As client, i'd like to have a stable behaviour for outputs of the `fm_fit_model` function

[Cervangirard]

As client, i'd like to have a stable behaviour for outputs of the fm_fit_model function so that I am warned if there are differences in outputs while changing other functions or while updating package dependencies

Client - Validation

After this is unblocked we plan to validate these points:

• [ ] There is a unit test for each result of the function. The unit test is properly chosen depending on the size of the output, so that it does not make a too heavy package.
  • [ ] time.step_df_output
  • [ ] loc_x
  • [ ] report_output with only a look at the spatial predictions (S_x) and total biomass (total_abundance).
  • [ ] samp_process_output

Dev - Tech

• [ ] Save only what is needed for the outputs

• [ ] Units tests for each result

• [ ] => We suggest to only look at the spatial predictions (S_x) and total biomass (total_abundance). You should get it by doing fm_model_results$report$S_p and fm_model_results$report$total_abundance.

Problem

We want to write stable unit tests that can, over time, identify real differences in the model's output. Checking on the entire model results can cause the test to fail on very small differences. Here is an example:

• What are the essential elements of the list of fm_fit_model that are stable over time if the parameters are unchanged?
  • For each sub-element of the list, what are the elements to be tested?

[balglave]

We suggest to only look at the spatial predictions (S_x) and total biomass (total_abundance). You should get it by doing fm_model_results$report$S_p and fm_model_results$report$total_abundance.

[Cervangirard]

Thanks. We close this ticket because the tests pass for a model with a small k and a short period. We will reprocess this issue later with a more consistent model

[Cervangirard]

Ajouter lambda_p et par_b