Re: problem with multicore package

[Olaf Till]

On Sat, Jul 03, 2010 at 12:49:51AM +0200, Martijn wrote:
> Today I did my first calculations with the multicore package. Since I
> used only the two cores of my own laptop, the results were not too
> spectacular but nevertheless very welcome.
> 
> I am now trying to go a step further, but in the slave session I get the
> message: "Warning: Either variable 'functionHandle' or 'parameters' not
> existing after loading file parameter_xxx.mat"
> 
> A few lines from my script
> 
> F = @myfunc;
> <code for creating a structure 'p' of parameter structures>
> result = startmulticoremaster(F, p, '/tmp/mydir');
> 
> In the slave session(s) I just run "startmulticoreslave('/tmp/mydir')
> p{1} = 
> {
>   [1] = 25
>   [2] = 0
>   [3] = 12500
> }
> F(p{1}) gives the expected result
> 
> If I look in the parameter_xxx.mat file, I see all sane values for both
> the 'functionHandle' and the 'parameters'. 
> 
> What can be wrong here?
> 
> with kind regards,
> 
> Martijn


Sorry, thats no answer to your question, but a hint: since the
multicore package uses files for data exchange you should not expect
it is an ideal solution. If possible, you should look for
alternatives, e.g. Jaroslav Hajeks parcellfun in the general package.

Olaf