize equation as a function. Can somebody help me??

Thanks in advance!
Jan
From: Walter Roberson on 16 May 2010 15:43
Jan wrote:
> Dear
>
> I want to make a function of a variable, I have a string variable with
> an equation in it and I want to construct a function which I can use for
> lsqcurvefit.
>
> I have this:
> % x0 are my initial parameters
> % ind is a column matrix with x and y values in it
>
> equation = x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4); % = a*x^c+b*y^d
> equation = strcat('@(x0,ind)',equation,',''x0'',''ind''');
> param = lsqcurvefit(f,x0,ind,z,-100,100,options);
>
> I got the following error:
> ??? Error using ==> inline.feval at 26
> Too many inputs to inline function.

You do not show a definition for f, and you do not use equation in your
lsqcurvefit call?

Your second line for equation does not make sense to me. You start with
the text for an anonymous function declaration, but you leave no space
between the body and the string 'x0' that you insert at the end, and you
leave no space between that and the string 'ind' that you finish with.
Your line would thus look like

@(x0,ind)<SomethingHere>'x0''ind'

which, if evaluated, would be something with the string x0'ind
at the end of it.

You appear to be confusing anonymous functions and inline functions. If
you have an inline function, it should not have @ at the beginning of
it; if you have an anonymous function, you should need to evaluate the
definition of that function and pass the resulting function handle to
lsqcurvefit.

The <SomethingHere> that I indicated before is problematic. If x0 and
ind are defined before the first line in which you define equation, then
unless they are defined as symbolic, that first line is going to
evaluate to a fixed numeric value (even if some of the items in it are
strings), and that numeric value is not going to be compatible with
strcat unless strcat converts the numeric value to characters in the
range 0 to 255. If x0 and ind are defined and symbolic, strcat will not
be able to convert the resulting symbolic expression into a string. If,
though, x0 and ind are not defined before that line, then you will get
an execution time error because of the undefined variables.

What I _suspect_ you want is just

equation = @(x0,ind) x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4);

without a second line and without any strings or inline functions at all.
From: Jan on 17 May 2010 03:06
Hi

Thanks for your reply, but the difficult point is that I work with 2 figures: 1 main figure and a small figures that pops up and asks for an equation to be solved. When the equation is filled in, I write it to a textfile and then I should use that equation to be solved.

But when I manually set up an equation, the following code works perfect:
f = @(x0,ind)x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4),'x0','ind';
param = lsqcurvefit(f,x0,ind,z,-100,100,options);

To clear everything up:
x0 = [0 0 0 0]; % Initial parameters
ind = [x y]; % Column vectors of x and y datapoints

equation = x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4); % = a*x^c+b*y^d
equation = strcat('@(x0,ind)',equation,',''x0'',''ind''');
param = lsqcurvefit(equation,x0,ind,z,-100,100,options);

Can anyone help me?
Thanks in advance
From: Steven Lord on 17 May 2010 10:26

"Jan " <jan_neyens(a)skynet.be> wrote in message
news:hspd83$h42$1(a)fred.mathworks.com...
> Dear
>
> I want to make a function of a variable, I have a string variable with an
> equation in it and I want to construct a function which I can use for
> lsqcurvefit.
>
> I have this:
> % x0 are my initial parameters
> % ind is a column matrix with x and y values in it
>
> equation = x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4); % =
> a*x^c+b*y^d
> equation = strcat('@(x0,ind)',equation,',''x0'',''ind''');

If you're using a relatively recent version of MATLAB (release R2009a or
later), use STR2FUNC.

equation = 'x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4)';
equation = ['@(x0,ind) ',equation];
fh = str2func(equation);
fh(1:4, [1 2;3 4;5 6])

http://www.mathworks.com/access/helpdesk/help/techdoc/rn/bry1ecg-1.html#bry2cut-1

--
Steve Lord
slord(a)mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ


From: Walter Roberson on 17 May 2010 10:51
Jan wrote:

> Thanks for your reply, but the difficult point is that I work with 2
> figures: 1 main figure and a small figures that pops up and asks for an
> equation to be solved. When the equation is filled in, I write it to a
> textfile and then I should use that equation to be solved.

In that case, use eval() on the string to construct the function handle.


> But when I manually set up an equation, the following code works perfect:
> f = @(x0,ind)x0(1)*ind(:,1).^x0(2)+x0(3)*ind(:,2).^x0(4),'x0','ind';

I am having difficulty figuring out what that means. Ah... That is
interpreted as 3 different commands. The first command is the assignment
to f, the second command is 'x0' which displays the string x0 to the
display, and the third command is 'ind'; which computes the string ind
and then does not send it to the display because the semi-colon tells it
not to output results.

I would suggest that all in all, there is no point in having the
,'x0','ind' in that line.

The syntax for inline functions takes variable names after it, but the
syntax for anonymous functions does not.
 | 
Pages: 1
Prev: Call for Papers:Special Issue on Cognitive Information Processing
Next: image processing