Using E-Surge

[Guillaume Souchay]

Hi,

I found the solution for the problem about my data: I delete all individuals who don't have an available BCI.
Ma data are loaded, I compute all my matrix in GEPAT (4 matrix, 2 steps for transition: survival + conditional survival transitions).
When I want to compute models in Gemaco, the Gemaco windows doesn't appear and I have the following error:

Error using ==> fsanf
Invalid file identifier -1
Error using ==> interface_gemaco_evt_V4
Too many output arguments
Error in ==> modme.contraint at 30
Error in ==> menumuse>Gemaco_Callback at 594
Error in ==> gui_mainfcn at 75
Error in ==> menumuse at 42
Error while evaluating uicontrol Callback.

What's the problem? Computing the matrix?
I'm using E-Surge v1.4.4

Thanks,

Guillaume

[CHOQUET]

I don't really understand the problem.
Perhaps, you have try to read an output file which is not valid.

sincerely,

Rémi

[Guillaume Souchay]

Hi,

I tried a second time, using the pattern matrix file I'saved the first time, and it's ok, so I don't really understand why there was a bug the first time.
But now, I'm thinking about GEMACO and I've some comprehension problem with the Initial state. To my data, birds are marked in only one site (I've got 4 sites), so for the first occasion of each birds, they can only be seen on the first site.
Therefore, the probability being in state 1 is 1 and in the others states, it's 0.
I have to set: to(1)+others for the initial state, isnt'it?

In my case, I don't think that Initial state is very useful (I only use E-Surge to integrate individual covariates into multisites models, I've already used M-Surge for all models without covariate).

Thanks,

Guillaume

[CHOQUET]

The simplest is to set the pattern matrix of the initial state vector as

* - -

which states that pi(2)=pi(3)=pi(4)=0 so pi(1)=1.

By this way, there is no parameter for the initial state.

Rémi

[Guillaume Souchay]

Hi,

I've tried your solution, but I still have a problem.
The idea of this model is to have a model without individual covariate with E-Surge and then to compare with one with covariate. The initial model is my best model of a first selection done with M-Surge.

I use the following matrix:
Initial state
* - - -

Transition
Step 1 : survival
σ - - - *
- σ - - *
- - σ - *
- - - σ *
- - - - *
Step 2 : transition
ψψψψ-
ψψψψ-
ψψψψ-
ψψψψ-
- - - - *
Event
*β - - -
*- β - -
*- - β -
*- - - β
*- - - -
For Gemaco,
I computed equations:
-IS: to
-Transtion survie: f(1).a(1)+f.a(2,3:6,7)+others
-Transition transition: f(1).to.ete.a(1)+f.to.a(2:7)
-Event: firste+nexte.to.t

I fixed some values: transition 2 2 1 1 1 1 to 1
event 1 2 1 1 1 1 to 1

I have no error message up to this point, but when I clicked on "RUN"
The following message appeared:
Error using ==> reshape
To reshape the number of elements must not change
Error in ==> modme.grad_F1_age_metat>gradPattern at 529

Error in ==> modme.grad_F1_age_metat at 489

Error in ==> modme.f_age_metat at 22

Error in ==> menumuse>Run_Callback at 902

Error in ==> gui_mainfcn at 75

Error in ==> menumuse at 42

Error while evaluating uicontrol Callback.

Thanks to be so patient and to answer all my newbi questions!!

Guillaume

[CHOQUET]

Replace the following incorrect pattern

Step 2 : transition
ψψψψ-
ψψψψ-
ψψψψ-
ψψψψ-
- - - - *

by

Step 2 : transition
ψψψ*-
ψψψ*-
ψψψ*-
ψψψ*-
- - - - *


Rémi

[Guillaume Souchay]

So, the problem was that I forget to indicate which parameter had not to be estimated directly, but as the n-1 parameter of the Markovian chain.

Thanks a lot!!

E-Surge is running!!

Guillaume