[ADMB Users] separable runs slower in Catch-at-Age model

[Thorvaldur Gunnlaugsson]

Thanks Hans for your response.

I have now identified what has probably been my main problem in running 
the RE programs.
When parameters get active they loose the values assigned in the 
PRELIMINARY section
and get a zero or mid point of range value, which then drastically 
changes the fit obtained.

Another strange thing is that if I try printing the value of the
   objective_function_value g;
in a SEPARABLE_FUNCTION
it has the value zero there and is actually set to zero
but then returns what it is assigned in the function
so it is not like a local variable.
If the function is not separable this does not happen.

Both of this can be seen immediately in:

http://www.hafro.is/~thg/langtn/re/care/caresnr.tpl

 > caresnr |& more

ThG

On 11/23/09 10:08, H. Skaug wrote:
> Hi,
>
> The user manual states that you should use SEPARABLE_FUNCTIONS in this
> situation, not FUNCTION.
> I have no idea what happens if you use FUNCTION.
>
> Hans
>
>
> On Fri, Nov 20, 2009 at 2:37 PM, Thorvaldur Gunnlaugsson<thg at hafro.is>  wrote:
>    
>> At http://www.hafro.is/~thg/langtn/re/care/
>> there are files caresnr.tpl and .dat
>> where catch at age data (7x41) is simulated
>> with errors in catch, recruitment and stock (i.e. M?).
>> Random walk in F and Selection.
>> Exactly the same model is fitted
>> with all variables initialised to their true value.
>> Random effects are N(age,year), F(year) and 50% Selection_age(year)
>> When I change all the SEPARABLE_FUNCTIONS to just FUNCTION
>> the run-time is slightly shorter.
>>
>> The run command is in file xrun on the directory.
>>      
>>> uname -a is:
>>>        
>> Linux hafskip 2.6.30.9-96.fc11.x86_64 #1 SMP Wed Nov 4 00:02:04 EST 2009
>> x86_64 x86_64 x86_64 GNU/Linux
>> 1.9GB memory.
>>
>>