Constraining age classes in Gemaco

[Alexlyon]

Hi,

I have noted one trouble when constraining the matrices in Gemaco in E-surge.
My data consist in 45 capture occasions and 12 groups (3 groups, two sexes in each group, 2 age-classes at the time of first capture in each sex and group)
First, I do not change the number of age-classes in "data status" (so, 44 possible age-classes)
I want one parameter to vary with 5 age classes. Group(1:6) is for individuals marked in the first age-classes (3 months of age), and group (7:12) is for individuals marked in the second age classes (9 months of age).
The different age-classes I use are 6 months-age classes, except the last, so:
a(1:6) ; a(7:12) ; a(13:18 ); a(19:24) ; a(25:...)

I thus constrain the associated matrix using gemaco like that
a(1:6).g(1:6)+a(1:6).g(7:12)&a(7:12).g(1:6)+a(7:12).g(7:12)&a(13:18 ).g(1:6)+a(13:18 ).g(7:12)&a(19:24).g(1:6)+a(19:44).g(7:12)&a(25:44).g(1:6)
I call Gemaco, the DOS window sends an error message and the constrain doesn't work

Now, I change the number of age-classes into 25 in "data status", and the same sentence in Gemaco works well...

So, first I don't understand why.
Second, I would know whether changing the number of age classes in "data status" would affect parameter estimates (especially for the parameter invoked for the last age-classe in my case)

thanks in advance

sincerly

Alexis

[Alexlyon]

Hi,

Another trouble I met (may be the same as the previous one...)

I want one parameter to depend on age-classes.
The data are quiet different than the previous ones
They consists in 44 capture occasions and 12 groups again
groups are defined as follows
g(1:6) : population 1
g(7:12) : population 2
g(1:3) and g(7:9) : males
g(4:6) and g(10:12) : females
and for each sex and population three age classes at the time of first capture :
marked between 30-60 days
marked between 60-90 days
marked as adults (more than 180 days)

I use monthly capture occasion

Constraining the parameter depending on two age classes (juveniles ie 30-180 days, and adults, ie more than 180 days) leads to the following sentence (tell me if I am wrong):
a(1:5).g(1 4 7 10)&a(1:4).g(2 5 8 11)+a(6:44).g(1 4 7 10)&a(5:44).g(2 5 8 11)&g(3 6 9 12)

here, this constrain doesn't work (DOS windows says "matlab error")

Then, when I use an another dataset, actually a new file.inp with the number of capture occasions reduced to 22, or when I do not consider two population and so I reduce the number of group to 6.. the respective following sentences work well and the model runs...
a(1:5).g(1 4 7 10)&a(1:4).g(2 5 8 11)+a(6:22).g(1 4 7 10)&a(5:22).g(2 5 8 11)&g(3 6 9 12)
or
a(1:5).g(1 4)&a(1:4).g(2 5)+a(6:44).g(1 4)&a(5:44).g(2 5)&g(3 6)

Again I do not understand why. It seems that this sentence doesn't work anymore when increasing the number of group or age classes...

many thanks again

Alexis

[Alexlyon]

Hi, again

just one mistake in the previous posts.
My data consists in 45 capture occasions (not 44), and so 44 possible age-classes

sorry

Alexis

[Alexlyon]

Hi

I tested the previous sentences in two different computers (both working with windows XP, but the one at the lab was an AMD64 and the other was older).
I had installed E-Surge 1.4.6 and matlab v7.2 on the two computers.
Actually the sentence works in the old computer but not on the AMD64

When searching for compatibility versions between matlab and computer, we found that matlab 7.2 (service pack 2) was not compatible with an AMD64 working with windows XP!

Could my trouble with Gemaco come from this ?

thanks in advance

Alexis

[CHOQUET]

Presently, the programs are not compiled for 64-bit computer.
If there is a option to execute the program in a 32-bit mode then
try it.
Rémi