problem with unequal time intervals

[shelly]

Hi,

I am beginning a study investigating disease infection patterns in tits, in which I wish to explicitly account for uncertainty in disease state assignment and so am making the switch from MARK (with which I am familiar) to ESURGE. My problem is that I cannot get models to run with unequal time intervals. The data consist of yearly captures (during the breeding season) over a period of 9 years (disease state is unknown for some individuals in some years). In two of the nine years there were two sampling trips undertaken within the breeding season – two weeks apart – so essentially we have 11 occasions with uneven time intervals between four of them.

The file to account for unequal time intervals looks like this (yearly estimates)
1 1 1 1 1 0.05 0.95 0.05 0.95 1
However I also tried a file with interval values specified to give weekly estimates
52 52 52 52 52 2 50 2 50 52

I import these files as described in the manual and set the #steps to be “1 2”. But none of the models (even the simplest ones) work properly.
With time intervals set to years I get the following error messages on the black screen:

???Error using ->svd. Input to SVD must not contain NaN or Inf.
Error in -> modme.rank_F1_age at 231
Error in -> modme.rank_F1 at 16
Error in -> rhicc at 46
Error in -> menumuse>Run_callback at 1623
Error in -> gui_mainfcn at 96
Error in -> menmuse at 45
??? Error while evaluating unicontrol Callback

With time intervals set to weeks the model seems to run happily enough but the output is nonsense (infinitely large deviance and QAIC estimates, no estimated Hessian eigenvalues and obviously no standard errors estimated).

When I just have equal time intervals throughout (ignoring the two second capture occasions) the models all run fine.

What is going on?

[CHOQUET]

What are steps of transitions 1 or 2 ?
Rémi

[shelly]

Pattern Matrices are as follows:

Initial
π *

Survival (step1)
φ _ *
_ φ *
_ _ *

Transition (step2)
* ψ _
ψ * _
_ _ *

Capture (step1)
* β _
* _ β
* _ _

Events (state assignment, step2)
* _ _ _
_ δ _ *
_ _ δ *

[CHOQUET]

Hello,
your approach looks fine, perhaps there is a problem with
the small size of two intervals. But I don't know what happen.
If possible, send me your session so that I will be able to reproduce
a model with this problem and see if it can be solved.
Sincerely,
Rémi