Run all models in model set w/different estimation options

[tds]

Is there a way to rerun all models in a model set with different numerical estimation options at the same time? (I don't want to have to retrieve and run each model individually.)

Then, is there a way to keep all these new models and delete all the original models? (I don't want to have to manually delete all the old models individually.)

For example, I have 64 models in my model set, and I want to run ~9 different "user-specified covariate values" for each model (say -4, -3, -2, -1, 0, 1, 2, 3, 4). I can do it now by retrieving each model, hitting the "user-specified covariate values" button, and specifying the individual covariate values (all 0 except for the parameter of interest, A (first round A = -4). Then, I need to delete the old model. It's possible, but this way takes a really long time (all 64 models for covariate = -4, then all 64 or covariate = -3, etc.).

Thank you.

[cooch]

Is there a way to rerun all models in a model set with different numerical estimation options at the same time? (I don't want to have to retrieve and run each model individually.)

No, there isn't a simple way to re-run all the models specifying a different optimization method (although I suspect you could cobble together something using RMark, with its 'scripting' capabilities.)

However, there is no reason for you to do that. You only need to change from the default optimization to simulated annealing if you have strong reason to suspect there is a numerical convergence problem. If that is all your models, then the optimization algorithm isn't thee issue, your data are (at least when confronted with the models you're trying to fit).

Then, is there a way to keep all these new models and delete all the original models? (I don't want to have to manually delete all the old models individually.)

No, capacity not built into MARK, since in practice, you only need to try SA for a few models -- so few, that manually keeping track of the ones you want to keep, and which ones you want to chuck out, is trivial.

For example, I have 64 models in my model set, and I want to run ~9 different "user-specified covariate values" for each model (say -4, -3, -2, -1, 0, 1, 2, 3, 4). I can do it now by retrieving each model, hitting the "user-specified covariate values" button, and specifying the individual covariate values (all 0 except for the parameter of interest, A (first round A = -4). Then, I need to delete the old model. It's possible, but this way takes a really long time (all 64 models for covariate = -4, then all 64 or covariate = -3, etc.).

Thank you.

That kind of capability is one example of a reason why some people have moved to RMark, which has a function to do what you describe (more or less).

[cooch]

Is there a way to rerun all models in a model set with different numerical estimation options at the same time? (I don't want to have to retrieve and run each model individually.)

Just version just uploaded -- gives you this capability. Simply select 'Run | re-run models', and you'll be shown an option to re-run using simulated annealing for *all* the models you select. When you re-run a model, it doesn't add a new model to the browser (even if you use a different optimization). It simply rewrite the results for that model.