fminunc with a one-variable function

[Jonathan Stickel]

Jaroslav Hajek wrote:
> On Fri, Jul 24, 2009 at 7:34 PM, Jonathan Stickel<jjstickel at vcn.com> wrote:
>>
>> Jaroslav Hajek wrote:
>>> On Thu, Jul 23, 2009 at 5:32 PM, Jonathan Stickel<jjstickel at vcn.com>
>>> wrote:
>>>> I see the following when I use fminunc to minimize a function of just one
>>>> variable:
>>>>
>>>>  function F = theF(x)
>>>>  F = x.^2 + exp(x);
>>>>  endfunction
>>>>  x0 = 1;
>>>>  [xmin,fmin] = fminunc(@theF,x0)
>>>>
>>>> error: Invalid call to cholupdate.  Correct usage is:
>>>>
>>>>  -- Loadable Function: [R1, INFO] = cholupdate (R, U, OP)
>>>>
>>>>
>>>> I know that there are other minimization functions that can be used for
>>>> single-variable functions, but fminunc should work too.
>>>>
>>>> Thanks,
>>>> Jonathan
>>>>
>>> Please try
>>> http://hg.savannah.gnu.org/hgweb/octave/rev/5fcf008c7c41
>>>
>> OK, I managed to compile chol.cc (thanks to jwe) from the indicated
>> changeset and put it in my path:
>>
>> octave:6> which chol
>> `chol' is a function from the file
>> /Users/jstickel/programming/octave/local/chol.oct
>>
>> However, I still see the same error as reported above.
>>
> 
> Of course you do. The error was in cholupdate et al., so you need also
> to override the autoloads. Or just replace the chol.oct file in
> Octave's directory.
> 

OK, it works now (after replacing the installed chol.oct).

What is the algorithm used in this version of fminunc?  It might be nice 
to mention in the help text.

Thanks,
Jonathan