We perform a priori identifiability investigation using (e,q)-identifibiability, ARE, and coverage criteria leveraging WENDy to generate large samples for model systems with unobserved compartments. Details of the (e,q) criterion and derivation of weak-form input-output equations can be found at:
File practical_ident.m will perform the identifiability analyses and generates Figures 4-5. Files bloodtissue_diffusion_parmaeter_estimation.m and sir_parameter_estimation.m perform a single parameter estimation for a given noise level and can be used to generate Figures 1-3. File performance_comparison.m tests the accuracy and speed of WENDy vs an output error based optimization and generates Figures 6-7.